Overcoming Controllable and Uncontrollable Work Difficulties: Change Environment or Self?

Elise Maher, B.Sc. (Hons)

Submitted in fulfillment of the requirements for the degree of Doctor of Philosophy Deakin University, December 2002

ii

DEAKIN UNIVERSITY
CANDIDATE DECLARATION

I certify that the thesis entitled Overcoming Controllable and Uncontrollable Work Difficulties: Change Environment or Self?

submitted for the degree of Doctor of Philosophy is the result of my own work and that where reference is made to the work of others, due acknowledgment is given.

I also certify that any material in the thesis which has been accepted for a degree or diploma by any other university or institution is identified in the text.

Full Name

Elise Catherine Maher

Signed ................................................................................... Date.......................................................................................

iii

Acknowledgements First and foremost, I would like to acknowledge my supervisor, Professor Robert Cummins. Bob offered endless support, guidance, and encouragement throughout my studies. His dedication to research and his tremendous work ethic motivated and inspired me. I am most grateful that he challenged me and allowed me the freedom and respect to develop my own ideas and theories. Second, I must thank all the organisations that allowed me to enter their workplaces, and all the participants involved in the studies. I especially want to thank staff at Australian Unity, members of the Australian Centre of Quality of Life, and the hundreds of supermarket workers, teachers and academics that assisted me. These people invested time and energy into completing my survey purely for the benefit of helping others. I am so appreciative of their efforts and I am determined to share the knowledge that I have gained from them. Third, I would like to acknowledge the academic and administrative staff, and fellow students at Deakin University. I especially want to acknowledge Rose-Anne and Helen (my surrogate parents) for taking me under their wings. Their delightful, vibrant personalities made work times pleasurable. Also, I would like to acknowledge Carolyn and Catherine for always being there to listen and share. Finally, I would like to acknowledge my family and friends. My parents, John and Frances, and my brothers have provided endless support throughout the last seven years. I also want to acknowledge Lauren, Taylah, Tyson and Buffy for always lighting up my life. My partner, Tim Davis, has been my tower of strength, and I am forever grateful for his love and support.

iv

Elise

Table of Contents
ACKNOWLEDGEMENTS.............................................................................................................................III LIST OF TABLES...................................................................................................................................XIII LIST OF FIGURES.................................................................................................................................XVII LIST OF APPENDICES.............................................................................................................................XIX ABSTRACT...........................................................................................................................................XXI CHAPTER 1 - LITERATURE REVIEW................................................................................................1 0.1 ABSTRACT........................................................................................................................................2 0.2 JOB SATISFACTION.............................................................................................................................3 0.2.1 Theories of Job Satisfaction: Environmental and Dispositional Predictors...........................3 0.3 MASLOWS (1954, 1970) HIERARCHY OF NEEDS.................................................................................5 0.3.1 Need Hierarchy Theory..........................................................................................................5 0.3.2 Applying Maslows (1954, 1970) Theory to Organisations....................................................6 0.3.3 Criticisms of Maslows (1954, 1970) Need Hierarchy Theory...............................................7 0.3.4 Conclusion...........................................................................................................................20 0.4 HERZBERG, MAUSNER AND SNYDERMANS (1959, 1993) TWO-FACTOR THEORY OF JOB SATISFACTION....21 0.4.1 How the Two-Factor Theory has Contributed to our Understanding of Job Satisfaction.....21 0.4.2 Development of the Two-Factor Theory...............................................................................21 0.4.3 Criticisms of Herzberg et als., (1959) Theory ....................................................................22 0.4.4 Conclusion...........................................................................................................................31 0.5 VROOMS (1964) EXPECTANCY THEORY OF JOB SATISFACTION.............................................................33 0.5.1 How Expectancy Theory has Contributed to our Knowledge of Job Satisfaction.................33 0.5.2 Description of Expectancy Theory........................................................................................33 0.5.3 Applications of the Valence Model.......................................................................................35 0.5.4 Studies of the Valence Model...............................................................................................35 0.5.5 Methodological Limitations..................................................................................................36

vi

0.5.6 Conclusion...........................................................................................................................42 0.6 DISCREPANCY THEORIES...................................................................................................................43 0.6.1 How Discrepancy Theories have Contributed to our Knowledge of Job Satisfaction...........43 0.6.2 Description of Discrepancy Theories...................................................................................43 0.6.3 Empirical Studies Investigating Discrepancy Theories........................................................43 0.6.4 Theoretical Problems with Discrepancy Theories................................................................44 0.6.5 Conclusion...........................................................................................................................45 0.7 JOB CHARACTERISTICS MODEL (JCM; HACKMAN & OLDHAM, 1976)..................................................46 0.7.1 How the Job Characteristics Model has Contributed to our Knowledge of Job Satisfaction46 0.7.2 Description of the Job Characteristics Model .....................................................................46 0.7.3 Empirical Studies of the Model............................................................................................49 0.7.4 Conclusion...........................................................................................................................55 0.8 JOB DEMAND-CONTROL MODEL (KARASEK, 1979; KARASEK & THEORELL, 1990)................................56 0.8.1 How the Job Demand-Control Model Contributes to our Understanding of Job Satisfaction ......................................................................................................................................................56 0.8.2 Description of the Job Demand-Control Model...................................................................56 0.8.3 Empirical Studies of the Job Demand-Control Model..........................................................58 0.8.4 Conclusion...........................................................................................................................61 0.8.5 Extensions on the Job Demand-Control Model....................................................................61 0.8.6 Addressing the Gaps in the Job Demand-Control Model................................................62 0.9 DEVELOPMENT OF A NEW EXPLANATION FOR THE RELATIONSHIP BETWEEN JOB AUTONOMY AND JOB SATISFACTION: INFLUENCING EMPLOYEES RESPONSES TO WORK DIFFICULTIES..............................................63 0.9.1 a) Primary Control Strategies and Secondary Control Strategies.......................................64 0.9.2 b) Amounts of Primary Control and Secondary Control .....................................................66 0.9.3 c) Which Control Strategies are more Adaptive for Employees?..........................................67 0.9.4 Summary..............................................................................................................................73

vii

0.10 EXPLAINING THE RELATIONSHIP BETWEEN JOB AUTONOMY AND JOB SATISFACTION: HOW JOB AUTONOMY INFLUENCES PRIMARY AND SECONDARY CONTROL......................................................................................73 0.10.1 1) Use of Primary and Secondary Control.........................................................................74 0.10.2 2) Adaptiveness of Primary and Secondary Control..........................................................75 0.10.3 Summary............................................................................................................................83 0.11 OTHER MAJOR PREDICTORS OF JOB SATISFACTION.............................................................................85 0.11.1 Personality.........................................................................................................................85 0.11.2 Life Satisfaction..................................................................................................................88 0.12 MODEL OF JOB SATISFACTION.........................................................................................................94 1 CHAPTER 2 - STUDY ONE.................................................................................................................97 1.1 ABSTRACT......................................................................................................................................98 1.2 PROPOSAL FOR STUDY ONE...............................................................................................................99 1.2.1 Identifying Employees with Low/High Job Autonomy..........................................................99 1.3 AIMS AND HYPOTHESES..................................................................................................................101 1.4 METHOD.......................................................................................................................................104 1.4.1 Participants........................................................................................................................104 1.4.2 Materials............................................................................................................................104 1.4.3 Procedure...........................................................................................................................110 1.5 RESULTS.......................................................................................................................................112 1.5.1 Data Screening and Checking of Assumptions...................................................................112 1.5.2 Descriptive Statistics and Inter-Correlations ....................................................................114 1.5.3 Factor Analyses..................................................................................................................116 1.5.4 Factor Analysis of the Job Descriptive Index.....................................................................116 1.5.5 Factor Analysis of the Primary and Secondary Control Scale...........................................117 1.5.6 Factor Analysis of the Job Autonomy Scale........................................................................121 1.6 HYPOTHESIS TESTING.....................................................................................................................123 1.6.1 Hypothesis One- Assumption Testing.................................................................................123

viii

1.6.2 Hypothesis Two- Occupational Differences in the Use of the Control Strategies..............124 1.6.3 Hypothesis Three- Examining how Job Autonomy Relates to the Control Strategies.........126 1.6.4 Hypothesis Four- Examining how Job Autonomy Influences the Adaptiveness of the Control Strategies....................................................................................................................................127 1.6.5 Hypothesis Five- Does Job Autonomy Moderate the Relationship Between the Control Strategies and Job Satisfaction?.................................................................................................128 1.6.6 Hypothesis Six- Do the Control Strategies Mediate the Relationship Between Job Autonomy and Job Satisfaction?..................................................................................................................133 1.6.7 Hypothesis Seven- Occupational Differences in Job and Life Satisfaction.........................138 1.6.8 Hypothesis Eight- Predictors of Job Satisfaction...............................................................142 1.6.9 Conclusion.........................................................................................................................145 1.7 DISCUSSION...................................................................................................................................146 1.7.1 Assumption Testing............................................................................................................146 1.7.2 Does Job Autonomy Influence the Use of the Control Strategies?......................................147 1.7.3 Does Job Autonomy Influence the Relationship Between the Control Strategies and Job Satisfaction?................................................................................................................................150 1.7.4 Do the Control Strategies Mediate the Relationship Between Job Autonomy and Job Satisfaction?................................................................................................................................152 1.7.5 Examining Occupational Differences in Job Satisfaction...................................................153 1.7.6 Examining Occupational Differences in Life Satisfaction..................................................158 1.7.7 Predicting Job Satisfaction from Job Autonomy, Control Strategies, Personality, and Life Satisfaction..................................................................................................................................160 1.7.8 Conclusion.........................................................................................................................161 2 CHAPTER 3 - STUDY TWO..............................................................................................................163 2.1 ABSTRACT....................................................................................................................................164 2.2 PROPOSAL FOR STUDY TWO............................................................................................................165 2.2.1 a) The Primary and Secondary Control Scale....................................................................165

ix

2.2.2 b) Job Autonomy Scale.......................................................................................................177 2.2.3 c) Occupational Groups.....................................................................................................178 2.2.4 d) Need for Job Autonomy..................................................................................................180 2.2.5 e) Addition of Social Support.............................................................................................182 2.3 MODEL OF JOB SATISFACTION.........................................................................................................185 2.4 AIMS AND HYPOTHESES..................................................................................................................187 2.5 METHOD.......................................................................................................................................191 2.5.1 Participants........................................................................................................................191 2.5.2 Materials ...........................................................................................................................192 2.5.3 Procedure...........................................................................................................................199 2.6 RESULTS.......................................................................................................................................200 2.6.1 Data Screening and Checking of Assumptions...................................................................200 2.6.2 Descriptive Statistics and Inter-Correlations ....................................................................201 2.6.3 Preliminary Examination of the Primary Control and Secondary Control Scale...............202 2.7 HYPOTHESIS TESTING.....................................................................................................................207 2.7.1 Hypothesis One: Levels of Job Autonomy and Job Satisfaction.........................................207 2.7.2 Hypotheses Two and Three: Examining how Job Autonomy Influences the Amount of Primary and Secondary Control Strategies.................................................................................208 2.7.3 Hypotheses Four and Five: Examining how Job Autonomy Influences the Relationship Between the Control Strategies and Job Satisfaction..................................................................209 2.7.4 Hypothesis Six: Examining the Proposed Explanation for the Relationship Between Job Autonomy and Job Satisfaction...................................................................................................213 2.7.5 Hypothesis Seven: Occupational Differences in Job Satisfaction and Life Satisfaction.....214 2.7.6 Hypothesis Eight: Examining how Social Support at Work Moderates the Relationship between Difficulties at Work and Job Satisfaction......................................................................218 2.7.7 Hypothesis Nine: The Moderating Role of Need for Autonomy on the Relationship Between Job Autonomy and Job Satisfaction ............................................................................................223 2.7.8 Hypothesis Ten: Major Predictors of Job Satisfaction.......................................................224

2.7.9 Conclusion.........................................................................................................................227 2.8 DISCUSSION...................................................................................................................................228 2.8.1 Assumption- The Academics Represent a High Job Autonomy Group and the Teachers Represent a Low Job Autonomy Group.......................................................................................228 2.8.2 Hypothesis Testing.............................................................................................................231 2.8.3 Job Autonomy Influences the Amount of the Control Strategies.........................................231 2.8.4 Job Autonomy Influences the Relationship Between the Control Strategies and Job Satisfaction..................................................................................................................................234 2.8.5 Limitations in the Hypotheses Examining Job Autonomy and Control Strategies ............235 2.8.6 Other Predictors of Job Satisfaction..................................................................................237 2.8.7 Occupational Differences in Job Satisfaction and Life Satisfaction...................................237 2.8.8 The Influence of Social Support at Work on the Relationship Between Work Difficulties and Job Satisfaction...........................................................................................................................240 2.8.9 The Influence that Need for Job Autonomy has on the Relationship Between Job Autonomy and Job Satisfaction....................................................................................................................242 2.8.10 Major predictors of Job Satisfaction................................................................................243 2.8.11 Conclusion.......................................................................................................................244 3 CHAPTER 4 - STUDY THREE..........................................................................................................245 3.1 ABSTRACT....................................................................................................................................246 3.2 PROPOSAL FOR STUDY THREE..........................................................................................................247 3.2.1 Specificity of Hypotheses Testing the Proposal that Job Autonomy Influences the Control Strategies....................................................................................................................................247 3.2.2 Examining how the Controllability of a Difficulty Influences the Use of the Control Strategies....................................................................................................................................248 3.2.3 Empirical Studies Examining if the Controllability of a Situation Influences the Use of Control Strategies.......................................................................................................................249 3.2.4 Examining how Controllability Influences the Adaptiveness of the Control Strategies......253

xi

3.2.5 Developing a Situation Specific Primary and Secondary Control Scale.............................257 3.2.6 Examining the Moderating Role of Primary and Secondary Control Strategies.................261 3.2.7 Examining the Moderating Role of Social Support at Work...............................................262 3.3 REVISED MODEL OF JOB SATISFACTION............................................................................................265 JOB SATISFACTION..............................................................................................................................267 3.4 HYPOTHESES.................................................................................................................................269 3.5 METHOD.......................................................................................................................................271 3.5.1 Participants........................................................................................................................271 3.5.2 Materials............................................................................................................................271 3.5.3 Procedure...........................................................................................................................277 3.6 RESULTS.......................................................................................................................................278 3.6.1 Data Screening and Checking of Assumptions...................................................................278 3.6.2 Descriptive Statistics and Inter-Correlations.....................................................................279 3.6.3 Factor Analyses..................................................................................................................281 3.6.4 Primary and Secondary Control Scale...............................................................................281 3.6.5 Social Support at Work.......................................................................................................286 ITEMS WITH LOADINGS LESS THAN 0.30 ARE NOT SHOWN..........................................................................287 3.7 HYPOTHESIS TESTING.....................................................................................................................288 3.7.1 Hypothesis One- Use of Control Strategies for Controllable and Uncontrollable Difficulties ....................................................................................................................................................288 3.7.2 Hypothesis Two- Adaptiveness of the Control Strategies for Controllable and Uncontrollable Difficulties..........................................................................................................292 3.7.3 Hypothesis Three- The Moderating Role of Primary and Secondary Control....................295 3.7.4 Hypothesis Four - Moderating Role of Instrumental Support.............................................300 3.7.5 Hypothesis Five- Moderating Role of Emotional Support..................................................304 3.7.6 Hypothesis Six- Major Predictors of Job Satisfaction........................................................307 3.7.7 Conclusion.........................................................................................................................308 3.8 DISCUSSION...................................................................................................................................310

xii

3.9 HYPOTHESES TESTING....................................................................................................................310 3.9.1 Primary Control, Self-Protective Secondary Control, and Self-Affirmative Secondary Control........................................................................................................................................311 3.9.2 Proposal One: The Controllability of the Difficulty Influences the Amount and Adaptiveness of the Control Strategies Used to Manage that Difficulty ..........................................................314 3.9.3 Proposal Two: Moderators of Controllable and Uncontrollable Difficulties on Job Satisfaction..................................................................................................................................325 3.9.4 Proposal Three: Predictors of Job Satisfaction..................................................................331 3.9.5 Conclusion.........................................................................................................................332 4 CHAPTER 5 - FINAL DISCUSSION................................................................................................334 4.1 ABSTRACT....................................................................................................................................336 4.2 THE DEVELOPMENT OF A NEW MODEL OF JOB SATISFACTION.............................................................337 1) Primary and Secondary Control Strategies Mediate the Relationship Between Job Autonomy and Job Satisfaction....................................................................................................................339 4.2.2 Conclusion: Do the Control Strategies Mediate the Relationship Between Job Autonomy and Job Satisfaction?.........................................................................................................................346 4.2.3 2) Social Support at Work and Life Satisfaction Directly Predict Job Satisfaction............348 4.2.4 3) The Control Strategies and Social Support at Work Moderate the Relationship Between Work Difficulties and Job Satisfaction........................................................................................350 4.3 REVISED MODEL OF JOB SATISFACTION............................................................................................353 JOB SATISFACTION..............................................................................................................................355 4.4 CONCLUSION.................................................................................................................................356 4.5 FINAL WORD................................................................................................................................357 4.6 REFERENCES..................................................................................................................................358 4.7 APPENDICES..................................................................................................................................394

xiii

List of Tables

TABLE 1- SOURCES OF GOOD/BAD TIMES FOR ACCOUNTANTS AND ENGINEERS (N=200).........................................................................................................................................................22 TABLE 2- SECONDARY CONTROL STRATEGIES.........................................................................72 TABLE 3- MEANS AND STANDARD DEVIATIONS OF MAJOR VARIABLES FOR ACADEMICS AND SUPERMARKET WORKERS..........................................................................114 TABLE 4- INTER-CORRELATIONS FOR THE ACADEMICS AND THE SUPERMARKET WORKERS...............................................................................................................................................115 TABLE 5- FACTOR ANALYSIS OF JOB SATISFACTION SCALE............................................117 TABLE 6- TOTAL VARIANCE EXPLAINED BY A FIVE-FACTOR SOLUTION....................119 TABLE 7- FACTOR ANALYSIS OF PRIMARY AND SECONDARY CONTROL SCALE......120 TABLE 8- FACTOR ANALYSIS OF JOB AUTONOMY SCALE..................................................121 TABLE 9- MEANS AND STANDARD DEVIATIONS OF CONTROL MEASURES FOR ACADEMICS AND SUPERMARKET WORKERS..........................................................................126 TABLE 10- MULTIPLE REGRESSION OF PRIMARY AND SECONDARY CONTROL ON JOB SATISFACTION FOR ACADEMICS AND SUPERMARKET WORKERS........................128 TABLE 11- MODERATING ROLE OF PRIMARY AND SECONDARY CONTROL ON THE RELATIONSHIP BETWEEN JOB AUTONOMY AND JOB SATISFACTION..........................133 TABLE 12 -HIERARCHICAL MULTIPLE REGRESSION TESTING THE MEDIATING ROLE OF THE CONTROL STRATEGIES........................................................................................135

xiv

TABLE 13- MEANS AND STANDARD DEVIATIONS OF JOB SATISFACTION SCALE FOR ACADEMICS AND SUPERMARKET WORKERS..........................................................................140 TABLE 14- MEANS AND STANDARD DEVIATIONS OF LIFE SATISFACTION FOR ACADEMICS AND SUPERMARKET WORKERS..........................................................................142 TABLE 15- MULTIPLE REGRESSION OF JOB AUTONOMY, CONTROL STRATEGIES, PERSONALITY, AND LIFE SATISFACTION FOR ACADEMICS AND SUPERMARKET WORKERS...............................................................................................................................................144 TABLE 16- FACTOR ANALYSIS OF PRIMARY AND SECONDARY CONTROL SCALE....167 TABLE 17- ORIGINAL AND REVISED PRIMARY CONTROL ITEMS.....................................171 TABLE 18- ORIGINAL AND REVISED SECONDARY CONTROL ITEMS..............................175 TABLE 19- FUNCTIONS OF THE SECONDARY CONTROL STRATEGIES...........................175 TABLE 20- DEMOGRAPHICS OF THE ACADEMICS AND TEACHERS.................................192 TABLE 21- FACTOR ANALYSIS OF THE NEED FOR JOB AUTONOMY SCALE.................195 TABLE 22- MEANS AND STANDARD DEVIATIONS OF MAJOR VARIABLES FOR ACADEMICS AND TEACHERS..........................................................................................................201 TABLE 23- INTER-CORRELATIONS FOR THE ACADEMICS AND TEACHERS.................202 TABLE 24- FREQUENCY OF PRIMARY AND SECONDARY CONTROL................................203 TABLE 25- FACTOR ANALYSIS OF THE REVISED PRIMARY AND SECONDARY CONTROL SCALE.................................................................................................................................205 TABLE 26- MULTIPLE REGRESSION OF PRIMARY AND SECONDARY CONTROL ON JOB SATISFACTION FOR ACADEMICS AND TEACHERS........................................................210

xv

TABLE 27- HIERARCHICAL MULTIPLE REGRESSION TESTING THE MODERATING ROLE OF THE CONTROL STRATEGIES ON THE RELATIONSHIP BETWEEN JOB AUTONOMY AND JOB SATISFACTION.........................................................................................212 TABLE 28- HIERARCHICAL MULTIPLE REGRESSION TESTING THE MEDIATING ROLE OF THE CONTROL STRATEGIES........................................................................................213 TABLE 29- MEANS AND STANDARD DEVIATIONS OF THE INTRINSIC AND EXTRINSIC JOB SATISFACTION ITEMS FOR ACADEMICS AND TEACHERS..........................................216 TABLE 30- MEANS AND STANDARD DEVIATIONS OF THE DOMAINS OF LIFE SATISFACTION FOR ACADEMICS AND TEACHERS.................................................................217 TABLE 31- HIERARCHICAL MULTIPLE REGRESSION ANALYSIS EXAMINING IF SUPERVISOR SUPPORT MODERATES THE RELATIONSHIP BETWEEN WORK DIFFICULTIES AND JOB SATISFACTION.....................................................................................219 TABLE 32- HIERARCHICAL REGRESSION ANALYSES EXAMINING WHETHER COWORKER SUPPORT MODERATES THE RELATIONSHIP BETWEEN WORK DIFFICULTIES AND JOB SATISFACTION. ...................................................................................222 TABLE 33- HIERARCHICAL REGRESSION ANALYSES EXAMINING WHETHER NEED FOR JOB AUTONOMY MODERATES THE RELATIONSHIP BETWEEN JOB AUTONOMY AND JOB SATISFACTION. .................................................................................................................224 TABLE 34- STANDARD MULTIPLE REGRESSION PREDICTING JOB SATISFACTION FOR EMPLOYEES WITH LOW AUTONOMY AND EMPLOYEES WITH HIGH AUTONOMY............................................................................................................................................226 TABLE 35- NORMATIVE DATA FOR HACKMAN AND OLDHAMS (1980) AUTONOMY SCALE.......................................................................................................................................................230

xvi

TABLE 36- DEMOGRAPHIC CHARACTERISTICS OF THE SAMPLE....................................271 TABLE 37- MEANS AND STANDARD DEVIATIONS OF THE MAJOR VARIABLES...........280 TABLE 38- INTER-CORRELATIONS AMONG MAJOR VARIABLES......................................280 TABLE 39- FACTOR ANALYSIS OF PRIMARY AND SECONDARY CONTROL ITEM IN CONTROLLABLE SITUATIONS........................................................................................................283 TABLE 40- FACTOR ANALYSIS OF PRIMARY AND SECONDARY CONTROL ITEMS IN UNCONTROLLABLE SITUATION....................................................................................................285 TABLE 41-SECONDARY CONTROL ITEMS INCLUDED IN ANALYSES...............................286 TABLE 42- FACTOR ANALYSIS OF THE SOCIAL SUPPORT AT WORK SCALE................287 TABLE 43 - CONTROLLABLE AND UNCONTROLLABLE DIFFICULTIES REPORTED BY EMPLOYEES...........................................................................................................................................288 TABLE 44- EMPLOYEES USE OF PRIMARY AND SECONDARY CONTROL IN CONTROLLABLE AND UNCONTROLLABLE SITUATIONS.....................................................290 TABLE 45- MEANS AND STANDARD DEVIATIONS OF INDIVIDUAL CONTROL STRATEGIES...........................................................................................................................................292 TABLE 46- STANDARD MULTIPLE REGRESSION ANALYSIS PREDICTING JOB SATISFACTION FROM PRIMARY AND SECONDARY CONTROL.........................................294 TABLE 47- CORRELATIONS BETWEEN INDIVIDUAL CONTROL STRATEGIES AND JOB SATISFACTION FOR CONTROLLABLE AND UNCONTROLLABLE DIFFICULTIES.......295 TABLE 48- HIERARCHICAL MULTIPLE REGRESSION TESTING THE MODERATING ROLE OF CONTROL STRATEGIES ON THE RELATIONSHIP BETWEEN WORK DIFFICULTIES AND JOB SATISFACTION.....................................................................................299

xvii

TABLE 49- HIERARCHICAL REGRESSION ANALYSES TESTING THE MODERATING ROLE OF INSTRUMENTAL SUPPORT............................................................................................304 TABLE 50- HIERARCHICAL REGRESSION ANALYSES TESTING THE MODERATING ROLE OF EMOTIONAL SUPPORT...................................................................................................306 TABLE 51- STANDARD MULTIPLE REGRESSION PREDICTING JOB SATISFACTION. .308

List of Figures

FIGURE 1-JOB CHARACTERISTICS MODEL.................................................................................48 FIGURE 2- MODEL OF JOB SATISFACTION...................................................................................96 FIGURE 3- EXPECTED MODERATED EFFECT OF JOB AUTONOMY ON A) PRIMARY CONTROL AND JOB SATISFACTION AND B) SECONDARY CONTROL AND JOB SATISFACTION .....................................................................................................................................129 FIGURE 4- JOB AUTONOMY MODERATES THE RELATIONSHIP BETWEEN A) PRIMARY CONTROL AND B) SECONDARY CONTROL, AND JOB SATISFACTION........131 FIGURE 5- MEDIATING ROLE OF CONTROL STRATEGIES ON THE RELATIONSHIP BETWEEN JOB AUTONOMY AND JOB SATISFACTION...........................................................134 FIGURE 6 -REVISED MODEL OF JOB SATISFACTION FOR STUDY 2 .................................186 FIGURE 7 - RELATIONSHIP BETWEEN WORK DIFFICULTIES AND JOB SATISFACTION FOR EMPLOYEES WITH LOW/HIGH SUPERVISOR SUPPORT..............................................221 FIGURE 8- REVISED MODEL OF JOB SATISFACTION ............................................................267 FIGURE 9 PRIMARY AND SECONDARY CONTROL MODERATE THE RELATIONSHIP BETWEEN WORK DIFFICULTIES AND JOB SATISFACTION.................................................297

xviii

FIGURE 10 - REGRESSION OF CONTROLLABLE WORK DIFFICULTIES ON JOB SATISFACTION FOR EMPLOYEES WITH LOW INSTRUMENTAL CO-WORKER SUPPORT AND EMPLOYEES WITH HIGH INSTRUMENTAL CO-WORKER SUPPORT . 301 FIGURE 11- REVISED MODEL OF JOB SATISFACTION ..........................................................355

xix

List of Appendices

APPENDIX A- PLAIN LANGUAGE STATEMENT FOR STUDY ONE.......................................395 APPENDIX B- JOB AUTONOMY SCALE USED IN STUDY ONE (REVISION OF GANSTER, 1989, CITED IN DWYER & GANSTER, 1991)...................................................................................396 APPENDIX C- PRIMARY AND SECONDARY CONTROL SCALE USED IN STUDY ONE (REVISION OF HEEPS ET AL., 2000) ...............................................................................................398 APPENDIX D- JOB SATISFACTION SCALE USED IN STUDY ONE (REVISION OF ROZNOWSKI, 1989) ..............................................................................................................................401 APPENDIX E- LIFE SATISFACTION SCALE USED IN STUDY ONE (CUMMINS, 1997).....403 APPENDIX F- PERSONALITY SCALE USED IN STUDY ONE (COSTA & MCCRAE, 1992) .....................................................................................................................................................................404 APPENDIX G-LEVELS OF JOB SATISFACTION REPORTED BY VARIOUS OCCUPATIONAL GROUPS.................................................................................................................407 APPENDIX H- PRIMARY AND SECONDARY CONTROL SCALE FOR STUDY TWO (MAHER ET AL., 2001) .........................................................................................................................410 APPENDIX I- PLAIN LANGUAGE STATEMENT FOR STUDY TWO.......................................414 APPENDIX J- JOB AUTONOMY SCALE FOR STUDY TWO (HACKMAN & OLDHAM, 1975) .....................................................................................................................................................................414 APPENDIX K- NEED FOR AUTONOMY SCALE FOR STUDY TWO (DE RIJK ET AL., 1998) .....................................................................................................................................................................416 APPENDIX L- JOB SATISFACTION SCALE FOR STUDY TWO (WEISS ET AL., 1967).......417

xx

APPENDIX M- SOCIAL SUPPORT SCALE FOR STUDY TWO (REVISION OF KARASEK & THEORELL, 1990)..................................................................................................................................420 APPENDIX N-PLAIN LANGUAGE STATEMENT USED IN STUDY THREE...........................421 APPENDIX O- PRIMARY AND SECONDARY CONTROL SCALE FOR STUDY THREE (MAHER & CUMMINS, 2002)..............................................................................................................422 APPENDIX P- LIFE SATISFACTION SCALE FOR STUDY 3 (CUMMINS ET AL., 2001)......428 APPENDIX Q- SOCIAL SUPPORT SCALE FOR STUDY 3 (DUCHARME & MARTIN, 2000) .....................................................................................................................................................................429

xxi

Abstract Although theories of job satisfaction have been extensively studied, researchers are yet to agree on the major predictors of job satisfaction. One theory, which is particularly appealing to the workplace, is Karasek and Theorells (1990) job demand-control model. Essentially, this model proposes that job autonomy can reduce the effects of job demands on job satisfaction by allowing workers to redirect the physiological arousal produced from job demands into an appropriate response. This explanation is criticised however for being tautological, and a new explanation is developed which incorporates the life span theory of control (Heckhausen & Schulz, 1995) and the discrimination model (Thompson et al., 1998). Specifically it is proposed that job autonomy influences the use, and the adaptiveness of primary and secondary control strategies. This proposal is developed into a model of job satisfaction that includes job autonomy, primary and secondary control, life satisfaction, personality, and social support at work. This model of job satisfaction is tested over three studies using university academic staff, secondary school teachers, supermarket workers and general employees. Overall, the results demonstrated that job autonomy did not influence the use or adaptiveness of the control strategies. These results suggest that employees have trait control strategies, and they also challenge the assumptions about primary control failure. The proposed model of job satisfaction was revised to include job autonomy, primary and secondary control strategies and their successfulness, life satisfaction, work difficulties, and personality.

Chapter 1 - Literature Review

0.1

Abstract

Although theories of job satisfaction have been extensively researched in the organisational psychology literature, researchers are yet to agree on the major predictors of job satisfaction. Several predictors have been investigated such as needs, values, expectations and specific job characteristics such as job autonomy and job demands. This chapter reviews such theories, focussing on the ones that have made the greatest contribution to the understanding of job satisfaction. Although these theories are well cited, many of them have theoretical and empirical problems as well as having limited applicability to the workplace. One theory, which is less problematic, and particularly appealing to the workplace, is the job demand-control model. This model proposes that job autonomy can reduce the effect of job demands on job satisfaction, and that the most satisfied workers are those with high job demands and high job autonomy. According to the model, job autonomy influences job satisfaction because it allows workers to redirect the physiological arousal produced from job demands into an appropriate response. This explanation is criticised however for being non-specific and tautological. A new explanation is developed, where it is proposed that job autonomy influences how employees respond to work difficulties. This explanation forms the basis of a model of job satisfaction, which includes the following predictors: job autonomy; primary control and secondary control; personality; and life satisfaction.

0.2

Job Satisfaction

Job satisfaction, the extent to which employees like their job and its components (Spector, 1997), is one of the most extensively researched topic in the industrial and organisational psychology literature (Highhouse & Becker, 1993). The number of articles and books investigating this construct has increased from over 3000 in 1976 (Locke, 1976), to over 5000 in 1992 (Harwood & Rice, 1992). Today, a review of psychology and business databases demonstrates that over 10,000 publications on job satisfaction are available. Although this increasing interest in job satisfaction is no doubt beneficial to the field of industrial and organisational psychology, the amount of research has become overwhelming to both researchers and practitioners. Nowhere is this more clearly evident than in the theories of job satisfaction. 0.2.1 Theories of Job Satisfaction: Environmental and Dispositional Predictors Theories of job satisfaction include dispositional and environmental predictors. The dispositional predictors of job satisfaction refer to characteristics of the employee, such as needs, values, and expectations. The environmental predictors refer to job characteristics, such as job control, workload, feedback, role ambiguity, and role conflict. Some theorists focus on the dispositional predictors, whilst others focus on the environmental predictors. More recent theorists recognise the importance of both types of predictors.

Dispositional and environmental theories of job satisfaction have been extensively researched, however researchers have still not reached consensus as to the major predictors of job satisfaction. As a result, researchers continue to rely on theories that have theoretical and empirical problems, or have limited applicability to the workplace. In order to determine which theories are valid and useful, this review will examine the theories that have made the greatest contribution to a shift in focus of the determinants of job satisfaction. These include Maslows (1970) need hierarchy theory, Herzberg, Mausner and Snydermans (1959) two-factor theory of job satisfaction, Vrooms (1964) expectancy theory, discrepancy theories, Hackman and Oldhams (1976) job characteristics model, and Karaseks (1979) job demand-control model.

0.3

Maslows (1954, 1970) Hierarchy of Needs

The need hierarchy theory was one of the first theories to focus on the dispositional predictors of job satisfaction. It proposed that employees needs determine their level of job satisfaction. 0.3.1 Need Hierarchy Theory The need hierarchy theory (Maslow, 1954, 1970) posits that individuals are born with a set of needs. There are five needs: physiological, safety, belongingness, esteem, and self-actualisation. These are arranged in a hierarchy of relative prepotency, meaning that lower-order needs are satisfied before higher-order needs are activated. The lowest need, physiological, refers to basic biological drives, such as hunger, thirst and sex. These physiological needs are the most prepotent of all, as an individual deprived of all needs would seek to gratify these needs first. They would not be concerned with safety, belongingness, esteem, or self-actualisation. Once they have gratified the physiological needs however, the strength of that need decreases, and the next highest need, safety, becomes important. The safety need refers to security, stability, dependency, protection, and need for structure, order, law and limits. To gratify the safety need, an individual requires a safe, orderly, predictable, lawful world. Once the safety need is gratified, its need strength is reduced, and the strength of the belongingness need increases. The individual will begin to hunger for affectionate relationships with people, and for a place in their group or family. Once these belongingness needs are gratified, the

strength of the esteem need increases, and the individual will desire a high evaluation of themselves, and others. Once an individual has gratified these four needs, collectively known as deficiency needs (D-needs), they may begin to feel restless. This restlessness is indicative of the need for self-actualisation. The need for self-actualisation refers to the need for the individual to become everything they are capable of becoming. When the strength of this need increases, the individual strives for self-fulfilment. This fifth need is referred to as a being need (Bneed) because it sustains an individuals interest without being driven by feelings of deprivation. Unlike the previous four needs, when the need for self-actualisation is gratified, it increases in need strength (Maslow, 1962). Growth is a continued upward development, where the more that one gets, the more that one wants. This growth is endless, and can never be attained or satisfied (Maslow, 1962, p. 31). 0.3.2 Applying Maslows (1954, 1970) Theory to Organisations In terms of applying this theory to organisations, the theory proposes that the lower-order needs must be gratified before the higher-order needs are activated. As such, employers must ensure that their employees physiological, safety, belongingness and esteem needs are satisfied. The employer can help the employee to gratify each need. For example, to help them gratify their physiological and safety needs, employers can increase their employees pay. Once these needs are satisfied, the relationship between the employee and their supervisors and co-workers takes on increased strength. The employer can help the employee to gratify this need through increasing the amount of social interaction among employees. This process needs to

be continued until the employees have gratified all of the lower-order needs, and are reaching for self-actualisation, should the nature of the job permit this level to be attained. 0.3.3 Criticisms of Maslows (1954, 1970) Need Hierarchy Theory Almost every aspect of Maslows (1954, 1970) work has been disputed on both theoretical and empirical grounds (Neher, 1991; Wahba & Bridwell, 1976). Five fundamental propositions of Maslows (1954, 1970) theory have been questioned, including: 1) the higher the deprivation of a need, the higher its need strength (i.e., deprivation/domination paradigm); 2) the higher the satisfaction with a need, the higher the need strength of the need at the next level (i.e., gratification/activation paradigm); 3) the measurement of self-actualisation; 4) the ability to achieve selfactualisation; and 5) the applicability of the theory to organisations. Each of these will now be considered. 0.3.3.1 Criticism One: Deprivation/Domination Paradigm The deprivation/domination paradigm postulates that the higher the deprivation of a need, the higher its need strength. An early review concluded that the deprivation/domination paradigm was only partially supported for self-actualisation, and not supported for safety, belongingness and esteem needs (Wahba & Bridwell, 1976). On the basis of this review, many researchers have assumed that the proposition is not supported (Wicker, Brown, Wiehe, Hagen & Reed, 1993). This assumption may be inaccurate however, as many of the studies included in the review

have methodological limitations. These limitations concern: a) the operationalisation of need strength; and b) establishing causality. 0.3.3.2 a) Operationalising Need Strength One of the main limitations in studies examining the deprivation/domination paradigm concerns the operationalisation of need strength. Some researchers measure need strength through desire, others through important or intention. Two studies have measured need strength through desire. In Alderfers (1969) study, subjects were asked to rate how much more of the following factors they would like to have in their jobs; pay, fringe benefits, love, status, and growth. Similarly, in Graham and Ballouns (1973) study, subjects were asked how much improvement they wanted in their physiological, security, social and self-actualisation needs. These measures of need strength were then correlated with corresponding measures of satisfaction. Both studies provided some support for Maslows (1954) theory suggesting that as satisfaction with a need increases, the strength of that need decreases. For example, in Graham and Ballouns (1973) study, the correlations between need strength and satisfaction ranged from r = 0.42 to r = 0.72. Furthermore, in Alderfers (1969) study, satisfaction and need strength were negatively correlated for the relatedness need, which was composed of a respect from co-workers need, and a respect from supervisors need. For the respect from co-workers need, the correlations were all significant, ranging from r = -0.21 to r = -0.38. For the respect from supervisors need, the correlations ranged from r = -0.06 to r = -0.49. Although the correlations in Alderfers (1969) study were in the expected direction, they were

often small, and the correlations between satisfaction and need strength for the belongingness need were insignificant (r = 0.02 to r = 0.07). These two studies appear to provide some support for Maslow's (1954) theory. Both of these studies assessed need strength ratings by desire, where participants were asked how much more they wanted of a need. It must be questioned however, if wanting or desiring more of a need is a measure of the strength of the need. Wanting more of a need may actually be another way of demonstrating dissatisfaction with the area covered by that need. Other researchers have overcome this limitation by assessing need strength using importance ratings, which may be less likely to measure satisfaction. For example, Hall and Nougaim (1968) conducted a longitudinal study on managers, interviewing them annually for five years. The participants rated the importance of, and satisfaction with a number of needs including safety, affiliation, achievement and esteem, and self-actualisation. Inconsistent with Maslows (1954) theory, the correlations between the satisfaction of needs and the importance of needs were positive. For safety, importance and satisfaction correlated r = 0.26, for affiliation r = 0.16, for achievement and esteem r = 0.54, and for self-actualisation r = 0.29. In addition, Hall and Nougaim (1968) also examined the longitudinal changes in satisfaction and importance for each need. According to Maslows (1954) theory, it would be expected that if satisfaction of a need increased from one year to the next, importance of that need would decrease. However, they found that the importance of a need in a given year was positively correlated with its own satisfaction in the previous year. These correlations were moderate for safety (r = 0.25), affiliation

10

(r = 0.21), achievement and esteem (r = 0.53) and self-actualisation (r = 0.28). Although Hall and Nougaim (1968) failed to discuss these correlations in detail, they clearly contradict Maslows (1954) theory. Importance was positively related to need satisfaction, suggesting that a satisfied need is an important need. This finding does not support Maslows (1970, p. 393) proposal that a satisfied need is not a motivator. Although Hall and Nougaims (1968) findings are inconsistent with Maslows (1954) theory, their validity has been questioned. Specifically, the study relied on a small sample, and the interview was not designed to produce data relevant to Maslows (1954) theory (Lawler & Suttle, 1972). Furthermore, the inter-rater reliability of the coding of interviews was low (0.55 to 0.59). A study designed to overcome the limitations identified in Hall and Nougaims (1968) study was conducted by Lawler and Suttle (1972). They employed a reasonably large sample of employees from government agencies and retail stores. Their questionnaire, developed by Porter (1963), was designed to measure Maslows (1954) needs. According to Maslows (1954) theory, the importance of a need should be negatively correlated with satisfaction of that need. Hence, as satisfaction with a need increases, the importance of that need decreases. Lawler and Suttles (1972) results did not support this proposal for either the government or retail organisations respectively, for social (r = -0.09, r = 0.07), esteem (r = 0.06, r = -0.04), autonomy (r = 0.07, r =0.01), and self-actualisation needs (r = 0.01, r = -0.10). There was however, some support for the security needs (r = -0.34, r = -0.12). As their study was longitudinal they also conducted change analyses. They correlated the change in need importance with the change in need satisfaction. It was

11

expected that these correlations would be negative, indicating that increases in the satisfaction of a need were associated with decreases in its importance. However, these correlations were also positive ranging from r = 0.07 to r = 0.24. Hence, the direction of the correlations were inconsistent with Maslows (1954) theory. In summary, Hall and Nougaims (1968) and Lawler and Suttles (1972) findings are inconsistent with those of Alderfer (1969) and Graham and Balloun (1972). The major difference between these studies is that the latter two measured need strength with desire or improvement, while the former two relied on measures of importance. Although the desire and improvement measures were criticised earlier for being too similar to measures of satisfaction, the use of importance as an indicator of need strength has also been criticised (Wicker et al., 1993). Although Maslow (1970) postulates that a need is important because of deprivation, it has been suggested that a person may report that a need is important because they have attained it and value it (Wicker et al., 1993). Indeed, Maslow (1954, p. 148) proposed that greater value is usually placed on higher-order needs by persons who have gratified both kinds (i.e., lower and higher-order needs). Hence, people who are self-actualising may report that all the higher needs are important because they value them. A person may thus report that a higher-order need is important because they are deprived of it, or because they have attained it and value it. If individuals report that a higher-order need is important because they have attained it, it would be positively related to satisfaction (Wicker et al., 1993). Although importance may be an ambiguous construct, the early studies conducted by Hall and Nougaim (1968) and Lawler and Suttle (1972) should still be

12

valid. The majority of participants in these studies would not have gratified both lower-order and higher-order needs. As such, they would only be expected to report that a need was important if they were deprived of the need. Hence, although the early studies tested the deprivation/domination paradigm using importance ratings, this is not expected to reduce the validity of the findings, which are inconsistent with Maslows (1954) theory. More recent researchers have found some support for Maslows (1954) theory using a different measure of need strength, namely intention. Wicker et al., (1993) examined how need strength relates to satisfaction when need strength is operationalised in a number of different ways. They used, among others, ratings of importance (i.e., To what extent is it an important goal) and ratings of intention (i.e., How much do you want to pursue it). They correlated these variables with attainment as a measure of deprivation (i.e., To what extent do you already have it). According to Maslows (1970) theory, it would be expected that as attainment of a need decreased, the intention of that need would increase. However, they found the correlations of past attainment (deprivation) and intention were positive, ranging from r = 0.39 to r = 0.96. This suggests that as attainment of a need increases, the intention to pursue the need also increases. Although Wicker et als., (1993) findings are inconsistent with Maslows (1970) theory, they suggest that the correlations may have been inflated by halo-effects or carryover rating bias. They postulate that the ratings may be affected by a general motivation factor, and by earlier ratings. To control for such effects, deviation scores were computed and correlated. Deviation scores are calculated by

13

subtracting the grand mean over all scales for a need from the mean of that need on each particular scale. This removed a need-means factor from the data, reducing any biasing effect on correlations resulting from mean differences among needs (Wicker et al., 1993, p. 126). Using these deviation scores, the direction of the correlations were reversed. For importance, two of the four correlations were in the expected negative direction, however they were very small (r = -0.13 and r = 0.07). For intention however, all four of the correlations were strong and negative, ranging from r = 0.62 to r = 0.74. This suggests that if need strength is measured through intention, and deviation scores are used, then it is negatively related to attainment. On this basis, Wicker et al., (1993) postulate that it is too early to discard the deprivation/domination paradigm. They propose that participants in earlier studies (e.g., Hall & Nougaim, 1968; Lawler & Suttle, 1972) may have reported that a lower order need was important because they had attained it and they valued it (high satisfaction), or because they were deprived of it (low satisfaction). As a result, the correlations between importance and satisfaction could be positive or negative, depending on how need strength was operationalised. Despite this, it remains concerning that the deprivation/domination paradigm is only supported when need strength is operationalised as intention. 0.3.3.3 b) Establishing Causality A second methodological problem, which may reduce the validity of the studies examining the deprivation/domination paradigm is that although the deprivation/domination paradigm is causal, the relationship is assessed through

14

correlational analyses (e.g., Alderfer, 1969; Graham & Balloun, 1972; Hall & Nougaim, 1968; Lawler & Suttle, 1972). Only one study has attempted to establish causality through experimentally manipulating deprivation and measuring subsequent need strength. Wicker and Wiehe (1999) divided forty students into two groups, where one group wrote about a past event where they felt especially close to another person and the other group wrote about a time when they tried to get close to someone, but felt unsuccessful. Both groups then rated their needs on each level of the hierarchy on prior attainment (i.e., To what extent to do you already have it), intention (i.e., How much do you intend to pursue it), and importance (i.e., To what extent is it an important goal). The interpersonal scenario was expected to affect their belongingness responses, where the unsuccessful group would report lower attainment, and higher need strength for the belongingness need. Inconsistently however, the two groups did not report different levels of attainment on the belongingness need. The two groups did report different levels of esteem attainment where the unsuccessful group reported less past attainment of esteem needs than the successful group. The unsuccessful group also reported higher intention on all levels of the hierarchy than the successful group. The two groups did not however differ on importance ratings. These data were interpreted as supporting Maslows (1970) theory, as when the past attainment of esteem needs were low, intentions were higher. The results must be interpreted with caution however as there were methodological limitations in the study. Aside from each group having a small sample size (N = 20), need strength was not assessed prior to the intervention. Hence, the differences in their intentions may have

15

been a pre-existing difference. Furthermore, although the groups were asked to report a story relating to belongingness needs, the two groups did not report different level of past attainment on belongingness needs. Hence, the belongingness manipulation was not successful. In summary, although Wicker and Wiehe (1999) present their study as supporting Maslows (1970) theory, the findings should be viewed with caution. 0.3.3.4 Summary: Deprivation/Domination Paradigm The deprivation/domination paradigm was rejected after several early studies failed to find supportive correlations. Wicker et al., (1993) re-introduced the proposition into the literature, attributing the inconsistent findings to the operationalisation of need strength. They demonstrated that positive correlations between attainment and need strength could be reversed if deviation scores were used, and need strength was measured by intentions rather than importance. The validity of these findings continues to be questioned however, as the relationship between need strength and satisfaction, although causal has been assessed through correlational analyses. In summary, the majority of research demonstrates that as deprivation increases, need strength does not necessarily increase. 0.3.3.5 Criticism Two: Gratification/Activation Paradigm The gratification/activation paradigm postulates that the higher the satisfaction with a need, the higher the need strength of the need at the next level of the hierarchy. The gratification/activation paradigm is different from the deprivation/domination paradigm as the former examines the correlation between the satisfaction of a need at

16

one level with the importance of the need at the next level, whereas the latter examines the correlation between satisfaction and need strength of a need on the same level. Two longitudinal studies have been conducted to evaluate the gratification/activation paradigm. As previously mentioned, Hall and Nougaim (1968) interviewed managers annually throughout a five-year period, coding their responses on need strength and satisfaction. For each year, they correlated the changes in need satisfaction from one year to the next with changes in need strength at the next highest level during the same period of time. According to Maslows (1954) theory, it was expected that high correlations would exist between the change in satisfaction of a given need level and the change in strength of the next highest level. The pooled correlations were low however, ranging from r = 0.05 to r = 0.22. Hence, there was little evidence to suggest that the increasing satisfaction of a need results in the increasing need strength of the next highest need. It must be noted however that this study relied on a small sample size, and the interview used in the study was not designed to produce data relevant to Maslows (1954) theory. These limitations were addressed in Lawler and Suttles (1972) study. As previously mentioned, Lawler and Suttle (1972) relied on Porters (1963) questionnaire, which was specifically designed to measure Maslows (1954) needs. According to Maslows (1954) theory, it was expected that the satisfaction of a need would be positively correlated with the need strength of the need in the next highest level. Lawler and Suttles (1972) results demonstrated that one correlation between security satisfaction, and social importance was significant for the retail group

17

(r = 0.21), however the rest were all low ranging from r = -0.01 to r = 0.10. These findings, as with Hall and Nougaims (1968) findings clearly raise questions concerning the validity of the gratification/activation paradigm. In summary, the gratification/activation paradigm proposes that as satisfaction with a need increases, the need strength of the next highest need increases. Studies investigating this paradigm generally demonstrate that the correlations between need satisfaction and need strength of the next highest need are low. 0.3.3.6 Criticism Three: Measurement of Self-Actualisation There is a poor level of concordance between the definition of the need for self-actualisation, and the measurement of the need for self-actualisation. Selfactualisation is defined as the full use of ones talents, capacities, potentialities (Maslow, 1970, p. 150). It is the need for the individual to become everything they are capable of becoming. Self-actualisers have a more efficient perception of reality, accept others, are autonomous, do not need others, are less concerned with themselves, and have deeper interpersonal relationships (Maslow, 1970). These characteristics must be regarded with caution however as they were based on a social discussion with a sample of 22 people whom Maslow (1954) believed to be self-actualisers. These people were selected as they seemed to be fulfilling themselves, and doing the best they were capable of. Perhaps as a consequence of this vague definition, operational definitions of the need for self-actualisation vary extensively. Several early studies measured self-actualisation using Porters (1963) need scale (i.e., Lawler & Suttle, 1972; Roberts, Walter & Miles, 1971). This scale includes

18

three items which assess the opportunity for personal growth and development in the job, the feelings of self-fulfilment a person gets from being in the job, and the feelings of worthwhile accomplishment in the job. One problem with these items however, is that they appear to assess how the person feels about their work rather than whether they feel they are have reached their potential. Although more recent scales tend to be more comprehensive, their validity is still questioned. For example, Shoura and Singh (1999) assessed self-actualisation through items measuring meaningfulness, self-sufficiency, effortlessness, creativity, professional creativeness, self-understanding, independence, and harmony with the universe. Examples of these items are do you think you have enough talents and capabilities to perform the job, does your work come as second nature to you and do you feel your job is in harmony with the universe. These items are criticised for being vague, and it is questioned whether they measure if a person has become all that they are capable of. Furthermore, these items only refer to self-actualisation on the job, and in some cases, self-actualisation may occur off the job. In summary, there seems to be a great deal of discrepancy between the definition and measurement of self-actualisation. 0.3.3.7 Criticism Four: Ability to Achieve Self-Actualisation The need for self-actualisation is the need for the individual to become everything that they are capable of becoming. This suggests that anyone performing their job to the best of their abilities is self-actualising. However, Maslow (1970) screened 3000 college students and concluded that only one student was

19

self-actualising. Following this study, Maslow (1970) proposed that self-actualisation of the sort he had found in older adults was not possible for younger developing people. He proposed that young people lack many of the experiences needed for self-actualisation such as identity, autonomy, and romantic relationships. The proposal that younger people do not self-actualise has not received empirical support. A study conducted on engineers demonstrated that the junior engineers reported higher scores on self-actualisation than the senior engineers (Shoura & Singh, 1999). Furthermore, in a study of academics, ranging in age from 30 to 68 years, age and self-actualisation were not related (Hawkins, Hawkins & Ryan, 1989). It must be noted however that, as previously mentioned, these studies relied on questionable measures of self-actualisation. 0.3.3.8 Criticism Five: Applicability of Maslows Hierarchy of Needs to Organisations Although some of the propositions in the need hierarchy theory have not received empirical support, the theory has been extensively accepted in the management literature (Roberts, 1982). Moreover, the general idea that the concepts of love, safety, self-esteem, and growth contribute to motivation and satisfaction are acceptable to both psychologists and management scientists (Shoura & Singh, 1999). The fundamental problem in applying Maslows (1970) theory to work organisations is that little is known about how to reach the ultimate goal of selfactualisation. Maslows (1970, p.46) definition of self-actualisation as what a man

20

can be, he must be is extremely vague, and there is no agreed upon way of operationalising the construct, or facilitating it in employees. 0.3.4 Conclusion The need hierarchy theory proposes that individuals strive to gratify five needs, namely physiological, safety, belongingness, esteem and self-actualisation needs. The theory proposes that the higher the deprivation of a need, the higher its need strength, and the higher the satisfaction with a need, the higher the need strength of the next highest need. Although early studies tended to reject these propositions, more supportive results were found when need strength was operationalised as intentions rather than importance or desire. Even with some supportive findings, the validity of the theory is still questioned as very little is known about the ultimate goal for humans, the need for self-actualisation.

21

0.4

Herzberg, Mausner and Snydermans (1959, 1993) Two-Factor Theory of Job Satisfaction

0.4.1

How the Two-Factor Theory has Contributed to our Understanding of Job Satisfaction The two-factor theory (Herzberg et al., 1959) questioned the assumption that

job satisfaction and job dissatisfaction lie on a single continuum. Rather, the theory proposed that job satisfaction and job dissatisfaction are separate continua, and that the factors which affect job satisfaction are different from the factors which affect job dissatisfaction. 0.4.2 Development of the Two-Factor Theory The two-factor theory is based on a study of accountants and engineers. Through an interview, employees recalled experiences about times when they felt especially good or bad about their jobs, and then rated how seriously their feelings (good/bad) about their jobs had been affected by what happened. Using content analysis, their responses were coded into 14 categories. As demonstrated in Table 1, employees reporting the sources of good times tended to recall events related to achievement, recognition, work itself, responsibility, and advancement. These sources of satisfaction were termed motivator factors. Employees reporting the sources of bad times tended to recall events related to company policy and administration, supervision-technical, salary, recognition, and interpersonal relations with supervisor. These sources of dissatisfaction were termed

22

hygiene factors. An obvious exception to this classification is for the factor salary. Salary was reported a similar number of times for employees reporting the source of good events and for those reporting the source of bad events. On the basis of these findings, Herzberg et al., (1959) proposed that paying attention to motivator factors will increase job satisfaction, but will not affect job dissatisfaction. Alternatively, paying attention to hygiene factors will decrease job dissatisfaction but will not increase job satisfaction. For example, increasing status is expected to reduce job dissatisfaction, but not increase job satisfaction. Table 1- Sources of Good/Bad Times for Accountants and Engineers (N=200) Factor Achievement Recognition Work Itself Responsibility Advancement Salary Possibility of Growth IR-subordinate Status IR-Supervisor IR-Peer Supervision-technical Company policy and administration Working conditions Personal life Job Security Time felt especially good 41** 33** 26** 23** 20** 15 6 6 4 4 3 3 3 1 1 1 Time felt especially bad 7 18 14 6 11 17 8 3 4 15** 8** 20** 31** 11** 6** 1

**p<0.01; Motivator factors are bolded From Herzberg, F., Mausner, B., & Snyderman, B. (1959). The Motivation to Work. (p.60, 72). New York: Wiley. 0.4.3 Criticisms of Herzberg et als., (1959) Theory The two-factor theory is criticised for deducing conclusions from a study that:

23

a) failed to test the main propositions; and b) was methodologically flawed. In regards to the first criticism, there is insufficient evidence to demonstrate how motivator and hygiene factors relate to job satisfaction. Although the study demonstrated that employees recalling good times tended to recall motivator factors, and employees recalling bad times tended to recall hygiene factors, there is no empirical evidence for the proposal that motivator factors can only contribute to job satisfaction and that hygiene factors can only contribute to job dissatisfaction. The study did not measure job satisfaction, and as such, there is no basis for assuming that the factors described in the incidents caused, or were even related to job satisfaction (Ewen, 1964). In regards to the second criticism of Herzberg et als., (1959) theory, concerning the methodology of the study, several problems have been identified. These include: 1) some of the findings contradict the theory; 2) the findings differ depending on the method of data collection; and 3) the hypotheses and criterion measures are ambiguous. These limitations will now be discussed more extensively. 0.4.3.1 Criticism One: Evaluation of Results The results from Herzberg et als., (1959) study did not completely support the theory. As can be seen in Table 1, employees often report motivator factors, such as recognition when they are recalling a time when they felt bad. Although they reported recognition significantly less for bad times than good times, recognition was still the third highest source of a bad time. Furthermore, some of the hygiene factors were reported only slightly more for bad events than good events (i.e., salary, status and job security). Hence, some of the findings are not supportive of the two-factor theory.

24

0.4.3.2

Criticism Two: The Interview Method Replications of Herzberg et als., (1959) study have produced mixed results.

Some researchers have found support for the theory (i.e., Schmidt, 1976), whilst others have contradicted the theory (e.g., Armstrong, 1971; Brenner, Carmack & Weinstein, 1971; Hill, 1986; King 1970; Waters & Waters, 1969). A commonality among the studies that have contradicted the theory is that they have departed from the traditional interview method (Gardner, 1977; Salancik & Pfeffer, 1977). The interview method is criticised for being retrospective, and selective (Gardner, 1977). The employees are expected to more readily recall positive events which reflect upon themselves, and negative events which can be attributed to external conditions (Vroom, 1964). As a result, many researchers have tested Herzberg et als., (1959) theory with rating scales. 0.4.3.3 Rating Scales One example of such a study is Waters and Waters (1969) study of office employees. Rather than using Herzberg et als., (1959) critical incidents interview, employees completed a job satisfaction scale, a job dissatisfaction scale (as these are proposed to be two separate dimensions), and a scale examining satisfaction with specific facets of work. They correlated facet satisfaction with overall satisfaction and overall dissatisfaction. According to Herzberg et als., (1959) theory, it was expected that the motivator factors (i.e., responsibility, work, sense of achievement etc.) would correlate with overall satisfaction more than overall dissatisfaction. This finding was not supported as the pattern of relationships with satisfaction and dissatisfaction were

25

similar (i.e., responsibility of job correlated with satisfaction r = 0.41 and with dissatisfaction, r = -0.37). Similar results were obtained for the hygiene factor, where for example, competent supervision correlated r = 0.44 with satisfaction and r = -0.40 with dissatisfaction, and salary correlated r = 0.43 with satisfaction and r = 0.28 with dissatisfaction. As motivator and hygiene factors acted as both satisfiers and dissatisfiers, this study did not provide support for the two-factor theory. Other researchers who have relied on rating scales have also found that their results fail to support the theory. For example, Brenner et al., (1971) conducted a study on accountants, assessing how much is there now for each motivator and hygiene factor. They correlated each of the items with a measure of overall job satisfaction. Consistent with the two-factor theory, the motivator factors were positively related to measures of job satisfaction, with the correlations ranging from r = 0.39 to r = 0.62. Inconsistently however, the hygiene factors were also positively related to job satisfaction, with the correlations ranging from r = 0.41 to r = 0.59. These findings, fail to conform with Herzberg et als., (1959) theory, and suggest that as motivator and hygiene factors increase, job satisfaction increases. Although Waters and Waters (1969) and Brenner et als., (1971) studies failed to support the two-factor theory using a rating scale, Hills (1986) study claims to offer more support. Hill (1986) developed a 45-item questionnaire to measure intrinsic and extrinsic factors of work in academia. The intrinsic factors (i.e., teaching, convenience, recognition-support) were similar to motivator factors, whilst the extrinsic factors (i.e., economic, administration, and collegial) were similar to the hygiene factors. It was expected that the intrinsic factors would lead to job

26

satisfaction and that the extrinsic factors would lead to job dissatisfaction. To test this proposal, Hill (1986) compared the mean level of satisfaction with each dimension. The employees were more satisfied with the intrinsic dimension (M = 4.43) than the extrinsic dimension (M = 4.18). Specifically, the following means were observed where one is very dissatisfied and six is very satisfied: teaching (M = 4.82), convenience (M = 4.52), recognition-support (M = 3.96), economic (M = 4.24), administration (M = 4.00), and collegial (M = 4.23). From these results, Hill (1986) concluded that the academics dissatisfaction with their work came from extrinsic factors (i.e., hygiene factors), whilst their satisfaction came from intrinsic factors (i.e., motivator factors). The validity of this conclusion is questioned however, as the mean level of satisfaction for the intrinsic and extrinsic factors were very similar. The difference was significant, however this may be due, in part, to the large sample size (N = 1000). More importantly however, it must be questioned whether Hills (1986) study is even testing Herzberg et als., (1959) theory. The two-factor theory did not propose that employees are more satisfied with the motivator factors than the hygiene factors, but rather that the motivators serve to bring about job satisfaction, and hygiene factors prevent job dissatisfaction. As such, although Hills (1986) study claims to support the two-factor theory using a rating scale, the validity of the findings are questioned. In summary, it appears that studies testing the two-factor theory using rating scales tend to be inconsistent with those using the interview method. The rating scale may be superior to the interview method, however it is still problematic (Herzberg,

27

1966; Silver, 1987; Whitsett & Winslow, 1967). Researchers propose that the rating scales may induce respondents to indicate an attitude towards every item, even on items that they have never thought about before (Herzberg, 1966). Furthermore, there is pressure for the respondents to appear rational when they report their satisfaction with the job facets and overall satisfaction, where they may attempt to keep their responses consistent. As a result of these limitations, some researchers have opted for free response scales (e.g., Silver, 1987). 0.4.3.4 Free Response Scales Studies attempting to overcome the limits of both interview and ratings scales have relied on free response scales. These scales are not retrospective, and allow the employee to develop their own answers. For example, Friesen, Holdaway and Rices (1983) study of school Principals relied on two questions including which two factors contribute most to your overall satisfaction with the principalship and which two factors contribute most to your overall dissatisfaction with the principalship. They then calculated how often the Principals mentioned motivator factors and hygiene factors when they referred to sources of their satisfaction and dissatisfaction. These were converted into ratios, which included the number of times each factor was mentioned as a satisfier, and the number of times each factor was mentioned as a dissatisfier (satisfier: dissatisfier). For example, sense of achievement was reported as a source of satisfaction 85 times, and a source of dissatisfaction 5 times (i.e., 85: 5). Other factors that were reported as satisfiers more than dissatisfiers included interpersonal relationship (77: 0), importance of the work (24: 0), and relationship with

28

central office (11: 0). These findings were generally consistent with the two-factor theory, the exception being factors involving relationships (i.e., interpersonal relationships and relationships with central office). Relationship factors are hygiene factors, and as such, are expected to be reported as dissatisfiers more than satisfiers. The factors that were mentioned more as dissatisfiers than satisfiers include amount of work (0: 68), overall constraints (0: 56), attitudes of society (0: 49), stress (0: 21) and impact on home life (0: 14). These were also generally consistent with the two-factor theory. It must be noted however that many other factors were identified as sources of both satisfaction and dissatisfaction, such as relationship with teachers (94: 42), responsibility (81: 20), autonomy (70: 19), student attitudes (51: 25), challenge of work (41: 36), relationships with parents (22: 51) and salary (6: 7). In fact, only eight of the 20 factors occurred uniquely as either satisfiers or dissatisfiers and two of these were in the wrong direction (i.e., interpersonal relationships, relationships with central office). Hence, although researchers have proposed that this study represents a major step in resolving the controversy in favour of Herzbergs assertion (Silver, 1987, p. 5), it provides at best, only partial support. A similar study was conducted on educators by Silver (1987). The participants were required to think of a time when they felt especially good/bad about their jobs, and write a paragraph describing what happened. It was hypothesised that the employees would cite motivator factors more often than hygiene factors when describing positive events, and cite hygiene factors more often than motivator factors when describing negative events. As hypothesised, the employees mentioned more

29

motivator factors (85) than hygiene factors (6), when recalling a positive event. Inconsistently however, the employees reported more motivator factors (48) than hygiene factors (40), when recalling a negative event. As such, Silvers (1987) study provides only partial support for the two-factor theory. Silver (1987) conducted a second study using a questionnaire developed by Wernimont (1966). The questionnaire contained two lists of statements, one positive and one negative, each referring to one of Herzberg et als., (1959) 16 categories. The participants were required to indicate whether an event had occurred, and then to indicate whether it was a positive or negative event. For example, for the pay facet, on the negative list was the pay increase I got was insufficient for putting some aside for the future and on the positive list was I received a substantial increase in pay. It was hypothesised that on the positive-feelings list, respondents would check more motivator than hygiene items, and on the negative-feelings list, respondents would check more hygiene than motivator items. On the positive list, the employees checked 322 motivator factors and 259 hygiene factors, whilst on the negative list, they checked 255 hygiene and 178 motivators factors. These results are assumed to be supportive of the two-factor theory as respondents checked more motivator than hygiene factors on the positive list and more hygiene than motivator factors on the negative list. However, it is concerning that motivator and hygiene factors were reported for both positive and negative events.

30

0.4.3.5

Summary Studies that contradict the two-factor theory tend to depart from the traditional

interview method. These studies, relying on rating scales or free response scales, claim to provide some support for the theory. Closer examination of the results however, demonstrates that these studies provide at best, partial support of the theory. 0.4.3.6 Criticism Three: Ambiguous Hypotheses and Criterion Measures Researchers testing the two-factor theory have been criticised for employing several different hypotheses and criterion measures (King, 1970). First, in regard to the hypotheses, King (1970) cites several different ways that researchers test the main propositions of the theory. Some researchers propose that all motivator factors combined together should contribute more to job satisfaction than job dissatisfaction, and that all hygiene factors combined should contribute more to job dissatisfaction than job satisfaction. Other researchers examine each factor separately, proposing that each motivator factor should contribute more to job satisfaction than job dissatisfaction, and each hygiene factor should contribute more to job dissatisfaction than job satisfaction. A more precise version of the theory proposes that only motivators determine job satisfaction, and that only hygienes determine job dissatisfaction. These examples serve to demonstrate that one researcher using a broad hypothesis may report that their findings support the theory, whilst another researcher using a specific hypothesis may report that their results are inconsistent with the twofactor theory.

31

In regard to the criterion measures, researchers tend to evaluate their findings differently (King, 1970). For example Sergiovanni (1967) conducted a study on teachers using the critical incident technique. The results indicated that teachers reported achievement as a source of a positive event (30) more than a source of a negative event (9). Some researchers, including Sergiovanni (1967) propose that this ratio is supportive of the two-factor theory as it is reported more in positive experiences than negative experiences. However, other researchers (e.g., Friesen et al., 1983) propose that it is not supportive as achievement was reported for some negative experiences. Most researchers opt for the former, proposing that if one part of the ratio is greater than the other part, the results are supportive of the two-factor theory (i.e., Silver, 1987). Even so, these different criterion measures certainly create confusion. It must also be questioned whether a study can provide support for the twofactor theory when some of the ratios are in the wrong direction (i.e., salary 20: 12). Herzberg et al., (1993) did not comment on the issue, however they accepted results that were not in the proposed direction in their study. King (1970) attempted to specify some guidelines, proposing that failure to conform one item would not contradict the whole theory unless that one item had a significant negative difference. However, it still remains unclear how many items would need to be inconsistent for the theory to be refuted. 0.4.4 Conclusion The two-factor theory was notable for proposing that job satisfaction and job dissatisfaction are separate continua, and that the factors which affect job satisfaction

32

are different to the factors which affect job dissatisfaction. The original study from which the theory developed was methodologically flawed, and as such, it is not surprising that empirical studies evaluating the two-factor theory often demonstrate that motivator and hygiene factors affect both job satisfaction and job dissatisfaction. Researchers that report supportive findings often rely on less stringent hypotheses and criterion measures. In conclusion, the two-factor theory of job satisfaction has received little empirical or theoretical support.

33

0.5

Vrooms (1964) Expectancy Theory of Job Satisfaction

0.5.1

How Expectancy Theory has Contributed to our Knowledge of Job Satisfaction Expectancy theory (Vroom, 1964) was one of the first theories to focus on the

cognitive processes that underlie job satisfaction. It has received considerable theoretical and empirical attention for over 30 years (Van Eerde & Thierry, 1996). The number of studies examining expectancy theory has decreased recently however, with only ten studies being conducted since the 1990s (Ambrose & Kulik, 1999). As such, this review will mainly be based on the earlier studies. 0.5.2 Description of Expectancy Theory Expectancy theory describes its major constructs and propositions using its own jargon. It refers to three major constructs, namely expectancy, valence, and instrumentality. Expectancy refers to how much a person perceives that an action will result in a certain outcome. For example, how much a person believes that if they work harder, they will get a pay rise. Valence refers to the degree of anticipated satisfaction or desirability of an outcome. Hence, in the previous example, the valence would be a measure of how much the person desires a pay rise. Instrumentality refers to the degree to which the person sees the outcome in question as leading to the attainment of other outcomes. Hence, in our example, instrumentality would be how much a person believes that a pay rise will result in other outcomes, such as buying a house.

34

The way these constructs are combined depends on the variable that is being predicted. Three dependent variables have been examined, namely job effort, job performance and job satisfaction. This review will only examine the model predicting job satisfaction, referred to as the valence model. This incorporates two of the abovementioned constructs, namely valence and instrumentality. It proposes that job satisfaction can be predicted by multiplying the valence of an outcome by its instrumentality. Hence, to predict job satisfaction, we would need to determine how much a person likes or values an outcome of their job (i.e., being promoted) and multiply this measure by how much they believe that this outcome will lead to other outcomes (i.e., being offered a partnership in a business). There is a great deal of ambiguity surrounding the measurement of the major constructs in the expectancy theory (Van Eerde & Thierry, 1996). The instrumentality construct has proved to be the most troublesome for researchers (Wahba & House, 1978). Vroom (1964) referred to instrumentality as the probability that an outcome will result in other outcomes (i.e., outcome-outcome relationship), and expectancy as the probability that an action will result in an outcome (i.e., action-outcome relationship). Researchers have confused these variables however, and have measured instrumentality through examining the probability that an action will result in an outcome (eg., Constantinople, 1967; Pulakos & Schmitt, 1983; Reinharth & Wahba, 1976). These different conceptualisations of instrumentality influence the application of the valence model to the workplace.

35

0.5.3

Applications of the Valence Model According to the valence model as defined by Vroom (1964), an employer can

increase their employees levels of job satisfaction through ensuring that employees value the outcomes of their job (i.e., gaining admiration from other workers, being promoted, feeling a sense of accomplishment, pay rise), and believe that these outcomes will lead to other outcomes. According to researchers who operationalise instrumentality as expectancy, employers should ensure that their employees value the outcomes of their jobs, and believe that their work will help them achieve those outcomes. 0.5.4 Studies of the Valence Model Several early studies examined the relationship between job satisfaction and the valence model (e.g., Constantinople, 1967; Ferris, 1977; Pulakos & Schmitt, 1983; Reinharth & Wahba, 1976; Sobel, 1971, Teas, 1981). A review of such studies demonstrates that correlations between the valence model (valence x instrumentality) and job satisfaction are generally positive, ranging from r = 0.03 to r = 0.57 (Mitchell, 1974). This demonstrates that together, valence and instrumentality predict job satisfaction. An example of a typical study conducted to assess how the valence model influences satisfaction, is that conducted by Constantinople (1967). This study examined how valence and instrumentality contributed to satisfaction in university students. The students were given a list of 14 outcomes of university (e.g., learning how to learn from books and teachers). Each outcome was rated in terms of its

36

importance (i.e., valence) and on the degree to which the university was helping the students to achieve the outcome (i.e., instrumentality). The product of these two ratings (i.e., instrumentality and valence) was obtained for each outcome, and the products were summed across all 14 outcomes. This measure was then correlated with a measure of satisfaction with college. According to the valence model, it was expected that the valence times instrumentality interaction would be positively related to satisfaction. The results were generally supportive of the model with the correlations ranging from r = 0.34 to r = 0.49. It must be noted however that Constantinople (1967) did not examine how much each component of the model contributed to satisfaction. 0.5.5 Methodological Limitations Although many studies testing Vrooms (1964) valence model claim to provide moderate support for Vrooms (1964) expectancy theory (e.g., Ferris, 1977; Pulakos & Schmitt, 1983; Reinharth & Wahba, 1976; Sobel, 1971, Teas, 1981), these studies have some methodological limitations. Three such limitations have been identified and will be discussed below as: 1) the finding that the components of the valence model account for more of the variance in satisfaction on their own than when combined; 2) violations of the assumptions of the multiplicative composite; and 3) inflated correlations due to common method variance.

37

0.5.5.1

1) The Finding that the Components of the Valence Model Account for more of the Variance in Satisfaction on their own than when Combined. The valence model proposes that job satisfaction can be predicted by the

product of valence and instrumentality. However, many studies have demonstrated that the components of expectancy theory account for more of the variance in satisfaction on their own than when included in the expectancy model (e.g., Pulakos & Schmitt, 1983; Reinharth & Wahba, 1976; Teas, 1981; Van Eerde & Thierry, 1996). In these studies, one of the components, either valence or instrumentality, has predicted job satisfaction as well, or better than, the valence times instrumentality interaction. An example of such a study is that conducted by Reinharth and Wahba (1976). They measured valence and expectancy in a sample of sales force employees. Although instrumentality should have been included in the model, their measure of expectancy was similar to a measure of instrumentality. They measured expectancy by assessing the extent of agreement with the following items; The harder I work, the more I produce, there are no rewards for working hard in this company and poor job performance may get me fired. Their results demonstrated that expectancy was as strongly correlated to job satisfaction (r = 0.43) as the expectancy times valence interaction (r = 0.40). Similar findings were reported in Pulakos and Schmitts (1983) study of graduating students. Valence of work outcomes was assessed through rating the importance of job facets (e.g., good pay, cooperative workers, opportunities for personal growth), and instrumentality was assessed through rating the likelihood of

38

each facet. They correlated these measures with internal job satisfaction and external job satisfaction. The results demonstrated that valence and instrumentality considered separately correlated with job satisfaction as well or better than the valence times instrumentality interaction. For example, in regard to the co-workers facet, the correlations between the valence times instrumentality interaction (r = 0.04, internal, r = 0.11, external) were lower than the correlation for instrumentality considered on its own, (r = 0.11, internal, r = 0.12, external). Hence, in this example, the valence model was not more strongly related to job satisfaction than the components considered separately. A recent meta-analysis of studies using the valence model to predict occupational choice reached similar conclusions (Van Eerde & Thierry, 1996). The results demonstrated that valence (r = 0.27) and instrumentality (r = 0.27) considered separately correlated as well with choice as the valence times instrumentality model (r = 0.28) In conclusion, these studies suggest that the components of the valence model often account for more of the variance in job satisfaction when considered separately rather than when combined into the valence model. These results not only question the usefulness of the two components of the valence model, but also how these components are combined. 0.5.5.2 2) Violations of the Assumptions of the Multiplicative Composite Although the valence model proposes that valence should be multiplied by instrumentality, many assumptions underlying the multiplicative process may not be

39

met. First, although it is assumed that for a multiplication to be valid, the two constructs are independent (Campbell & Pritchard, 1970), instrumentality and valence are related to each other (e.g., r = 0.47; Pritchard & Sanders, 1973). Second, although it is assumed that multiplicative composites are based on a ratio scales with a true zero point (Evans, 1991), most researchers rely on interval scales (Mitchell, 1974). Some researchers have attempted to establish a zero-point on a likert scale by having a scale that ranges from 0 to 10 (i.e., Pritchard & Sanders, 1973). This scale does not have a true zero point, and rather, to establish a true zero point, a complex and timeconsuming process needs to be undertaken, that requires the scaling of pairs, as well as individual outcomes or objects (Thurstone & Jones, 1957). In summary, although the valence model proposes that the components of the model should be multiplied, two major assumptions underlying multiplicative composites may not be met. 0.5.5.3 3) Inflated Correlations due to Common Method Variance Although the assumptions of the multiplicative composites are often ignored, the correlations between the components, considered either on their own or in the valence model, with job satisfaction, are still moderate. Critics suggest that these moderate correlations occur as the measures of instrumentality, valence, and satisfaction are all based on self-report (Schwab, Olian-Gottlieb & Heneman, 1979). It has been proposed that when both the independent variables and dependent variables are measured through self-report, they correlate higher than if one of the variables is observed (Mitchell, 1974; Schwab et al., 1979). The problem with this

40

reasoning however is that self-report measures are expected to differ from objective measures. Objective life satisfaction, for example, is poorly correlated with subjective life satisfaction (r = 0.12; Cummins, 2000a). Thus, the subjective measures cannot be verified through objective measures. Furthermore, it is the subjective measures, which are important to the individuals levels of satisfaction. As long as the employee perceives that by working hard, they will receive a pay rise (instrumentality), and value a pay rise (valence), their satisfaction will be influenced. As such, there is no evidence for the proposal that the correlations among variables in the valence model are inappropriately inflated through common method variance. Rather, the correlations used to make such claims are based on invalid comparisons between objective and subjective variables. Although common method variance is not deemed to be a problem in this regard, researchers have tested the valence model using measures other than self-report. Sobel (1971) conducted a study with students, experimentally manipulating instrumentality. Two groups were formed; one with high instrumentality and one with low instrumentality. Both groups were told that they were required to complete a task of mental agility. Before completing this task, they rated the valence of this task to themselves. They then completed the task, and their score was calculated. They were given a table of norm probabilities which indicated how likely it was that they would perform well on the next task. One group was given a table of norms, which contained high probabilities (i.e., high instrumentality group), whilst the other group was given a table of norms which contained low probabilities (i.e., low instrumentality group). Both groups were then asked to rate their satisfaction whilst

41

considering the importance of the task, and the probability that they would do well in the next task. According to the valence model, it was expected that people with higher instrumentality and higher valence would report higher satisfaction. In regard to instrumentality, the high instrumentality group consistently reported higher satisfaction (M = 19.5) than the low instrumentality group (M = 13.5), thus supporting the model. In regard to the proposed interaction effect, it was expected that for the high or low instrumentality group, people who reported high valence would also report higher satisfaction than people who reported low valence. Inconsistently however, the results demonstrated that for the high instrumentality group, there was no difference in the level of satisfaction reported by the high valence group (M = 20.3) and the low valence group (M = 18.7). Furthermore, for the low instrumentality group, the low valence group reported significantly higher satisfaction (M = 15.0) than the high valence group (M = 12.1). Although these results are generally inconsistent with the valence model, there was a major limitation in the study. The researchers failed to measure instrumentality after the subjects had completed the intervention. As such, they failed to demonstrate that their intervention altered levels of instrumentality. In summary, researchers have suggested that instrumentality and valence correlate well with job satisfaction because they are measured by self-report. Although there is no evidence for this proposition, Sobels (1971) study suggests that when the variables are experimentally manipulated, the results are inconsistent with the valence model.

42

0.5.6

Conclusion Expectancy theory proposes that job satisfaction can be predicted by

multiplying the valence of an outcome by its instrumentality. Reviews conducted on the valence model have demonstrated that the two major components of the model correlate well with job satisfaction. However, the individual components of the model often account for more of the variance in job satisfaction than the multiplicative composite. This has led researchers to not only question the validity of the individual components of the model, but also the validity of the multiplicative composite. In conclusion, while the valence model appears to be simple, it combines a set of complex variables in a problematic manner.

43

0.6

Discrepancy Theories

0.6.1

How Discrepancy Theories have Contributed to our Knowledge of Job Satisfaction Discrepancy theories of job satisfaction focus on the cognitive processes that

underlie job satisfaction. These theories are particularly notable for proposing that employees levels of job satisfaction are dependent on this source of comparison. 0.6.2 Description of Discrepancy Theories Discrepancy theories propose that job satisfaction is a result of a comparison between the perception of the current situation and some standard of comparison (Lawler & Suttle, 1973; Locke, 1969; Michalos, 1985; Porter, 1961). Researchers have defined this standard of comparison in various ways, including what they want, what they feel they are entitled to, what they see others as getting, what they had in the past, or what they expected to have (Harwood & Rice, 1992; Michalos, 1985). In all of these theories however, the larger the difference between the perceptions of the current situation and the standard of comparison, the lower the level of job satisfaction. 0.6.3 Empirical Studies Investigating Discrepancy Theories Although only a few empirical studies have examined the relationship between discrepancy and job satisfaction, they have generally been supportive. For example, Rice, McFarlin and Bennett (1989) measured how much employees have, and want, thirteen job facets. They then calculated the amount of discrepancy between what the

44

employee has and what they want. They found that the perceived have-want discrepancies were moderately negatively correlated with facet satisfaction, where r = -0.48. Hence, as the have-want discrepancy increases, satisfaction decreases. Although Rice et als., (1989) study only examined have-want discrepancies, similar results have also been found for other discrepancies. For example, Harwood and Rice (1992) examined comparisons with: a) co-workers; b) what the person believed that they should have; c) what they expected; and d) what they currently expect. The correlations between these discrepancies and satisfaction, although all in the predicted direction, varied depending on the comparison. The have-want discrepancy was most highly correlated with satisfaction, where the average correlation was r = -0.51. For the have-should have, r = -0.42, have-expected, r = -0.33, have-expect, r = -0.25, and have-co-workers discrepancy, r = -0.22. In summary, studies examining the discrepancies theories are generally supportive. 0.6.4 Theoretical Problems with Discrepancy Theories Although the discrepancy between what a person has and some standard of comparison correlates well with job satisfaction, there are difficulties in using discrepancies to explain satisfaction (Cummins & Nistico, in press). When the discrepancy theory is used to explain job satisfaction in the workplace, the explanation becomes tautological. For example, the theory would propose that an employee has a low level of job satisfaction because they want more from their job. As such, these discrepancies may define job satisfaction rather than explain it.

45

0.6.5

Conclusion Discrepancy theories propose that job satisfaction can be determined by

cognitive comparative processes. Empirical studies have demonstrated that the discrepancy between what an employee has and some standard of comparison is moderately correlated with job satisfaction. However, when discrepancies are used as an explanation of job satisfaction, the explanation becomes tautological.

46

0.7

Job Characteristics Model (JCM; Hackman & Oldham, 1976)

0.7.1

How the Job Characteristics Model has Contributed to our Knowledge of Job Satisfaction The job characteristics model (Hackman & Oldham, 1976) was one of the first

theories to focus on the environmental determinants of job satisfaction. 0.7.2 Description of the Job Characteristics Model The job characteristics model proposes that complex jobs are associated with increased job satisfaction, motivation and performance. It postulates that five core job characteristics are associated with positive outcomes (refer to Figure 1). These include skill variety, task identity, task significance, autonomy, and feedback. Skill variety is the degree to which the job requires employees to use different skills. Task identity is the degree to which the job requires completion of a whole piece of work. Task significance is the degree to which the job has an effect on other peoples lives, and autonomy is the degree to which the job provides freedom. Finally, feedback is the degree to which the job provides clear information about the effectiveness of the employees performance. These five variables do not directly relate to job satisfaction, rather the relationship is mediated by three critical psychological states, including experienced meaningfulness of the work, responsibility for outcomes, and knowledge of results. Scores on these critical psychological states are determined by the five job characteristics. Experienced meaningfulness of the work refers to the degree to which

47

the individual experiences the job as being meaningful and worthwhile. It is determined by skill variety, task identity, and task significance. Experienced responsibility, determined by autonomy, is the degree to which the individual feels accountable for their work. Knowledge of results, determined by feedback, refers to the degree to which the individual is aware of how they are performing the work. These critical psychological states are expected to predict a number of personal and work outcome measures including work motivation, work performance, work satisfaction, absenteeism and turnover. However, the relationship between the critical psychological states and outcomes is mediated by growth need strength. Growth need strength is the need for personal growth and development. It is proposed that individuals with high growth need strength will respond more positively to their critical psychological states than those with low growth need strength.

48

Figure 1-Job Characteristics Model

Core job dimensions

Critical Psychological States

Personal and work outcomes

Skill Variety Task Identity Task Significance Autonomy Feedback

Experienced meaningfulness of work Experienced responsibility for outcomes of the work Knowledge of the actual results of work activities

High internal work motivation High quality work performance High satisfaction with work Low absenteeism and turnover

Employee Growth Need Strength

Source: Hackman, J.R., & Oldham, G.R. (1975). Development of the Job Diagnostic Survey. Journal of Applied Psychology, 60, p. 161.

49

0.7.3

Empirical Studies of the Model The job characteristics model has been extensively researched in over 200

studies (Renn & Vandenberg, 1995), and at least three reviews (Fried & Ferris, 1987; Loher, Noe, Moeller & Fitzgerald, 1985; Roberts & Glick, 1981). These studies have examined four major propositions of the model. These include that: 1) the five core job characteristics contribute to the three critical psychological states; 2) the critical psychological states will mediate the relationship between the job characteristics and the outcome variables; 3) the model is moderated by growth need strength; and 4) the model can be applied to the workplace. These propositions will be examined for only one outcome variable, general satisfaction. General satisfaction is an overall measure of the degree to which the employee is satisfied and happy with their job (Hackman & Oldham, 1975). 0.7.3.1 Proposal One: The Five Core Job Characteristics Contribute to the Three Critical Psychological States As demonstrated in Figure 1, each job characteristic contributes to one critical psychological state. The first three job characteristics (skill variety, task identity and task significance) contribute to experienced meaningfulness. Autonomy contributes to experienced responsibility, and feedback contributes to knowledge. It is proposed that each job characteristic should only correlate with its designated critical psychological state. However, studies examining the relationships among the core job characteristics and the critical psychological states have provided only moderate support for this proposal.

50

Generally, the job characteristics correlate with their designated critical psychological state, but they also correlate with other critical psychological states. For example, autonomy, as expected, correlates with experienced responsibility (r = 0.40, Fox & Feldman, 1988; r = 0.41, Hackman & Oldham, 1976; r = 0.45, Wall, Clegg & Jackson, 1978). However, autonomy also correlates with experienced meaningfulness (r = 0.46, Hackman & Oldham, 1976; r = 0.37, Wall et al., 1978), and knowledge of results (r = 0.25, Fox & Feldman, 1988; r = 0.26, Hackman & Oldham, 1976; r = 0.32, Wall et al., 1978). Another example of a job characteristic that correlates with more than one critical psychological state is skill variety. Skill variety correlates with experienced meaningfulness (r = 0.46, Fox & Feldman, 1988; r = 0.51, Hackman & Oldham, 1976; r = 0.30, Wall et al., 1978) and with experienced responsibility (r = 0.35, Fox & Feldman, 1988; r = 0.40, Hackman & Oldham, 1976; r = 0.22, Wall et al., 1978). These results suggest that, inconsistent with the job characteristics model, the core job characteristics may predict several critical psychological states. 0.7.3.2 Proposal Two: The Degree to which the Critical Psychological States Mediate the Relationship Between the Job Characteristics and the Outcome Variables Although the three critical psychological states are proposed to be the causal core of the model (Hackman & Oldham, 1976, p. 8), only a few researchers have examined the mediation hypothesis (e.g., Arnold & House, 1980; Fox & Feldman, 1988; Hackman & Oldham, 1976; Renn & Vandenberg, 1995; Wall et al., 1978). This

51

hypothesis proposes that the relationship between the five core job characteristics and outcome variables is mediated by the three critical psychological states. This hypothesis has been tested by examining the correlations between the job characteristics and the outcome variables before, and after, the relevant critical psychological states have been controlled for. These results have provided some support for the mediation hypothesis. For example, Hackman and Oldham (1976) found that the correlations between the job characteristics and the outcome variables were lower after controlling for the critical psychological state for skill variety and task significance. However, for autonomy and feedback, the correlations remained moderate (r = 0.29, r = 0.23). These results suggest that the critical psychological states may be partial mediators for only some of the job characteristics. The mediation hypothesis has also been tested using multiple regression analyses. To support the mediation hypothesis, these analyses should demonstrate that the critical psychological states account for sizeable proportions of the variance in each of the dependent variable, and that the core job dimensions add little to this when considered in the same analysis (Hackman & Oldham, 1976). Results have demonstrated that the critical psychological states have accounted for sizeable amounts of the variance in job satisfaction, where R = 0.68 (Hackman & Oldham, 1976), and R = 0.54 (Wall et al., 1978). When the five core job dimensions were added to these analyses, the value of R increased by 0.01 in Hackman and Oldhams (1976) study and by 0.10 in Wall et als., (1978) study. This increase was significant in Hackman and Oldhams (1976) study, suggesting that the variance in the five core job characteristics is explained by the three critical psychological states. It must be noted however that

52

the increase in R was small, and the significance may have reflected that they employed a large sample size (N=658). Although the above studies examined the five job characteristics together, Renn and Vandenberg (1995) examined the job characteristics separately. They examined the effects of the job characteristics before and after the relevant critical psychological state were controlled for. They demonstrated that the effects of the job characteristics were lower in magnitude when the critical psychological states were controlled for, than when considered on their own. For example, when predicting general satisfaction, the partial regression coefficient of task identity on its own was 0.27, and when meaningfulness of work was controlled for, the partial regression coefficient was 0.20. However, the partial regression coefficient representing task identity effects on general satisfaction after meaningfulness was controlled for was still significant (0.20). This was the case for three of the five job characteristics. Specifically, after the relevant critical psychological state was controlled for, the partial regression coefficients for skill variety was 0.08, task identity was 0.20, task significance was 0.15, autonomy was 0.53, and feedback was 0.11. These results concur with the earlier studies that the critical psychological states are only partial mediators of the relationship between job characteristics and outcomes. In summary, although only a few studies have tested the mediation hypothesis, they generally suggest that the critical psychological states are, at best, only partial mediators of the relationship between the core job characteristics and general satisfaction.

53

0.7.3.3

Proposal Three: The Degree to which the Model is Moderated by Growth Need Strength. Growth need strength is a need for personal growth and development. It is

postulated that people who have a high need for personal growth will respond more positively to the critical psychological states than people who have a low need for personal growth. Although an early study conducted by Hackman and Oldham (1976) demonstrated that the relationship between the critical psychological states and general satisfaction was significantly higher for employees with high growth need strength than for those with low growth need strength, later studies have been less supportive (Champoux, 1980; Fried & Ferris, 1987; Tiegs, Tetrick, & Fried, 1992). For example, Tiegs et al., (1992) tested the moderating role of growth need strength with over 6,000 subjects. Using univariate and multivariate hierarchical moderated regression analyses, they demonstrated that growth need strength did not moderate the relationships among job characteristics, critical psychological states, and motivation and affective outcomes. In summary, more recent studies have questioned the moderating role of growth need strength. 0.7.3.4 Proposal Four: Applying the Job Characteristics Model to Work Organisations According to the job characteristics model, an employer can increase job satisfaction through increasing the five job characteristics (e.g., skill variety, task identity etc). Through increasing these job characteristics, the employees critical psychological states will increase, and job satisfaction will subsequently increase.

54

However, as tests of the theory have not examined changes in job characteristics, and the theory does not specify how to make changes to the job characteristics, it may be problematic to apply the job characteristics model to the workplace. Tests of the theory tend focus on naturally occurring variations rather than examining changes in job characteristics. However, the effects of changing the job characteristics for an employee through job re-design may have different effects than if the person was recruited into the already re-designed job (Kelly, 1992). This is important because if the model were applied to a workplace, the five job characteristics would be changed in an attempt to increase job satisfaction. As researchers have not examined the effects of changing the job characteristics, there is little research specifying how to change the job characteristics (Roberts & Glick, 1981). Researchers have attempted to change them using their own techniques, however these have not been particularly successful. Kelly (1992) reviewed such studies, demonstrating that job re-design led to improvements in job satisfaction in 17 out of 30 cases, a distribution that was not significantly different from chance. This suggests that job re-design did not consistently lead to increased job satisfaction. It must be noted however, that in many of the studies, the employees did not alter their perceptions of the job after job re-design. When this finding was taken into account, perceptions of job content and job satisfaction were associated. The important finding from this review however is that job re-design may not change employees perceptions of their jobs. This finding has serious implications for employers intending to implement job re-design. It may be costly and time-consuming

55

to change the job characteristics, particularly if only a few employees recognise and benefit from the changes. 0.7.4 Conclusion The job characteristics model focuses on the environmental determinants of job satisfaction. The model proposes that five job characteristics relate to job satisfaction through influencing three critical psychological states. Empirical tests of the model have provided partial support for the main propositions, however these tests have also demonstrated that many of the relationships that exist between variables were excluded from the model. Even if these relationships were added to the model, practical difficulties in applying the findings to the workplace reduce the usefulness of the theory.

56

0.8

Job Demand-Control Model (Karasek, 1979; Karasek & Theorell, 1990)

0.8.1

How the Job Demand-Control Model Contributes to our Understanding of Job Satisfaction The job demand-control model, developed by Karasek (1979) is one of the

most well known models in the occupational job stress literature (Fletcher & Jones, 1993). Like the job characteristics model (Hackman & Oldham, 1976), it focuses on the characteristics of the job rather than the person. Unlike the job characteristics model however, it proposes that job satisfaction can be increased without altering work demands. 0.8.2 Description of the Job Demand-Control Model The job demand-control model proposes that job satisfaction is a function of the job demands placed on the worker (job demands), and the discretion permitted in deciding how to address those demands (job decision latitude; Karasek & Theorell, 1990). Job demands are the psychological stressors in the work environment (i.e., high pressure of time, high working pace, difficult and mentally exacting work). Job decision latitude is the workers potential control over his/her tasks and conduct during the working day. Using the job demand and job decision latitude dimensions, the job demand-control model predicts four outcomes. Two of these outcomes occur when job demands are high (i.e., active model, high-strain model), whilst the other two occur when job demands are low (i.e., low-strain model, and passive model).

57

The most positive outcomes, including learning and growth, result from active jobs, where both job demands and job decision latitude are high. Although high job demands increase physiological arousal (i.e., increase heart rate or adrenaline), high job decision latitude allows this arousal to be reduced. Workers with high job decision latitude redirect the arousal into an appropriate response. They can choose how they deal with their demands. Through dealing with demands in their own way, they can reduce the arousal. A high strain job is one in which job demands are high and job decision latitude is low. This type of job results in the most adverse reactions of psychological strain (i.e., fatigue, anxiety, depression, physical illness). This is because the arousal from the high job demands cannot be redirected. As the employees have low job decision latitude, they cannot choose how to handle their work demands. As a result, their arousal increases, producing a larger physiological reaction. The two other models are the low-strain model and the passive model. Low strain jobs are those in which job demands are low and job decision latitude is high. These low-strain jobs, although clearly not common, may characterise some selfemployed workers, who only have the occasional customer. Employees in these jobs have a low risk of job strain as they have few demands that produce arousal. Even when they do have the occasional demand, they can redirect the arousal into an appropriate response. Finally, a passive job is one in which both job demands and job decision latitude are low. Employees in these jobs face few challenges and are unable to test ideas for improving the work environment. As a result, they often suffer from reduced work motivation.

58

More recently, in addition to job demands and job decision latitude, Karasek and Theorell (1990) added social support at work. Social support at work is defined as the overall levels of helpful social interaction available on the job from both coworkers and supervisors (Karasek & Theorell, 1990, p. 69). It is proposed that social support at work is positively related to job satisfaction, and that job demands, job decision latitude and social support at work interact to predict job satisfaction. 0.8.3 Empirical Studies of the Job Demand-Control Model Initial tests of the job demand-control model demonstrated that both job demands and job decision latitude predicted a number of dependent variables, including exhaustion, depression, job dissatisfaction, life satisfaction, pill consumption and sick days (Karasek, 1979). Job demands were positively related to these variables, whilst job decision latitude was negatively related to these variables. Replications of Karaseks (1979) study have demonstrated that job demands and job decision latitude separately predict the dependent variables (Dwyer & Ganster, 1991; Fletcher & Jones, 1993; Payne & Fletcher, 1983; Spector, 1987; Warr, 1990). Although these results are supportive of the model, the central proposition of the job-demand control model is that job demands and job decision latitude interact to predict job strain. This interaction effect was tested through regression analyses where the interaction term was added (Karasek, 1979). These analyses demonstrated that job demands and job decision latitude interacted to predict exhaustion, job dissatisfaction, and life dissatisfaction. The following beta values were observed for exhaustion, (decision latitude = -0.004, demands = 0.07, interaction = 0.11), job dissatisfaction,

59

(decision latitude = -0.22, demands = 0.001, interaction = 0.12), and life dissatisfaction, (decision latitude = -0.13, demands = -0.03, interaction = 0.11). Although these interaction terms were significant, the method of analysis was subsequently criticised (Fletcher & Jones, 1993; Ganster & Fusilier, 1989). Researchers propose that Karasek (1979) rejected the traditional tests of interaction based on partialed product terms in regression analyses, and rather relied on variables that reflected differences between demands and control (Fletcher & Jones, 1993; Ganster & Fusilier, 1989). When researchers have replicated these analyses using an appropriate test of the interaction effect specified by Cohen and Cohen (1983), the interaction effect tends to be insignificant (Fletcher & Jones, 1993; Payne & Fletcher, 1983; Warr, 1990). For example, Payne and Fletcher (1983) tested the job demand-control model on secondary school teachers. Using multiple regression they demonstrated that the interaction term did not predict the dependent variables, including depression, anxiety, obsession, somatic symptoms, and cognitive failures. Although these studies suggest that job demands and job decision latitude do not interact to predict job satisfaction, a major problem has been identified in the measurement of job decision latitude. Job decision latitude is defined as the working individuals potential control over his tasks and his conduct during the working day (Karasek, 1979, p. 289-290). However, the most recent measure of job decision latitude, developed by Karasek and Theorell (1990), includes items reflecting decision latitude and decision authority. Decision latitude refers to whether the job involves learning new things, and developing skills. Decision authority refers to whether the

60

person has the freedom to make their own decisions and if they can choose how they perform their work. Although the decision authority items are consistent with the definition of job decision latitude, the decision latitude items have been criticised for measuring skill level, skill variety, and job complexity (Ganster, 1989). Factor analyses of this scale have confirmed that two factors emerge (Smith, Tisak, Hahn, & Schmeider, 1997), namely decision latitude and decision authority. However, only the decision authority items are consistent with the definition. Many researchers have proposed that the definition of job decision latitude is similar to the definition of job autonomy (de Jonge, Breukelen, Landeweerd & Nijhuis, 1999; Ganster & Fusilier, 1989; Spector, 1986). Indeed, Ganster and Fusilier (1989, p. 256) propose that the definition of job decision latitude mirrors job autonomy. Job decision latitude is the working individuals potential control over his tasks and his conduct during the working day (Karasek, 1979, p. 289-290), whilst job autonomy is the degree to which the job provides substantial freedom, independence, and discretion to the individual in scheduling the work and in determining the procedures to be used in carrying it out (Hackman & Oldham, 1976, p.258). As a result of this similarity, researchers have tested the job demand-control model using measures of job autonomy (de Jonge, Mulder & Nijhuis, 1999; Dwyer & Ganster, 1991). It must be noted that these researchers may refer to their scales as measuring job control, however job control and job autonomy appear to be interchangeable (de Jonge et al., 1999b). For example, the interaction effect of the job demand-control model was tested using Gansters (1989, cited in Dwyer & Ganster, 1991) multidimensional control scale. This scale examines the amount of choice employees have in several areas of

61

their work, such as their work tasks, work pacing, work scheduling, physical environment, decision making, interaction and mobility. Using regression analyses, Dwyer and Ganster (1991) demonstrated that the interaction term predicted absenteeism, satisfaction with work, tardiness and sick days. Specifically, the interaction term contributed an additional 15% to explaining the variance in absences, 4% in satisfaction with work, 26% in tardiness, and 4% in sick days. These findings suggest that further research on the job demand control model is required using Gansters (1989, cited in Dwyer & Ganster, 1991) scale. 0.8.4 Conclusion The job demand-control model is intuitively appealing, proposing that job decision latitude can ameliorate job demands. This theory has received partial support as job demands and job decision latitude have separately predicted the dependent variables. Whether these two variables interact to predict job satisfaction continues to be debated. 0.8.5 Extensions on the Job Demand-Control Model Although the evidence for the job demand-control model has been equivocal, there are two main reasons why this theory, over the other reviewed theories, deserves further attention. First, the proposition that job autonomy can somehow ameliorate job demands is certainly appealing to employers (Ganster & Fusilier, 1989). It suggests that employers can increase job satisfaction without altering work demands. Second, although few researchers are continuing to investigate the other theories, the job demand-control model continues to be the subject of many papers (e.g., de Jonge et al.,

62

1999a; Dollard, Winefield, Winefield, & de Jonge, 2000; Hallqvist, Diderichsen, Theorell, Reuterwall, & Ahlbom, 1998; Lu, 1999; Parker & Sprigg, 1999). 0.8.6 Addressing the Gaps in the Job Demand-Control Model Although the job demand-control model deserves further attention, it must be recognised that, in addition to the operationalisation of job decision latitude, there is a major gap in the theory. This involves the explanation of how job decision latitude results in positive outcomes. 0.8.6.1 The Explanation of how Job Decision Latitude Results in Positive Outcomes The model proposes that job decision latitude increases job satisfaction by allowing employees to redirect the physiological arousal produced from job demands. Specifically, Karasek and Theorell (1990) propose that employees with high job decision latitude can translate the physiological arousal produced from job demands into action through effective problem solving. They propose that workers with high job autonomy are given the freedom to decide what is the most effective course of action in response to a stressor (Karasek & Theorell, 1990, p. 36). Job decision latitude gives employees the freedom of action in accomplishing the formal work taskand the freedom to engage in the informal rituals (Karasek & Theorell, 1990, p. 34). A major problem with this explanation however is that it is tautological. This explanation proposes that job decision latitude, or the ability to choose at work, is beneficial because it allows people to choose how they deal with their work demands.

63

Furthermore, the model is non-specific, failing to discuss how the physiological arousal produced from job demands is redirected, and failing to define the most effective course of action. As such, it is unknown how a person with low job decision latitude handles a job demand, and how this is different from a person with high job decision latitude. In response to this criticism, a new explanation for the relationship between job decision latitude and job satisfaction is developed. This explanation specifies how employees with low job autonomy differ from employees with high job autonomy. 0.9 Development of a new Explanation for the Relationship Between Job Autonomy and Job Satisfaction: Influencing Employees Responses to Work Difficulties The job demand-control model proposes that workers with higher job autonomy have higher job satisfaction because they can channel the arousal produced from job demands into an appropriate response. A new explanation is developed which proposes that employees with low job autonomy respond differently to work difficulties than employees with high job autonomy. It must be noted that this explanation focuses on work difficulties rather than job demands. Job demands are the psychological stressors in the work environment (i.e., high pressure of time, high working pace, difficult and mentally exacting work; Karasek & Theorell, 1990). It is expected that job autonomy will influence employees responses to these job demands, but that the hypothesis can be extended to any type of work difficulty. Thus, job autonomy is expected to influence employees

64

responses to their supervisors, co-workers, pay, opportunities for promotion, and so forth. In response to a work difficulty, employees can change the situation to suit themselves, or they can change themselves to suit the situation (Heckhausen & Schulz, 1995; Rothbaum, Weisz & Snyder, 1982). These two strategies are referred to as primary control and secondary control strategies respectively. Before discussing how job autonomy influences the control strategies that employees use, the two strategies will firstly be examined. Specifically, the nature of the strategies will be examined, followed by a discussion of the strategies that people generally use and the most adaptive strategies. 0.9.1 a) Primary Control Strategies and Secondary Control Strategies Two strategies implemented by employees when they face difficult situations are primary control strategies and secondary control strategies (Rothbaum et al., 1982). Primary control involves changing the work environment to suit ones needs, whilst secondary control strategies involve changing oneself to suit the work environment. For example, if an employee felt they were being underpaid, they could use a primary control strategy, such as confronting their employer, or they could use a secondary control strategy and compare themselves to others who are worse off. This conceptualisation of primary and secondary control is similar to Lazarus and Folkmans (1984) conceptualisation of problem-focussed coping and emotionfocussed coping. In this case, coping refers to the constantly changing cognitive and behavioral efforts to manage specific external and/ or internal demands that are

65

appraised as taxing or exceeding the resources of the person (Lazarus & Folkman, 1984, p. 141). Coping strategies are employed to manage the problem causing the distress (i.e., problem-focussed coping) and to regulate the accompanying emotions (i.e., emotion-focussed coping; Folkman & Lazarus, 1980). The theory underlying problem-focussed and emotion-focussed coping and the questionnaire designed to assess these strategies (i.e., Ways of Coping Questionnaire; WCQ; Folkman & Larazus, 1980) is shrouded in methodological limitations (Edwards & ONeill, 1998). First, the definition of coping focuses on managing demands that tax or exceed personal resources. Thus, coping strategies should manage or reduce demands and enhance personal resources to meet demands. The Ways of Coping Questionnaire (Folkman & Larazus, 1980) examines how an individual can cope with a situation by changing the environment or the self, however it is not specified how these strategies manage demands or enhance personal resources (Edwards & ONeill, 1998). Second, there is often a great deal of overlap among the coping dimensions, where some problem-focussed coping strategies resemble emotion-focussed coping strategies (Edwards & ONeill, 1998). Problem-focussed coping is aimed at problem solving, or doing something to alter the situation, however it also includes strategies that alter the self. For example, the problem-focussed coping strategy of shifting ones aspiration level involves the person attempting to move ones goals to be more in line with the current situation (Lazarus & Folkman, 1984). Furthermore, reducing ego involvement involves the person attempting to reduce the overall significance of the situation to oneself (Lazarus & Folkman, 1984). These strategies alter the self and

66

are more consistent with emotion-focussed coping strategies, which aim to reduce the emotional distress associated with the problem. Third, and perhaps most concerning is that factor analyses of the Ways of Coping Questionnaire are highly inconsistent. Researchers have found support for three factors (Parkes, 1984), six factors (Vitaliano, Russo, Carr, Maiuro & Becker, 1985) and eight factors (Aldwin & Revenson, 1987; Folkman et al., 1986; Lazarus & Folkman, 1984). Edwards and ONeill (1998) used confirmatory factor analysis to evaluate these alternative factor structures, concluding that these models yielded poor fit. The conceptualisation of primary control and secondary control is superior to problem-focussed and emotion-focussed coping because it maintains the distinction between changing the environment (i.e., primary control), and changing the self (i.e., secondary control). The control model not only addresses responses to threat and negative events, but also behaviour directed at growth and potential (Schulz & Heckhausen, 1996). Furthermore, the items on the scale are consistent with the definitions of the control strategies. 0.9.2 b) Amounts of Primary Control and Secondary Control The life span theory of control, developed by Heckhausen and Schulz (1995) specifies which kind of strategies people rely on throughout their life. They propose that adults implement both primary and secondary control strategies, however in Western samples, primary control strategies tend to be implemented first, and are generally preferred over secondary control strategies.

67

Research examining the frequency of primary control and secondary controltype strategies in the work environment is generally supportive of Heckhausen and Schulz's (1995) propositions (e.g., Boey, 1998; Koeske, Kirk & Koeske, 1993). For example, Boeys (1998) study on nurses demonstrated that, using a scale ranging from 0 to 4, on average, problem-focussed strategies (M = 2.47) were reported more than emotion-focussed strategies (M = 1.63). In general, theoretical and empirical research suggests that people tend to rely on primary control more than secondary control. The next step is to determine whether primary control strategies are also the most adaptive strategies. 0.9.3 c) Which Control Strategies are more Adaptive for Employees? To determine which control strategies are more adaptive for employees, theoretical and empirical research is examined. The theoretical propositions are based on the life span theory of control (Heckhausen & Schulz, 1995), and the empirical studies specify the correlations between the control strategies and job satisfaction. 0.9.3.1 Theoretical Propositions: The Life span Theory of Control The life span theory of control (Heckhausen & Schulz, 1995) proposes that primary control is more adaptive than secondary control as it allows individuals to meet their own needs. If a person successfully changes their environment using primary control, they overcome their difficulty and also enhance their general perceptions of control. Secondary control strategies are less adaptive than primary control strategies, however they have two main benefits (Heckhausen & Schulz, 1995). They

68

compensate for primary control failure and they assist individuals to focus on goals that expand primary control (Heckhausen & Schulz, 1995). Secondary control compensates for primary control failure, which may threaten self-esteem, self-efficacy, and general perceptions of control (Heckhausen, Schulz & Wrosch, 1997). If an individual experiences repeated primary control failure, they may become vulnerable to experiencing learned helplessness. However, if they implement secondary control after primary control failure, they can protect their selfesteem, and reduce the likelihood of experiencing repeated primary control failure. For example, an individual may face a difficulty at work, where a co-worker is working at a slow pace. To handle this difficulty, they could use a primary control strategy or a secondary control strategy. It is expected that they would firstly implement a primary control strategy, where they may confront their co-worker. They may discuss the problem with them, and the co-worker may agree to put in more effort. If this primary control strategy is successful, they overcome their difficulty. If the strategy fails however and the co-worker continues to work at the same pace, the employee is likely to experience a loss in their general perception of primary control. To avoid repeating this situation, they could implement a secondary control strategy, such as wisdom control, where they think, I cant always get what I want. Through implementing this strategy, they avoid risking repeated primary control failure. Secondary control strategies are also beneficial in assisting individuals to focus on goals that expand primary control. An individual may continue to persist to solve a difficulty if they implement secondary control strategies such as focussing on past success. Through such a focus, the individual may feel more confident in their ability

69

to overcome the problem. In summary, theoretically, primary control is more adaptive than secondary control. 0.9.3.2 ii) Empirical Studies Examining the Adaptiveness of Primary and Secondary Control Empirical studies examining the relationship between primary control and secondary control-type strategies and job satisfaction/job stress have provided some support for the life span theory of control (Boey, 1998; Burke & Greenglass, 2000; Koeske et al., 1993; Kohn, Hay & Legere, 1994; Norman, Collins, Conner, Martin & Rance, 1995). These studies generally demonstrate that primary control strategies are more positively related to job satisfaction than secondary control strategies. For example, Norman et als., (1995) study of teleworkers demonstrated that problem-focussed coping was positively correlated with job satisfaction (r = 0.33) and emotion-focussed coping was negatively related to job satisfaction (r = -0.22). In Burke and Greenglasss (2000) study of nurses, control coping was also positively related to job satisfaction (r = 0.14) and escape coping was negatively related to job satisfaction (r = -0.12). Furthermore, in Kohn et als., (1994) study of teachers, task-oriented coping was negatively related to perceived stress (r = -0.38) and emotion-oriented coping was positively related to stress (r = 0.63). Although these results suggest that, consistent with the life span theory of control, primary control is more adaptive than secondary control, it is important to note

70

however that these studies have conceptualised primary control and secondary-type strategies using several different constructs and scales. For primary control, many of the scales are poorly designed, including items that do not appear to measure primary control-type strategies. For example, Burke and Greenglass (1995) relied on Latacks (1986) measure of control coping. Control coping refers to actions and cognitive reappraisals that are proactive. Many of these items refer to secondary control-type strategies that make the person feel better about the problem. For example, the item try to see the situations as an opportunity to learn and develop new skills is measuring a secondary control strategy known as positive re-interpretation. Furthermore, the items try to think of myself as a winner- as someone who always comes through and tell myself that I can probably work things out to my advantage refers to another secondary control strategy known as illusory optimism. Another scale which confounds primary control strategies with secondary control type strategies is the control coping scale implemented in Koeske et als., (1993) study. Many of the items included in this scale appear to measure secondary control strategies. For example, talked with spouse or other relative about the problem, tried to see the positive side of the situation, got busy with other things to keep my mind off the problem, told myself things that helped me feel better, let my feelings out somehow, and exercising more. These strategies attempt to make the person feel better, rather than change a situation. The secondary control scales have also been criticised for confounding secondary control strategies with primary control strategies. For example, Burke and

71

Greenglass (2000) relied on Latacks (1986) measure of escape coping. This scale included primary control-type strategies, such as delegate work to others and set my own priorities based on what I like to do. Many other scales focused on negative responses, such as avoidance and denial. For example, Norman et al's., (1995) revised version of the COPE scale relied on only five emotion-focussed coping items, such as I use alcohol or drugs to make me feel better and I give up the attempt to get what I want. Furthermore, Boey (1998) measured avoidance coping through items involving suppression of feelings, blaming others, and getting mad at people (i.e., taking more tranquillising drugs, drinking more, avoided being with people in general). Although avoidance and denial are two types of secondary control, there are many other ways that people can change the self to fit in with the environment. Fourteen secondary control strategies have actually been identified in the Primary and Secondary Scale (Heeps, Croft & Cummins, 2000). These strategies, displayed in Table 2, concur with Rothbaum et als., (1982) and Heckhausen and Schulz's (1995) definition of secondary control. In summary, the empirical studies suggest that secondary control-type strategies are negatively related to job satisfaction. However, these findings may not be generalised to secondary control as conceptualised by Heckhausen and Schulz (1995), since these authors recognise that there are many positive secondary control strategies.

72

Table 2- Secondary Control Strategies Item I can see that something good will come of it. I remember you can't always get what you want. I know things will work out OK in the end. I remember I am better off than many other people. I remember I have already accomplished a lot in life. I remember the success of my family and friends. I think nice thoughts to take my mind off it. I tell myself it doesn't matter. I don't feel disappointed because I knew it might Happen. I can see it was not my fault. I ignore it by thinking about other things. I realise I didn't need to control it anyway. I think about my success in other areas. Secondary control strategy Positive re-interpretation Wisdom Illusory-optimism Downward social comparison Past success Vicarious Positive approach Goal disengagement Predictive-negative Attribution Active avoidance Sour grapes Present success

Source: Heeps, L., Croft, C., & Cummins, R.A. (2000). Primary control and Secondary Control Scale (2nded.). Melbourne: Deakin University. 0.9.3.3 Comparing the Life Span Theory of Control and Empirical Studies Examining the most Adaptive Control Strategy The life span theory of control (Heckhausen & Schulz, 1995) proposes that primary control strategies are more adaptive than secondary control strategies as they allow individuals to meet their own needs, and they facilitate general perceptions of control. Secondary control strategies are still useful however in compensating for primary control failure and assisting individuals to focus on goals that expand primary control. The empirical studies partly concur with these propositions, demonstrating that primary control-type strategies are positively related to job satisfaction. Inconsistently however, several studies demonstrate that secondary control type strategies are negatively related to job satisfaction (Boey, 1998; Burke & Greenglass,

73

2000; Friedman et al., 1992; Koeske et al., 1993; Kohn et al., 1994; Norman et al., 1995). It must be noted however, that these empirical studies have relied on many different scales, some of which are methodologically flawed. 0.9.4 Summary Employees implement primary control and secondary control strategies when they face a difficulty at work, however they tend to rely on primary control more than secondary control. Primary control strategies allow individuals to meet their own needs, and are positively related to job satisfaction. Secondary control strategies are assumed to compensate for primary control failure, and assist individuals to focus on goals that expand primary control. Although they have been negatively related to job satisfaction in previous studies, the scales have been criticised for focussing on negative strategies. It is expected that secondary control strategies, as assessed through the Primary Control and Secondary Control Scale (Heeps et al., 2000), will be beneficial for people after they have experienced primary control failure. 0.10 Explaining the Relationship Between Job Autonomy and Job Satisfaction: How Job Autonomy Influences Primary and Secondary Control The explanation for the relationship between job autonomy and job satisfaction proposes that job autonomy influences the way employees respond to their work difficulties. It is expected that job autonomy will influence the use and adaptiveness of primary and secondary control strategies. Past research has tended to confuse job autonomy and primary control (Thompson, Collins, Newcomb & Hunt, 1996) and as such, these two will be

74

differentiated. Job autonomy refers to whether employees perceive that they can control aspects of their work environment, whereas primary control is a strategy that employees use to change their work environment. An employee who has high job autonomy perceives that they can choose, or control many aspects of their work. An employee who has high primary control perceives that they change their environment when they face a difficulty at work. Although job autonomy and primary control are different, they are expected to be related to each another. Specifically, job autonomy should influence: 1), which control strategies employees rely on; and 2) the adaptiveness of the control strategies (i.e., the relationship between the control strategies and job satisfaction). 0.10.1 1) Use of Primary and Secondary Control It is expected that all individuals, with either low or high job autonomy, will implement both primary control and secondary control strategies. Both groups will implement primary control strategies first, and if they experience primary control failure, they will then implement secondary control strategies. The difference between the two groups lies in the amount of primary control failure that they experience. Employees with low job autonomy have little influence over their work environment are expected to experience more primary control failure than employees with high job autonomy. As they need to compensate for this failure, it is expected that these employees (i.e., low job autonomy) will implement more secondary control than employees with high job autonomy. Hence, it is proposed that the ability to choose is

75

inversely related to the probability of primary control failure, which in turn, influences the use of secondary control strategies. These propositions are based on Heckhausen and Schulzs (1995) life span theory of control. This theory proposes that as people age, they experience reduced autonomy, and they begin to experience primary control failure more often. To compensate for this primary control failure, they need to increase their reliance on secondary control strategies. For example, an older individual with restricted mobility may experience primary control failure when working hard to maintain their garden. To reduce the amount of primary control failure that they experience, they can rely on secondary control strategies such as downward comparison (e.g. I am better off than others my age). The proposal that older people rely on more secondary control strategies than younger people has been confirmed in several studies (i.e., Chipperfield, Perry & Menec, 1999; Maher & Cummins, 2001; McConatha & Huba, 1999). 0.10.2 2) Adaptiveness of Primary and Secondary Control In addition to influencing the relative use of primary and secondary control strategies, job autonomy may also influence the adaptiveness of such strategies. Although it was previously demonstrated that primary control-type strategies were positively related to job satisfaction, and secondary control-type strategies were negatively related to job satisfaction, it has been suggested that these relationships may change if the person perceives that the situation is uncontrollable (Thompson et al.,

76

1996; Thompson, Nanni & Levine, 1994; Thompson, Sobolew-Shubin, Galbraith, Schwankovsky & Cruzen, 1993; Thompson et al., 1998). Two models have been developed to explain this relationship, namely the discrimination model and the primacy/back-up model (Thompson et al., 1998). The discrimination model proposes that primary control is more adaptive than secondary control when the situation is controllable, and that secondary control is more adaptive than primary control when the situation is uncontrollable. This model underlies the philosophy of the serenity prayer; Grant me the strength to change what I can, the patience to accept what I cannot, and the wisdom to know the difference (Thompson et al., 1998, p. 587). In regard to job autonomy, this model suggests that primary control strategies are more adaptive for employees with high job autonomy, and secondary control strategies are more adaptive for employees with low job autonomy. The primacy/back-up model, on the other hand, proposes that primary control is more adaptive than secondary control in controllable and relatively uncontrollable situations. The role of secondary control is only to compensate for low primary control, and help increase feelings of overall control (Thompson et al., 1998, p. 587). Thus secondary control is only beneficial when primary control is low. In regard to job autonomy, the primacy/back-up model proposes that primary control strategies are the most adaptive strategy for employees with high job autonomy and for employees with low autonomy, unless primary control is low. It is difficult to differentiate the primacy/back-up model from the discrimination model. The primacy/back-up model proposes that primary control is more adaptive than secondary control unless primary control is low. If a person has

77

low primary control, they believe that they cannot change the environment using an active strategy, such as working hard. However, this means that they perceive the situation to be uncontrollable. Hence, the primacy/back-up model is proposing that secondary control is only useful when primary control is low, however primary control is low when the situation is perceived as being uncontrollable. This is indeed similar to the discrimination model, which proposes that secondary control is best in uncontrollable situations. As such, it appears that there may be some overlap in the models. In order to reduce the overlap in the models, the primacy/back-up model should be revised to propose that primary control is the most adaptive strategy in controllable and uncontrollable situations. The proposal that secondary control is beneficial when primary control is low needs to be excluded as it overlaps with the discrimination model. Researchers who have tested the primacy/back-up model generally focus on the proposal that primary control is adaptive in low-control and high-control situations. 0.10.2.1 Empirical Studies Examining the Discrimination Model and the Primacy/Back-Up Model Only a few studies have examined the most adaptive control strategies in lowcontrol situations (Thompson et al., 1996; 1994; 1993; 1998). In a review of these studies, Thompson et al., (1998) concluded that they generally supported the primacy/back-up model. As several serious methodological problems have been identified in these studies, they will be reviewed.

78

0.10.2.2

Primacy/Back-Up Model

The first study that Thompson et al., (1998) cites as supporting the primacy/back-up model is Thompson et als., (1993) study. According to Thompson et al., (1998), this study demonstrated that cancer patients with higher levels of primary control were less depressed than those with lower levels of primary control. Control was negatively related to maladjustment (r = -0.46) and positively related to physical functioning (r = 0.39) and marital satisfaction (r = 0.24). As such, it was concluded that this study supported the primacy/back-up model (Thompson et al., 1998). However, this conclusion is inaccurate as the study did not measure primary control, rather it measured perceived control. The participants were firstly asked about how much control they had over various facets of their lives (i.e., perceived control over emotions, physical symptoms, relationship with family). They were then asked what type of things they have done to control their feelings over each facet, and how effective these were. The items measuring amount of perceived control were then added to the effectiveness item for each facet. The resulting scale assessed perceived control, and the effectiveness of the control strategies, but clearly failed to measure primary control. When the findings are reinterpreted using perceived control rather than primary control, they are intuitive. It is not surprising that it is beneficial for cancer patients to believe that they can control areas of their lives. Indeed, a fundamental belief about human nature is that we have a need to control events, people and situations (DeCharms, 1968; White, 1959). However, perceived control is not the same as

79

primary control. Whereas autonomy or control refers to whether a person perceives that can change the environment, primary control refers to the specific strategies people use to change the environment to suit their needs. As such, Thompson et als., (1993) study does not adequately test the primacy/back-up model. Another study which claims to support the primacy/back-up model is Thompson et als., (1996) study on HIV-positive men in prison. They examined the relationship between primary and secondary control and distress. Regression analyses demonstrated that primary control was negatively related to distress and secondary control was positively related to distress. Although these findings suggest that people in a low-control environment should rely on primary control, the measurement of primary and secondary control in the study is questionable. To measure primary control, the participants were asked how much control they had over a variety of outcomes, such as their feelings, day-to-day activities, nutrition, and HIV-related symptoms. This measure is criticised however, as the items do not refer to primary control strategies, but rather refer to levels of perceived control. As mentioned previously, primary control is not the same as perceived control. As such, a person may report that they can control their relationship with their cellmates and how their correctional officers treat them, however this does not indicate that they use primary control strategies when they have a difficulty with their cellmates or correctional officers. A further problem with Thompson et als., (1996) study concerns the measurement of secondary control. Secondary control was measured by the following item; How much do you feel okay about (an outcome) because you just accept it and

80

dont try to change it? Although secondary control generally involves acceptance of the situation, this item is criticised as it fails to make respondents aware of the different ways they can accept a situation. For example, they can believe that it will work out okay in the end (i.e., illusory optimism), or they can think that they can't always get what they want (i.e., wisdom). The respondents in Thompson et als., (1996) study were not made aware of these different strategies, and as such may have underestimated their use of secondary control. A further problem with this measure of secondary control is that it does not just ask if the person uses acceptance, rather it confounds acceptance with feeling okay. One final study which claims to support the primacy/back-up model is Thompson et als., (1998). They examined whether adults (young, middle, and older) use primary or secondary control to handle their appearance-related changes due to aging. The youngest group was expected to have the most perceived control over agerelated changes, whilst the oldest group was expected to have the least. Averaging over all age groups, primary control (r = 0.46) and secondary control (r = 0.42) were positively related to satisfaction with physical appearance, and primary control (r = -0.20) and secondary control (r = -0.24) were negatively related to emotional distress. Although these correlations suggest that secondary control is adaptive, multiple regression analyses indicated that secondary control was only beneficial when primary control was low. There was no relationship between secondary control and distress for those with high primary control, but for those with low primary control, secondary control was negatively related to distress.

81

Although these results appear to provide support for the primacy/back-up model, the items measuring primary and secondary control were poorly constructed. Primary control was measured by the following five items rated on a scale from strongly agree to strongly disagree: 1) I feel that I have some control over the effects of aging on my appearance; 2) I dread the thought of aging, but there is not much I can do about it (reversed coded); 3) As long as I put the effort in I can keep looking attractive; 4) I can stay attractive and youthful as long as possible if I just work at it"; and 5) I get depressed when I think about whats coming as I get older (reverse coded). These items are criticised as some of them are based on the assumption that aging is a negative process (items two and five). For example, although Thompson et al., (1998) proposed that a person who disagreed with the item I dread the thought of aging, but there is not much I can do about it has high primary control, it may be that they do not dread the thought of aging. Furthermore, the item I get depressed when I think about whats coming as I get older does not refer to a secondary control strategy, and simply refers to the persons attitude towards aging. Other items are based on the assumption that people perceive themselves as being attractive (items three and four). For example, a person may disagree with the item I can stay attractive and youthful as long as possible if I just work at it, not because they have low primary control, but because they do not believe that they are attractive. The items in the secondary control scale are also criticised for being based on the assumption that aging is a negative process. The scale includes items such as as long as I know whats coming, it doesnt bother me too much to get older and I am

82

not worried about getting older, because I trust that God will take care of me. These items confound the perceptions of aging with the secondary control strategy. As such, it is impossible to tell if the person is referring to the part of the question referring to aging or the part referring to the strategy. For example, a respondent may report that they strongly agree they are "not worried about getting older, because they trust that God will take care of them because they are not worried about getting older, or because they trust that God will take care of them. In summary, as with the other reviewed studies, several measurement issues limit the validity of Thompson et als., (1998) findings. 0.10.2.3 Discrimination Model

One study conducted by Thompson et al., (1994) supported the discrimination model. This study examined the relationship between primary and secondary control and depression for men with a diagnosis of HIV. Both strategies appeared to be adaptive for people who presumably were in a low-control situation. Primary control (r = -0.36) and secondary control (r = -0.41) were negatively related to depression. Furthermore, for the group that was low in primary control, secondary control was negatively related to depression. For those high in primary control, secondary control was weakly related to depression. The values of these correlations cannot be discussed however, as the authors only reported them in graphical form. This study provided some support for the discrimination model, however the measurement of primary control and secondary control strategies was once again limited. The items measuring the control strategies were the same as Thompson et

83

als., (1996) study, where primary control strategies were measured by perceived control and secondary control strategies were measured by acceptance. 0.10.2.4 Conclusion: Does Research Support the Primacy/Back-Up Model or the Discrimination Model? Most of the studies reviewed thus far have concluded that their findings support the primacy/back-up model. However as these studies have often failed to validly measure primary control and secondary control strategies, further research is required to determine whether primary control is adaptive in low-control situations. This research must rely on a measure of primary and secondary control that concurs with Heckhausen and Schulz's (1995) conceptualisation of control. It is expected that employees with low job autonomy will rely on less primary control and more secondary control than employees with high job autonomy. Employees with low job autonomy are expected to have a higher probability of failing when implementing primary control. To compensate for this primary control failure, they can rely on secondary control. 0.10.3 Summary Job autonomy refers to the perceived ability to exert choice in the work environment. It may influence employees use of primary and secondary control, and the adaptiveness of the control strategies. In regard to the use of the control strategies, it is expected that the ability to choose facilitates the probability of primary control failure, which in turn, influences the use of secondary control strategies. In regard to the adaptiveness, empirical studies suggest that primary control strategies are more

84

positively related to job satisfaction than secondary control strategies in low-control situations. These studies are criticised however for their measurement of primary and secondary control, and it is expected that when they are measured validly, primary control is more adaptive in controllable situations, and secondary control is more adaptive in uncontrollable situations.

85

0.11 Other Major Predictors of Job Satisfaction In addition to job autonomy and the control strategies, two other major predictors of job satisfaction are examined. They are personality and life satisfaction. 0.11.1 Personality Researchers have recently paid considerable attention to the role of personality in predicting job satisfaction. The most common taxonomy of personality, the fivefactor model (Costa & McCrae, 1985) includes neuroticism, extroversion, conscientiousness, agreeableness and openness to experience. Researchers have examined how some of these personality variables, namely neuroticism and extroversion, influence levels of job satisfaction. 0.11.1.1 Personality and Job Satisfaction

Personality may directly influence job satisfaction. As evidence for this proposal, researchers have demonstrated that job satisfaction is consistent over time and across situations. For example, Staw and Ross (1985) demonstrated that job attitudes remained consistent over time, even if the person changed employer, and/or occupation. They conducted a longitudinal survey, administering a one-item measure of job satisfaction to over 5000 men in 1966, 1969 and 1971. They correlated the scores on this measure of job satisfaction over time. The correlation between satisfaction scores when the employer and occupation were the same, were moderate (r = 0.37 to r = 0.48). When the employer or the occupation had changed, the correlations were only slightly lower (r = 0.19 to r = 0.34). These correlations provide

86

support for the stability of job satisfaction, however more supportive results were provided by the regression analyses. In the regression analyses, the authors used the 1966 and 1969 job satisfaction scores to predict the 1971 job satisfaction scores. The prior job satisfaction scores (i.e., 1966, 1969 data) were almost always a better predictor of the 1971 job satisfaction scores than situational variables, such as changes in pay and job status. This was even the case when the sample had changed employers but had the same occupation, and when the sample had changed occupation but still had the same employer. Situational variables, including change in pay was a significant predictor of the 1971 job satisfaction score only when the employer and occupation had changed. However, the strength of the relationship was considerably less than prior job attitudes. Hence, Staw and Rosss (1985) study demonstrated that job satisfaction scores could be predicted five years later by earlier job satisfaction scores, even if the individual had changed their employers or changed their occupation. Although Staw and Rosss (1985) study demonstrated that job satisfaction remained stable, the authors did not specifically examine the relationship between personality and job satisfaction. However, Staw, Bell and Clausen (1986) used measures of childhood personality to predict adulthood levels of job satisfaction. They combined three longitudinal surveys, and compared the subjects at early adolescence (12-14 years), late adolescence (15-18 years) and adulthood. They correlated childhood measures of personality with facet job satisfaction, and an overall one-item measure of career satisfaction. The correlations were all positive, ranging from r =

87

0.04 to r = 0.45. Hence, these results suggest that childhood personality is related to job satisfaction in adulthood. To add support to the proposal that job satisfaction is influenced by dispositional variables such as personality, researchers have more recently tested whether there is a genetic component to job satisfaction. Arvey, Bouchard, Segal and Abrahams (1989) studied monozygotic twins who were reared apart. They completed the Minnesota Satisfaction Questionnaire (Weiss, Dawis, England & Lofquist, 1967), which consists of an intrinsic satisfaction scale, an extrinsic satisfaction scale, and a general satisfaction scale. Intraclass correlations, adjusted for age and sex, were significant for intrinsic satisfaction (r = 0.32) and for general satisfaction (r = 0.31). Similar findings were found by Arvey, McCall, Bouchard, Taubman and Cavanaugh (1994) where r = 0.27, and Lykken and Tellegen (1995) where r = 0.44 to r = 0.52. In summary, these findings suggest that job attitudes are consistent over time, that personality measured in adolescence predicts job satisfaction in adulthood, and that there is a genetic component to job satisfaction. The next step is to examine the relationship between specific personality characteristics and levels of job satisfaction. 0.11.1.2 The Relationship Between Neuroticism and Extroversion and Job Satisfaction Neuroticism tends to be negatively related to job satisfaction, where r = -0.29 (Judge, Bono & Locke, 2000), r = -0.18 (Tokar & Subich, 1997), r = -0.25 (Terry, Nielsen & Perchard, 1993), r = -0.21 (Smith, Organ & Near, 1983), and r = -0.40,

88

r = -0.26, r = -0.34 (Hart, 1999). The relationship between extroversion and job satisfaction tends to be much weaker than that of neuroticism. The following correlations between extroversion and job satisfaction have been reported; r = 0.25, r = 0.08, r = 0.18 (Hart, 1999), and r = 0.16 (Tokar & Subich, 1997). In summary, people reporting higher extroversion and lower neuroticism tend to report higher job satisfaction. 0.11.1.3 Summary

Personality appears to be an important predictor of job satisfaction. Research has demonstrated that job attitudes are consistent over time, and that personality measured in adolescence predicts job satisfaction in adulthood. People high on extroversion and low on neuroticism tend to report higher job satisfaction. 0.11.2 Life Satisfaction Researchers have long been interested in the relationship between life satisfaction and job satisfaction (Judge & Watanabe, 1994). Although varying definitions and theories of life satisfaction have been proposed, theoretical and empirical support has been provided for seven domains of life satisfaction. These include material well-being, emotional well-being, productivity, health, intimacy, safety, and community (Cummins, 1996; Felce & Perry, 1995). Before the relationship between life satisfaction and job satisfaction is examined, it will be demonstrated that life satisfaction, like job satisfaction, is influenced by personality.

89

0.11.2.1

Personality and Life Satisfaction

In addition to job satisfaction, neuroticism and extroversion also predict life satisfaction (DeNeve, 1999). Neuroticism is negatively related to life satisfaction where r = -0.29 to r = -0.37 (McCrae & Costa, 1991), r = -0.42 (Costa & McCrae, 1989), and r = -0.46 (Judge et al., 2000). Extroversion is positively related to life satisfaction, (e.g., r = 0.19 to r = 0.22 (McCrae & Costa, 1991), and r = 0.17 to r = 0.20 (Costa & McCrae, 1989). On the basis of these low to moderate correlations, extroversion and neuroticism have been proposed as the key to the relationship between personality and life satisfaction (DeNeve, 1999; Diener, Suh, Lucas & Smith, 1999). 0.11.2.2 Life Satisfaction and Job Satisfaction

Job satisfaction is expected to be related to life satisfaction, as work is a significant and central aspect of many peoples lives. Two models have been developed to explain the linkage between job satisfaction and life satisfaction, namely the spillover model, and the compensatory model (Wilensky, 1960). The spillover model assumes that satisfaction in one domain of an individuals life extends into other areas. Life satisfaction may spillover into job satisfaction or job satisfaction may spillover into life satisfaction. Either way, life satisfaction and job satisfaction would be positively related. Alternatively, the compensatory model proposes that job satisfaction and life satisfaction would be negatively related. An employee with low job satisfaction would be expected to compensate for this by engaging in satisfying non-work activities.

90

A meta-analysis of 34 studies examining the relationship between job satisfaction and life satisfaction demonstrated that the two variables were positively correlated, with an average correlation of r = 0.44 (Tait, Padgett & Baldwin, 1989). Several more recent studies found correlations of similar magnitudes. For example, Iverson and Maguire (2000) found a correlation of r = 0.23, and Beutell and Wittig-Berman (1999) reported a correlation of r = 0.39. Judge, Locke, Durham and Kluger (1998) found that r = 0.68 and r = 0.42, and Landry (2000) found that r = 0.44. These findings demonstrate that job satisfaction and life satisfaction are moderately related, and as such, support the spillover model. Researchers have also examined how life satisfaction relates to the specific facets of job satisfaction (Wright, Bennett & Dun, 1999; Judge & Locke, 1993). Judge and Lockes (1993) study of clerical workers demonstrated that life satisfaction was positively related to all facets of job satisfaction, including nature of work (r = 0.39), co-workers (r = 0.17), supervision (r = 0.26), pay (r = 0.35), and promotion (r = 0.24). In Wright et als., (1999) study of professional card dealers, only satisfaction with pay (r = 0.33) and satisfaction with the work itself (r = 0.28), were related to life satisfaction. Satisfaction with supervision (r = 0.20), satisfaction with promotional opportunities (r = 0.16), and satisfaction with co-workers (r = -0.04) were not significantly related to job satisfaction. Studies generally provide support for the spillover model, and most researchers tend to rely on this model (Rain, Lane & Steiner, 1991). Although it has been suggested that this model may not be appropriate for everyone, Judge and Watanabe

91

(1994) concluded that job satisfaction and life satisfaction were positively related for approximately 80% of the participants in their study. Although these studies have supported the spillover model, the methodology has been criticised. First, common method variance has been identified as a problem as both job satisfaction and life satisfaction are measured by self-report (Rain et al., 1991). This issue is extremely difficult to avoid however as there is no acceptable way to measure attitudes other than self-report. Objective measures of life satisfaction correlate poorly with self-reported life satisfaction (Cummins, 2000a), and behavioural measures of job satisfaction correlate only weakly with self-reported measures of job satisfaction (Iaffaldano & Muchinsky, 1985). The second methodological limitation concerns the cross-sectional study designs, which cannot determine the direction of causality between two variables. Cramer (1995) used cross-lagged correlations to examine a time-related relationship between job satisfaction and life satisfaction over 13 months. Job satisfaction and life satisfaction were positively related at the initial testing and also 13 months later, suggesting that the two variables may be causally related. In summary, the direction of the relationship between job satisfaction and life satisfaction continues to be debated (Iverson & Maguire, 2000). It is generally assumed that job satisfaction contributes to life satisfaction, but it is possible that life satisfaction influences job satisfaction, or that the relationship is reciprocal. It is clear however, that life satisfaction and job satisfaction are positively related. This suggests that people with high job satisfaction will also have high life satisfaction, and that people with low job satisfaction will also have low life satisfaction. However, the

92

relationship may not be quite so straightforward as life satisfaction is held under homeostatic control. 0.11.2.3 Consistency of Life Satisfaction

Recent publications have proposed a model for the homeostatic maintenance of life satisfaction (Cummins, 2000b). The basis of this model is the finding that life satisfaction, when measured either by a single question about satisfaction with life as a whole or by satisfaction averaged across a number of domains, is remarkably predictable. The demonstration of this phenomenon has rested on a statistic called a percentage of scale maximum (%SM) which converts Likert scale data into a range from 0 to 100. Applying this statistic to the combined mean values from large population surveys has revealed that they average 75 + 2.5%SM. In other words, using two standard deviations to define the normative range, it can be predicted that the mean level of life satisfaction of Western population samples will lie within the range 70-80%SM (Cummins, 1995). The consistency of these data provides a basis for the proposal that life satisfaction is held under homeostatic control. The model that describes how such homeostasis can be achieved proposes two levels of influence. The first involves an affective set-point range which is determined by personality. The second level involves a buffering system comprising the three processes of perceived control, optimism, and self-esteem (Cummins, 2000b). Thus, it is proposed, through the interaction of these mechanisms, the average life satisfaction for normative population samples is held within the range 70-80%SM.

93

This model of homeostasis can be used to make predictions about the life satisfaction of employees with low job autonomy and employees with high job autonomy. Provided that their homeostatic systems are operating normally, their life satisfaction is predicted to lie within the normal range. However, the homeostatic system can be defeated by a substantial source of negative input, and the low job autonomy group may have an increased probability of encountering such circumstances. This may be, for example, through exposure to circumstances of reduced personal control. Thus it is predicted that a sample of employees with low job autonomy will contain more people experiencing homeostatic defeat than a sample of employees with high job autonomy. The employees experiencing such defeat are expected to report an average level of life satisfaction that approximates the lower boundary of the normative range (70%SM) or even falls below this level. 0.11.2.4 Summary

As postulated by the spillover model, life satisfaction and job satisfaction are positively related. Although they are expected to co-vary, life satisfaction is held under homeostatic control and may not be free to vary. The average level of life satisfaction reported by employees is expected to lie within 70-80%SM. They may report a lower level of job satisfaction however, if they are experiencing homeostatic defeat.

94

0.12 Model of Job Satisfaction This review has identified five main predictors of job satisfaction. As demonstrated in Figure 2, these include job autonomy, primary control, secondary control, personality, and life satisfaction. The major proposal of the model is that primary and secondary control mediate the relationship between job autonomy and job satisfaction. This is represented by the arrows from job autonomy, through primary and secondary control, to job satisfaction. Primary and secondary control may not account for all of the variance in job autonomy, and thus job autonomy is also directly related to job satisfaction. It is expected that job autonomy influences the use and adaptiveness of the control strategies. In terms of the use of the control strategies, employees with high job autonomy are expected to rely on more primary control and less secondary control than employees with low job autonomy. Employees with high job autonomy are expected to be more successful when implementing primary control, and thus have less need for secondary control, which serves to compensate for primary control failure. In Figure 2, this relationship is represented by the arrow from job autonomy to the control strategies. In regard to the adaptiveness of the control strategies, it is proposed that employees who match their level of job autonomy with their control strategies will be most satisfied with their jobs. It is expected that primary control will be more adaptive for employees with high job autonomy and that secondary control will be more adaptive for employees who cannot control their environment. It is thus expected that job autonomy moderates the relationship between the control strategies and job

95

satisfaction. In Figure 2, this moderation effect is represented by the interaction terms (i.e., job autonomy x primary control, job autonomy x secondary control). These interaction terms are expected to predict job satisfaction. In addition to the control strategies, personality and life satisfaction are expected to directly influence job satisfaction. People higher on extroversion and lower on neuroticism are expected to report a higher level of job satisfaction and life satisfaction. Life satisfaction and job satisfaction are also proposed to influence one another. In summary, the model proposes that job satisfaction can be predicted from job autonomy, primary and secondary control, personality and life satisfaction. This model will be tested in study one, with employees that are low in job autonomy and employees that are high in job autonomy.

96

Figure 2- Model of Job Satisfaction

Job Autonomy x Primary Control

Primary Control

Job Autonomy Secondary Control

Job Satisfaction

Job Autonomy x Secondary Control

Personality

Life Satisfaction

97

1 Chapter 2 - Study One

98

1.1

Abstract

This study tests the model of job satisfaction developed in chapter 1. The major proposal of this model is that job autonomy influences the use of the control strategies and the relationship between the control strategies and job satisfaction (refer to Figure 2). Employees with high job autonomy are expected to rely on more primary control strategies and less secondary control strategies than employees with low job autonomy. Furthermore, primary control is expected to be the most adaptive strategy for employees with high job autonomy, whilst secondary control is expected to be the most adaptive strategy for workers with low job autonomy. These propositions were tested by comparing a sample of high job autonomy employees (university academic staff) with a sample of low job autonomy employees (supermarket register operators). As hypothesised, the academics reported higher job autonomy and lower secondary control than the supermarket workers, however the two groups did not report different levels of primary control. Additionally, primary control appeared to be the most adaptive strategy for both occupational groups, and secondary control was not related to job satisfaction. These findings are discussed in relation to the life span theory of control and the discrimination model.

99

1.2

Proposal for Study One

The model of job satisfaction developed in chapter 1 proposes that job autonomy relates to job satisfaction through influencing the use of the control strategies, and the relationship between the control strategies and job satisfaction. The model also proposes that personality and life satisfaction predict job satisfaction (refer to Figure 2). In order to test these propositions, study one will compare workers with low job autonomy with workers with high job autonomy. The first step is to identify what type of employees fit into these two groups. 1.2.1 Identifying Employees with Low/High Job Autonomy According to Gansters (1989, cited in Dwyer & Ganster, 1991) scale, employees with high job autonomy can exert choice in several domains of their work, such as in the scheduling of their rest breaks, procedures and policies, and in the variety of tasks they perform. An occupational group that appears to exemplify high job autonomy, is university academic staff. Their level of job autonomy has rarely been assessed (Leung, Siu, & Spector, 2000), however academics have traditionally had flexibility in their work, and freedom to pursue their own research interests (Winefield, 2000). They can often choose among a variety of tasks, including research, teaching, and administration (Fisher, 1994). It is particularly important to study academics level of job autonomy as researchers have recently suggested that although in theory, the freedom indicative of high control still exists, in practice, there has been a steady erosion of job control (Fisher, 1994, p. 61). This has been attributed to the increasing demands placed on

100

academics, where their workloads have increased and there is increasing pressure to attract external funding (Winefield, 2000). However, the current study proposes that even if their level of job autonomy is decreasing, they should still be in the upper range. Employees in the lower range of job autonomy are those that have little opportunity to exert choice in their work. They tend to have routinised jobs and have few tasks from which to choose. Supermarket register operators were selected as representing such low autonomy workers. These workers are expected to have little control over many aspects of their job, such as their rest breaks, the tasks they work on, and their working pace.

101

1.3

Aims and Hypotheses

This study aims to compare the levels of job autonomy, control strategies, and job satisfaction of supermarket register operators with academics. A number of hypotheses have been developed as follows:

1) Job autonomy will be positively related to job satisfaction, and academics will report higher job autonomy than supermarket workers. This hypothesis tests the basic assumptions of the study. The study aims to extend Karaseks (1979) job demand-control model, proposing that employees with high job autonomy have high job satisfaction because they rely on different control strategies. As such, it needs to be demonstrated that job autonomy is related to job satisfaction, and that the two groups selected for this study differ in their levels of job autonomy as expected.

2) The academic group will report less secondary control and more primary control than the supermarket workers. This hypothesis examines how job autonomy influences the use of the control strategies. As the academics are expected to have more control over their working environment than the supermarket workers, they are more likely to successfully change it using primary control. As secondary control is used to compensate for, and avoid future primary control failure, it is expected that the supermarket workers will report more secondary control than the academics.

102

3) Job autonomy is positively related to primary control and negatively related to secondary control. This hypothesis also tests whether job autonomy influences the use of the control strategies, however, unlike hypothesis two, it is based on the measured level of job autonomy rather than the assumed level.

4) Primary control will be more positively related to job satisfaction than secondary control for the academics, and secondary control will be more positively related to job satisfaction than primary control for the supermarket workers. This hypothesis tests whether job autonomy influences the relationship between the control strategies and job satisfaction. According to the discrimination model (Thompson et al., 1998), primary control is most adaptive in controllable situations and secondary control is most adaptive in uncontrollable situations. Although empirical studies have generally failed to support this model, the studies have been criticised for their measurement of primary and secondary control strategies.

5) The relationship between primary and secondary control and job satisfaction will be moderated by job autonomy. It is expected that the relationship between primary and secondary control and job satisfaction will change depending on the level of measured job autonomy. This hypothesis is similar to hypothesis four, however, rather than being based on the assumed level of job autonomy, it is based on the measured level of job autonomy.

103

6) The relationship between job autonomy and job satisfaction is mediated by primary and secondary control strategies. This hypothesis tests the proposed explanation for the relationship between job autonomy and job satisfaction. It offers an alternative to Karasek and Theorells (1990) explanation of the job demand-control model. They propose that job decision latitude (i.e., similar to job autonomy) is positively related to job satisfaction because it gives workers the freedom to choose how they complete their work and thereby reduces the arousal produced from job demands. An alternative explanation, to be tested here, is that workers with high job autonomy mostly rely on the preferred control strategies, namely primary control.

7) Academics will report higher job satisfaction and higher life satisfaction than the supermarket workers. As the academics are expected to report higher job autonomy and more primary control than the supermarket workers, they are expected to report a higher level of job satisfaction. This level of job satisfaction is expected to be positively related to life satisfaction.

8) Primary control, secondary control, job autonomy, personality, and life satisfaction will predict job satisfaction. These variables are assumed to be the major predictors of job satisfaction, as depicted in Figure 2.

104

1.4

Method

1.4.1

Participants The sample consisted of 104 university academic staff, and 96 supermarket

register operators. The academic group was obtained from seven Schools within Deakin University. The response rate was 32%. The supermarket workers group was obtained from two supermarket chains, with 16 stores being involved. As these employees only collected a questionnaire if they were interested in participating in the study, a response rate could not be calculated. 1.4.2 Materials Both the academics and the supermarket workers received a plain language statement (refer to Appendix A) and an anonymous questionnaire. The questionnaire consisted of several scales, which measured job autonomy, job related primary control and secondary control, job satisfaction, life satisfaction and personality. 1.4.2.1 Job Autonomy Although this study is examining the job demand-control model, Karasek and Theorells (1990) scale of job decision latitude was not used. This scale is criticised for confounding job control with skill level, skill variety, and job complexity (Ganster, 1989). In response to this criticism, Ganster (1989, cited in Dwyer & Ganster, 1991) developed an work control scale which examined the amount of choice an employee

105

has in several areas of their work, such as their work tasks, work pacing, work scheduling, physical environment, decision making, interaction, and mobility. Although the scale has good reliability (Fox, Dwyer & Ganster, 1993; Ganster, Dwyer & Fox, 2001; Schaubroeck & Merritt, 1997), a factor analysis demonstrated that two factors emerged (Smith et al., 1997). One factor included items on job autonomy (16 items), while the other factor included items on predictability (5 items). The predictability items include how much can you generally predict the amount of work you will have to do on a given day and how much are you able to predict what the results of decisions you make on the job will be. As these predictability items load on a different factor from the job autonomy items, they should be excluded from the scale. Hence, for the purpose of the present study, only the former items were used. A further potential problem with this scale is that some of the items directly refer to control. In an attempt to disguise the purpose of the scale, these items were changed from control to choice. For example, the item how much control do you have over the quality of your work was changed to In my job, I can choose the quality of my work. Furthermore, to reduce the number of items in the scale, two repetitive items were deleted. The items how much control do you have over when you come to work and leave and how much control do you have over when you take vacations or days off were deemed to be too similar to the following item; how much control do you have over the scheduling and duration of your rest breaks. All items refer to the timing and scheduling of rest breaks, and as such, only the latter item was retained.

106

This revised job autonomy scale consisted of 14 items (refer to Appendix B). They were rated on a 10-point scale ranging from 1 (do not agree at all) to 10 (agree completely). 1.4.2.2 Primary Control and Secondary Control The control strategies were assessed by the Primary and Secondary Control Scale, developed by Heeps et al., (2000). This includes five items assessing primary control strategies and 14 items assessing secondary control strategies. All of these items were revised to be relevant to the workplace (refer to Appendix C). They were measured on a 10-point scale, ranging from 1 (do not agree at all) to 10 (agree completely). Although this scale was only developed recently, early factor analyses suggest that two factors emerge (Maher & Cummins, 2001; Misajon & Cummins, in press). This scale was deemed to be superior to Heckhausen et als., (1997) Optimisation in Primary and Secondary Control Scale (OPS). The reason this scale was not selected merits discussion, as the proposed model of job satisfaction is partly based on the life span theory of control. In this theory, Heckhausen and Schulz (1995) propose that humans face two challenges in life; the need to be selective and the need to compensate for failure. On this basis, the Optimisation in Primary and Secondary Control Scale (Heckhausen et al., 1997) measures two types of primary and secondary control; selective and compensatory. Selective primary control is the investment of resources to reach goals, whilst compensatory primary control is used when internal resources are insufficient

107

(i.e., others help, technical aids). Selective secondary control refers to selfmanagement directed at enhancing commitment to goals, and compensatory secondary control serves to buffer the negative effects of failure. An alternative simpler explanation for the use of primary and secondary control is offered. Rather than assisting with the need to be selective and the need to compensate for failure, it is proposed that people use control strategies whenever they risk losing control. Examples of such situations are when people are unable to solve a problem, or when something bad happens to them. Primary control provides a sense of control derived from changing ones realities, whereas secondary control provides a sense of control derived from accepting or adjusting to ones realities (Halliday & Graham, 2000; Thompson et al., 1994). As this new explanation is not consistent with the Optimisation in Primary and Secondary Control Scale (Heckhausen et al., 1997), this scale was not appropriate for this study. The scale includes some situations, which do not appear to prompt the use of control strategies, such as when I have decided on something. The scale also includes statements that are assessing general beliefs rather than strategies. For example, I invest my time in developing broad skills that can be used in many areas, I stay active and involved in several different areas of life, and many life goals become important to me because it is the right time for them. As a new explanation for the use of control strategies has been developed, the Optimisation in Primary and Secondary Control Scale (Heckhausen et al., 1997) is no longer appropriate. As such, the Primary and Secondary Control Scale (Heeps et al.,

108

2000) will be used in this study. This scale examines how people react to situations where they risk losing control. 1.4.2.3 Job Satisfaction Two scales of job satisfaction were administered; a facet scale and a global scale. The facet scale is a revision of the Job Descriptive Index (Smith, Kendall & Hulin, 1969). This scale, reported to be the most frequently used measure of job satisfaction (Ironson, Smith, Brannick, Gibson & Paul, 1989), assesses five facets of job satisfaction. The scale contains 72 items assessing nature of work, supervision, pay, co-workers, and opportunity for promotion. This scale is reliable and convergent and discriminant validity has been demonstrated (Gillet & Schwab, 1975; Johnson, Smith & Tucker, 1982). This scale has been criticised however, as the items have not been revised since the scale was developed. In response to this criticism, Roznowski (1989) developed a revised scale by calculating the discriminating power of the existing items, as well as some new items. This revised scale had higher reliability with the alpha coefficient ranging from 0.81 to 0.91. Although this revised scale may be more relevant to todays workforce, it still contains 72 items. To reduce the number of items for the current study, a further revision was made. Only three items were selected to measure each facet (refer to Appendix D). These items were selected as they had the highest discrimination power. This facet measure of job satisfaction is useful to diagnose the strengths and weaknesses of organisations, however it cannot be summed to produce an overall measure of job satisfaction (Ironson et al., 1989). Many researchers continue to use

109

facet scales to obtain an overall measure of job satisfaction (e.g., ODriscoll & Beehr, 2000; Schappe, 1998), however facet scales have been criticised as they may exclude areas that are important to the respondent, or include areas that are unimportant to the respondent. Therefore, in addition to the facet measure, a global item of job satisfaction was also used. The global measure of job satisfaction is a one-item measure. The item is taking into consideration all the things about your job, how satisfied are you with it? This item was rated on a 10-point scale ranging from 1 (completely dissatisfied) to 10 (completely satisfied). This global scale requires the respondent to combine their reactions to various aspects of the job into a single response. When answering this question, the respondent may incorporate aspects of their job not included in the facet scale. Although internal consistency cannot be established, a meta-analysis of singleitem measures of job satisfaction has demonstrated that single-item measures correlate with other measures, such as the Job Diagnostic Survey (Hackman & Oldham, 1976), the Job in General Scale (Ironson et al., 1989), and the Minnesota Satisfaction Questionnaire (Weiss et al., 1967). On average, the correlation between other scales and single item scales was r = 0.63 (Wanous, Reichers & Hudy, 1997). 1.4.2.4 Life Satisfaction The subjective dimension of the Comprehensive Quality of Life Scale (ComQOL) developed by Cummins (1997) assesses satisfaction with seven domains of life, including material well-being, health, productivity, intimacy, safety, community, and emotional well-being (refer to Appendix E). An 11-point scale was utilised, ranging

110

from 0 (completely dissatisfied) to 10 (completely satisfied). The scale is psychometrically sound, with internal reliability ranging from 0.5 to 0.8 (Cummins, 1997) and validity has been established using data from a review of the QOL domains (Cummins, 1996). 1.4.2.5 Personality The neuroticism and extroversion subscales of the NEO Five Factor Inventory (short form; Costa & McCrae, 1992) was used to measure personality. This scale contains 12 items to measure extroversion and 12 items to measure neuroticism (refer to Appendix F). Six facet scales are measured in each factor. Neuroticism is the sum of scales measuring anxiety, angry hostility, depression, self-consciousness, impulsiveness, and vulnerability. Extroversion is the sum of warmth, gregariousness, assertiveness, excitement-seeking, and positive emotions. Convergent and discriminant validity of both of these factors has been established (Costa & McCrae, 1992, Leong & Dollinger, 1991; Tinsley, 1994). 1.4.3 Procedure Ethics approval was obtained from Deakin University, and consent was obtained from the Heads of School for the academics, and the Human Resource Managers for the supermarket workers. The recruitment procedure differed depending on the group. The academics were sent the questionnaire via the internal mail system. They returned the questionnaire by post. The supermarket workers questionnaires were left in the staff room. As they were expected to complete the questionnaire outside of work time, a $5 lottery ticket was given to each participant to thank them for

111

their time. The lottery tickets were given to the managers of the store. The supermarket workers that returned the questionnaire collected their lottery ticket from the service desk. At the conclusion of the study, the participating Heads of School and the Human Resource Managers received a summary of the aggregated results for all academics and supermarket workers.

112

1.5

Results

1.5.1

Data Screening and Checking of Assumptions Procedures for data screening, and checking the procedural analytic

assumptions for all dependent variables followed those appropriate for group data. The data set was initially examined for missing values, acquiescence, outliers, normality and linearity. In regard to missing values, less than 4% of the values for academics, and less than 5% of the values for supermarket workers were missing for any one item. As there was no pattern to these missing values, they were replaced with the group mean. Although this reduces the variance of the variables and bivariate correlations (Tabachnick & Fidell, 1996), the replacement is a conservative estimate. Once the missing values were replaced, the data set was examined for participants consistently reporting extreme scores (i.e., 1 or 10), in an attempt to reduce the influence of acquiescence. One participant was omitted from the entire sample for consistently reporting extreme scores on every scale. Other participants reporting extreme scores on just one scale were excluded from that particular analysis. Specifically, seven participants (all supermarket workers) responses were deleted from the life satisfaction analyses, and nine participants responses were omitted from the primary control analyses (three academics, six supermarket workers). Univariate outliers were identified on the facet job satisfaction scale, the life satisfaction scale, and the control scales. Specifically, five cases of job satisfaction, 12 cases of life satisfaction, three cases of primary control, and nine cases of job autonomy, lay outside three standard deviations from the mean. As these cases are

113

from the intended population, yet have more extreme values than the normal distribution, they were recoded to three standard deviations from the mean. On completion of the screening process, normality was assessed using the skew/standard error <3, Kolmogorov-Smirnof values, frequency histograms, and normal probability plots. In the academic group, overall life satisfaction (-3.91) and the one-item measure of job satisfaction (-5.60) were mildly negatively skewed. In the supermarket workers group, primary control was negatively skewed (-3.27). As transformations are not recommended for data that are mildly and naturally skewed (Tabachnick & Fidell, 1996), these data were not transformed. Finally, homoscedasticity and linearity were assessed through bivariate scatterplots and these appeared to demonstrate reasonable linear relationships between the variables.

114

1.5.2

Descriptive Statistics and Inter-Correlations Table 3 contains the means and standard deviations for the major variables in

the study for each occupational group. In this table, and the tables thereafter, all mean scores are converted to a percentage of scale maximum (%SM) which ranges from 0100. The formula is:

%SM = (mean score for the original domain 1) x 100/ (number of scale points 1). Table 3 demonstrates that the academics report slightly higher job satisfaction, primary control and job autonomy, and lower secondary control than the supermarket workers. Table 4 displays the correlations among all of the major variables for the academics and the supermarket workers. This demonstrates that job autonomy and primary control are positively related to job satisfaction for both occupational groups. Table 3- Means and Standard Deviations of Major Variables for Academics and Supermarket Workers Variable Job Satisfaction - 1 item Job autonomy Primary Control Secondary Control Life Satisfaction Neuroticism Extroversion M 66.05 51.94 71.56 36.63 78.22 36.60 61.71 Academics SD 21.09 14.63 11.95 15.64 10.96 15.78 12.41 Supermarket Workers M SD 59.71 25.69 34.50 20.24 67.06 18.62 46.74 19.77 73.30 15.97 39.25 17.88 65.28 13.99

115

Table 4- Inter-Correlations for the Academics and the Supermarket Workers JS JS JA PC SC LS Neu Ext 0.41** 0.44** 0.04 0.20* -0.27 0.09 JA 0.25* 0.43** 0.08 0.15 -0.19* 0.09 PC 0.38** 0.43** -0.02 0.11 -0.14 0.25** SC 0.14 0.03 0.07 -0.01 0.05 -0.12 LS 0.07 0.06 -0.02 -0.07 -0.59** 0.25** Neu -0.23* -0.16 -0.03 0.04 -0.50** -0.17* Ext 0.17 0.17 0.42** 0.02 0.29** -0.30**

* p<0.05 , ** p>0.01; Correlations for supermarket workers are bolded. JS = Job satisfaction; JA = Job autonomy; PC = Primary control; SC = Secondary control; LS = Life satisfaction; Neu = Neuroticism; Ext = Extroversion

116

1.5.3

Factor Analyses Prior to testing the hypotheses, factor analyses were conducted on the revised

scales of job satisfaction, primary and secondary control, and job autonomy. 1.5.4 Factor Analysis of the Job Descriptive Index To ensure the 15 job satisfaction items represented each of the five facets, a principle components analysis with direct oblimin rotation was conducted. The assumptions of sample size, normality, outliers, linearity, and the factorability of the correlation matrix were initially examined. Factor analysis requires a minimum of five subjects per variable (5 x 15 = 75) (Coakes & Steed, 1999), hence the sample size of 199 was adequate. Some of the job satisfaction items were not normally distributed however the solution is still worthwhile if normality is not met (Tabachnick & Fidell, 1996). Five outlying cases were recoded to three standard deviations from the mean. Reasonably linear relationships existed among the variables. In regard to the factorability of the correlation matrix, all of the correlations exceeded 0.30. The measures of sampling adequacy (MSA) were > 0.50. Bartletts test of Sphericity was large and significant (1574.89), and Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy exceeded 0.60. A principal components analysis, with direct oblimin rotation, yielded four eigenvalues over 1. With this four factor solution 22% of the nonredundant residuals had absolute values > 0.05, suggesting the presence of another factor. When a principal components analysis with direct oblimin rotation was conducted with five

117

factors, only 7% of the nonredundant residuals had absolute values > 0.05. As such, a five-factor model was deemed to be most appropriate. The loadings of the items on each of the five factors are presented in Table 5. These loadings demonstrate that Factor 1 refers to promotion, Factor 2 to nature of work, Factor 3 to pay, Factor 4 to supervisors, and Factor 5 to co-workers. Table 5- Factor Analysis of Job Satisfaction Scale Item There is a good chance for promotion in my job. There is a fairly good chance for promotion in my job. There are good opportunities for advancement in my job. My work is dull. My work is boring. My work is interesting. My pay is bad. I am well-paid. My pay is unfair. My supervisors know how to supervise. My supervisors are bad. My supervisors are annoying. My co-workers are stupid. My co-workers are responsible. My co-workers are a waste of time. Eigenvalues % of variance Cumulative % Cronbach's Alpha (total scale) F1 0.94 0.94 0.90 0.97 0.90 0.81 0.91 0.82 0.81 0.81 0.79 0.78 0.79 0.73 0.79 0.94 6.27 77.04 F2 F3 F4 F5

4.32 28.77 28.77 0.82

2.78 18.56 47.32

1.87 12.46 59.79

1.65 10.99 70.77

Items with loadings less than 0.30 are not shown. 1.5.5 Factor Analysis of the Primary and Secondary Control Scale To ensure that the primary and secondary control items loaded on two separate factors, a factor analysis was conducted on the primary and secondary control scale.

118

The sample size was adequate (N = 190), and the secondary control items were normally distributed for both groups. The primary control items were mildly negatively skewed for both the academics (pc4 = -4.41, pc5 = -3.17) and the supermarket workers (pc1 = -3.5, pc2 = -3.77, pc4 = -4.26, pc5 = -3.17; refer to Table 7 for items). However, as factor analysis is robust to violations of normality, the resulting solution was still deemed to be worthwhile (Tabachnick & Fidell, 1996). Linearity among the variables as assessed through scatterplots was reasonable. The correlation matrix was factorable with all correlations exceeding 0.30. The measures of sampling adequacy exceeded 0.50 for all variables. Bartletts test of sphericity was large and significant (1429.34), and Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy exceeded 0.60. A principal components analysis with direct oblimin rotation yielded 5 factors. The total variance explained by these five factors is demonstrated in Table 6.

119

Table 6- Total Variance Explained by a Five-Factor Solution Initial Eigenvalues Factor 1 2 3 4 5 Total 5.125 2.846 1.615 1.476 1.121 % of variance 26.975 14.980 8.498 7.766 5.901 Cumulative % 26.971 41.956 50.454 64.120 69.269

This five-factor solution demonstrated that four of the five primary control items loaded on one factor, and that the rest of the secondary control items loaded on the other four factors. However, as there was no clear pattern in the other four factors, a four-factor and three-factor solution were also requested. In both of these analyses however, many of the items loaded on more than one factor. To investigate the hypothesised two-factor solution, a principal components analysis with direct oblimin rotation was requested. More than two factors are present however, as 67% of the nonredundant residuals had absolute values > 0.05. As demonstrated in Table 7, all of the primary control items loaded on Factor 2. Seven of the 14 secondary control items loaded on Factor 1, and the remaining seven secondary control items loaded only on Factor 2 or on both factors. As such, in subsequent analyses, the scale will include all five primary control items and only the seven noncomplex secondary control items. With only these items, a factor analysis reveals that the primary control factor accounts for 19.84% of the variance, and the secondary control factor accounts for 28.30% of the variance.

120

Table 7- Factor Analysis of Primary and Secondary Control Scale No. pc1 pc2 pc3 pc4 pc5 sc1* sc2 sc3 sc4 sc5 sc6 sc7 sc8 sc9 sc10 sc11 sc12 sc13 sc14 Item When I have a goal at work that is difficult to reach, I think about different ways to achieve it. When I want something at work to change, I think I can make it happen. When a work task really matters to me, I think about it a lot. When I really want to reach a goal at work, I believe I can achieve it. When faced with a difficult work situation, I believe I can overcome it. I can see that something good will come of it. I remember you can't always get what you want. I know things will work out okay in the end. I remember I am better off than many other people. I remember I have already accomplished a lot in life. I remember the success of my family or friends. I think nice thoughts to take my mind of it. I remind myself the situation will change if I am just patient. I tell myself it doesnt matter. I think about my success in other areas. I dont feel disappointed because I knew it might happen. I can see it was not my fault. I ignore it by thinking about other things. I realise I didnt need to control it anyway. Eigenvalues % of variance Cumulative variance Cronbach's Alpha (for revised scale) F1 F2 0.70 0.39 0.51 0.71 0.51 0.32 0.32 0.44 0.36 0.53 0.55 0.74 0.65 0.72 0.74 0.61 0.40 0.69 0.61 5.13 14.98 14.98 0.82 0.65 0.45 0.67 0.51 0.39 0.34 0.38

2.85 26.98 41.96 0.70

Items with loadings less than 0.30 are not shown. pc= primary control; sc=secondary control; Bolded items are included in the scale. *All secondary control items preceded by When something bad happens that I cannot change

121

1.5.6

Factor Analysis of the Job Autonomy Scale To ensure the items on the job autonomy scale were measuring a single

construct, a factor analysis was conducted. The assumption of normality was violated with items 1, 2, 4, and 10 being mildly negatively skewed for academics. Items 7, 8, 9, and 11 were mildly positively skewed for the supermarket workers. As before, these variables were not transformed. Nine univariate outliers were recoded to three standard deviations from the mean. All correlations exceeded 0.30 and all of the variable MSA exceeded 0.50. Barlett's test of sphericity was significant (1077.97), and Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy exceeded 0.60. A principal components analysis with direct oblimin rotation demonstrated that the job autonomy items loaded on three factors. There was no meaningful pattern within these factors however, and all but three items loaded on more than one factor. In an attempt to find a pattern among the items, a two-factor principle components analysis with direct oblimin rotation was conducted. This analysis, presented in Table 8, demonstrated that six items loaded only on the first factor, one item loaded only on the second factor, and the remaining seven loaded on both factors. Factor 1 contained items that were directly related to the nature of the work (i.e., tasks, order of work, working pace), whereas the items that loaded on Factor 2 related to organisational structure (i.e., pay, evaluation). Although this factor analysis demonstrates that two factors emerge, all items will be retained in this scale as the overall measure of job autonomy should be based on the nature of the work and the organisational structure. Table 8- Factor Analysis of Job Autonomy Scale

122

No 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Item In my job I can choose among a variety of tasks or projects to do. In my job I can choose the order in which I do my work. In my job I can choose how quickly I do my work. In my job I can choose how I schedule my rest breaks. In my job I can choose the physical conditions of my workstation. In my job I can choose when I interact with others. In my job I can choose the amount I earn. In my job I can choose the number of times I am interrupted at work. In my job I can choose how my work is evaluated. In my job I can choose the quality of my work. In my job I can choose the policies and procedures in my work unit. In my job I can choose among a variety of methods to complete my work. In my job I can choose how much work I get done. In general, how much are you able to influence work and work-related matters. Eigenvalues % of variance Cumulative variance Cronbach's Alpha (revised scale)

F1 0.76 0.84 0.71 0.71 0.56 0.74 0.42 0.56 0.54 0.82 0.67 0.64 0.58 41.37 41.37 0.86

F2 0.33

0.46 0.66 0.58 0.76 0.59 0.36 0.41 0.50 1.27 9.04 50.41

Items with loadings less than 0.30 are not shown.

123

1.6

Hypothesis Testing

In order to test the proposed model of job satisfaction, multivariate analyses of variance were conducted to investigate how the academics and the supermarket workers differed in their levels of job autonomy, control strategies, and job satisfaction. Multiple regression analyses were also conducted to predict job satisfaction from job autonomy, control strategies, personality, and life satisfaction. In order to test these hypotheses, 22 p-values must be computed, and as such, familywise error rate must be considered. Familywise error rate is the probability of making at least one Type I error in a set of analyses (Keppel, 1991). Increasing the number of statistical tests can potentially increase the familywise error. The formula for familywise error is FW= (alpha level) x (number of comparisons). In this study, the familywise error rate is (0.05) x (22) = 1.1. One solution to reduce this familywise error rate is adjust the alpha level using the Bonferroni test (Keppel, 1991). The desired alpha level (0.05) is divided by the number of tests (22), yielding a recommended alpha level of 0.002. Although reducing the alpha level decreases the probability of Type I errors, it also increases the probability of Type II errors (Keppel, 1991). The solution therefore is to strike a balance between the two errors. Thus, the alpha level will be reduced to 0.01. 1.6.1 Hypothesis One- Assumption Testing In order to test the first part of hypothesis one, proposing that job autonomy and job satisfaction are positively related, the correlation coefficients for each occupational group were examined. Consistently, job autonomy was positively related

124

to job satisfaction for both the academics (r = 0.41) and the supermarket workers (r = 0.25). In order to test the second part of hypothesis one, proposing that the academics would report higher job autonomy than the supermarket workers, an analysis of variance was employed. The assumption of univariate homogeneity of variance, assessed using Levenes test was not met, F (1, 197) = 12.77, p = 0.00, however as this assumption is of little concern when the sample sizes are similar (Tabachnick & Fidell, 1997), the analysis proceeded with caution using an alpha level of 0.01. The univariate test of significance demonstrated that, consistent with hypothesis one, the academics (M = 51.94, SD = 14.63) reported significantly higher job autonomy than the supermarket workers, (M = 34.50, SD = 20.24), F (1, 197) = 49.51, p = 0.00. 1.6.2 Hypothesis Two- Occupational Differences in the Use of the Control Strategies In order to examine hypothesis two proposing that the academics will report less secondary control and more primary control than the supermarket workers, a multivariate analysis of variance was performed. The assumptions of normality, linearity, multicollinearity, and homogeneity of variance-covariance were examined for the variables. All of the variables were normally distributed, and reasonably linear relationships were evident. Two univariate outliers were recoded to three standard

125

deviations from the mean and there were no multivariate outliers. There was no evidence of multicollinearity, as the determinant of the within-cell correlation was > 0.0001 (i.e., 0.798). The assumption of univariate homogeneity of variance, as assessed by Levenes test, was met for secondary control, F (1, 188) = 2.84, p > 0.05. Equality of error variance was not found however for primary control, F (1, 188) = 17.07, p < 0.05. Levenes test is sensitive to non-normality however, and this can lead to overly conservative rejection (Tabachnick & Fidell, 1997). As such, the analyses will proceed with caution using an alpha level of 0.01. The assumption of multivariate homogeneity of variance-covariance, assessed through Boxs M test was also violated. Boxs M test is a notoriously sensitive test of homogeneity of variance-covariance, and it is recommended that if the test is violated, the multivariate tests be examined by Pillais criterion rather than Wilks lamba. The multivariate test of significance, using Pillais criterion, demonstrated that occupational differences existed, F (2, 187) = 10.03, p = 0.00. As demonstrated in Table 9, the supermarket workers reported significantly higher secondary control than the academics, F (1, 188) = 15.50, p = 0.00. The two groups did not report significantly different levels of primary control, F (1, 188) = 3.99, p = 0.04. It must be noted however that the difference in primary control was significant at 0.05, but not at the more stringent alpha level of 0.01. Hence, only partial support was provided for the second hypothesis.

126

Table 9- Means and Standard Deviations of Control Measures for Academics and Supermarket Workers Variable Primary Control Secondary Control M 71.56 36.63 Academic SD 11.95 15.64 Supermarket SD 18.62 19.77

M 67.06 46.74

Bolded constructs demonstrate significant occupational differences. 1.6.2.1 Summary Multivariate analyses of variance demonstrated that the academics report higher job autonomy, and lower secondary control than the supermarket workers. The two groups did not report significantly different levels of primary control. 1.6.3 Hypothesis Three- Examining how Job Autonomy Relates to the Control Strategies To examine hypothesis three, proposing that job autonomy will be positively related to primary control and negatively related to secondary control, the correlation coefficients were examined. It was necessary to examine whether job autonomy influences the control strategies using the measured level of job autonomy because there was some variability in the level of job autonomy reported within occupational groups. A median split was conducted on job autonomy and the employees were split into two groups. The majority of academics were in the high job autonomy group (66%), however 34% were in the low job autonomy group. Similarly, 70% of the

127

supermarket workers were in the low job autonomy group, however 30% were in the high job autonomy group. Job autonomy was positively related to primary control (r = 0.46), but not related to secondary control (r = - 0.18). These results provide partial support for hypothesis three, suggesting that job autonomy influences primary, but not secondary control. 1.6.4 Hypothesis Four- Examining how Job Autonomy Influences the Adaptiveness of the Control Strategies To examine hypothesis four, proposing that i) primary control will be more positively related to job satisfaction than secondary control for the academics, and ii) secondary control will be more positively related to job satisfaction than primary control for the supermarket workers, a standard multiple regression analysis was conducted on each occupational group. The assumptions of normality, linearity and homoscedasticity of residuals were assessed through examination of the residual scatterplots. These assumptions were met, and there was no evidence of multicollinearity. As demonstrated in Table 10, R was significantly different from zero for both the academics, R = 0.44, F (2, 102) = 12.53, p = 0.00, and the supermarket workers, R = 0.31, F (2, 82) = 4.37, p = 0.01. Primary control predicted job satisfaction for both groups, accounting for 20% of the variance in job satisfaction for the academics and almost 10% for the supermarket workers. Secondary control did not predict job satisfaction for either group. Hence, consistent with hypothesis four, primary control

128

was more positively related to job satisfaction than secondary control for the academics. Inconsistently however, secondary control was not related to job satisfaction for the supermarket workers. Table 10- Multiple Regression of Primary and Secondary Control on Job Satisfaction for Academics and Supermarket Workers Group Acad Variable PC SC Super PC SC 0.31 -0.02 0.08 0.43 -0.05 R =0.31*
**

JS 0.44 0.04

PC -0.02

B 0.78 0.006 R =0.44**

0.44 0.05 R2=0.20 0.31 -0.04 R2=0.10

sr2 (unique) 19.62** Adj R2=0.18 9.61** Adj R2=0.07

p<0.01; Acad Academics; Super Supermarket workers; JS Job satisfaction; PC Primary control; SC Secondary control For academics, R2 composed of shared variance (1.9%) and unique variance (98.1%) For supermarket workers, R2 composed of shared variance (3.9%) and unique variance (96.1%). 1.6.5 Hypothesis Five- Does Job Autonomy Moderate the Relationship Between the Control Strategies and Job Satisfaction? In order to examine hypothesis five, proposing that the relationship between the control strategies and job satisfaction is moderated by job autonomy, two hierarchical multiple regression analyses were conducted. Job autonomy is proposed to be a moderator, which means that it affects the direction and/or the strength of the relationship between the control strategies and job satisfaction. Specifically, the

129

relationship between primary control and job satisfaction is expected to be positive when job autonomy is high and negative when job autonomy is low. Furthermore, the relationship between secondary control and job satisfaction is expected to be positive when job autonomy is low, and negative when job autonomy is high. These expected relationships are demonstrated below in Figure 3. Figure 3- Expected Moderated Effect of Job Autonomy on a) Primary Control and Job Satisfaction and b) Secondary Control and Job Satisfaction

a)

High Job Autonomy

JS Low Job Autonomy Primary Control b)

Low Job Autonomy

JS High Job Autonomy Secondary Control

A moderation effect can be tested in a number of ways depending on whether the variables are continuous or discrete (Baron & Kenny, 1986). In this hypothesis, the moderator variable and the independent variable are both continuous. When both

130

variables are continuous, and when the effect of the independent variable on the dependent variable varies linearly with respect to the moderator, a hierarchical multiple regression analysis is conducted to test the presumed relationship (Baron & Kenny, 1986). As demonstrated in Figure 4, the dependent variable is regressed on the independent variable, the moderator variable, and the product of the independent variable and the moderator (Baron & Kenny, 1986). Moderator effects are demonstrated if the interaction term is significant when the independent variable and the moderator variables are controlled (Baron & Kenny, 1986).

131

Figure 4- Job autonomy Moderates the Relationship between a) Primary control and b) Secondary Control, and Job Satisfaction. Order of Variable Entry a) Step 1 Step 2 Step 3 Primary control Job autonomy Primary control x Job autonomy Job Satisf actio n

b) Step 1 Step 2 Step 3 Secondary control Job autonomy Secondary control x Job autonomy Job Satisf actio n

In order to test the moderating effect of job autonomy on primary control and secondary control, two hierarchical multiple regression analyses were conducted on the combined sample. By using the combined sample, there was more range in the levels of job autonomy. In these analyses the control strategies were entered first, then job autonomy, and then the interaction term. For primary control, R was significantly different from zero after the first step (i.e., primary control), R = 0.37, F (1, 188) = 30.34, p = 0.00, and the second step (i.e., job autonomy), R = 0.40, Finc (1, 187) = 4.89, p = 0.03. However, the addition of the interaction term was not significant, R =0.40, Finc (1,186)= 0.33, p = 0.57.

132

For secondary control, R was not significantly different from zero after the first step, R = 0.01, F (1, 197) = 0.04, p = 0.85. After job autonomy was entered, the value of R increased, R = 0.33, F (1, 196) = 24.01, p = 0.00. There was no further increase however when the interaction term was entered in step three, R = 0.34, F (1, 195) = 1.58, p = 0.21. These analyses, displayed in Table 11, demonstrate that inconsistent with hypothesis five, job autonomy did not moderate the relationship between the control strategies and job satisfaction.

133

Table 11- Moderating Role of Primary and Secondary Control on the Relationship Between Job Autonomy and Job Satisfaction Step 1 2 3 IV Primary control Primary control Job Autonomy Primary control Job autonomy Primary control x job autonomy DV JS B 0.57 0.46 0.20 0.55 0.41 -0.003 R =0.40 1 2 3 Secondary control Secondary control Job Autonomy Secondary control Job autonomy Secondary control x job autonomy JS 0.02 0.02 0.40 0.26 0.63 -0.005 R =0.34
**

0.37 0.30 0.17 0.36 0.34 -0.21 R2=0.16 0.01 0.02 0.33 0.21 0.52 -0.28 R2=0.12

sr2 (unique) 13.91** 6.96** 2.19* 3.46**

AdjR2=0.15

10.89** 4.54*

AdjR2=0.10

p<0.01, *p<0.05; JS Job satisfaction Hypothesis Six- Do the Control Strategies Mediate the Relationship Between Job Autonomy and Job Satisfaction? Hypothesis six proposes that the relationship between job autonomy and job

1.6.6

satisfaction is mediated by the control strategies. In this hypothesis, the control strategies are acting as mediators because they are explaining why job autonomy is related to job satisfaction. That is, employees with high job autonomy are expected to rely on more primary control and less secondary control than employees with low job autonomy. As primary control strategies are more positively related to job satisfaction

134

than secondary control strategies, employees with higher job autonomy report higher job satisfaction. It must be noted that although secondary control strategies are less positively related to job satisfaction than primary control, it is proposed that for workers with low job autonomy, secondary control strategies are superior to primary control strategies. If these workers use primary control, they are expected to experience primary control failure. According to Baron and Kenny (1986), in order to establish mediation, three standard regression analyses must demonstrate that: a) job autonomy predicts primary and secondary control; b) primary and secondary control and job autonomy together predict job satisfaction; and c) job autonomy predicts job satisfaction. For a mediation effect to be significant, all three regression equations must be significant, and the effect of the independent variable on the dependent variable must be less in b) than in c) (Baron & Kenny, 1986). This mediation analysis is demonstrated in Figure 5. Figure 5- Mediating Role of Control Strategies on the Relationship Between Job Autonomy and Job Satisfaction

Primary/Secondary Control

Job autonomy

Job Satisfaction

This method will not be used however as there is an easier way to test the mediating role of the control strategies. Rather than conducting three regression

135

analyses, only one hierarchical regression analysis is needed (M. Stokes, personal communication, August 16, 2002). In this analysis, primary and secondary control strategies are entered first, followed by job autonomy. It is expected that once primary and secondary control strategies have been entered, there would be no relationship between job autonomy and job satisfaction. As such, primary and secondary control would explain the relationship between job autonomy and job satisfaction. The assumptions of normality, linearity and homoscedasticity of residuals were met, and there was no evidence of multicollinearity. As demonstrated in Table 12, R was significantly different from zero after primary and secondary control were entered, R = 0.37, F (2, 187) = 15.23, p = 0.00. Primary control accounted for 13% of the variance in job satisfaction, and secondary control was not significant. R did not significantly increase after job autonomy was added to the equation, R = 0.40, Finc (3, 186) = 11.92, p = 0.03. Even if the less stringent alpha level of 0.05 was used, job autonomy only accounts for 2% of the unique variance in job satisfaction. As such, it appears that when primary and secondary control are entered first, there is no relationship between job autonomy and job satisfaction. This suggests that partial support is provided for hypothesis six as primary control, but not secondary control, mediates the relationship between job autonomy and job satisfaction.

Table 12 -Hierarchical Multiple Regression Testing the Mediating Role of the Control Strategies Step IV DV B sr2 (unique)

136

Primary Control Secondary Control

JS

0.57 -0.04

0.37 -0.03

13.91**

R =0.37**

R2=0.14

AdjR2=0.13

Primary Control Secondary Control Job Autonomy

JS

0.46 -0.02 0.20 R =0.40**

0.30 -0.02 0.17 R2=0.16

6.96** 2.01* AdjR2=0.15

**p>0.01 *p>0.05; JS - Job satisfaction It must be noted that although the results demonstrate that primary control is a partial mediator of the relationship between job autonomy and job satisfaction, the use of multiple regression to estimate a mediational model is based on the assumption that there is no measurement error in the mediator. This assumption is particularly concerning as the mediator is likely to be measured with error. The presence of such error tends to produce an underestimation of the effects of the mediator, and an overestimation of the effects of the independent variable on the dependent variable (Baron & Kenny, 1986, p. 1177). One statistical method that models the measurement error is structural equation modeling. Structural equation modeling is based on the analysis of sample variances and covariances rather than individual cases. This approach is particularly useful for

137

latent variables, which are hypothetical constructs that cannot be directly measured, such as job satisfaction. Although structural equation modelling has some advantages over multiple regression, it will not be used in this thesis for a number of reasons. First, unlike hierarchical multiple regression, structural equation modelling is a confirmatory technique. The current study, although grounded in theory, is exploratory, attempting to combine the propositions of the job demand-control model (Karasek & Theorell, 1990) with propositions of the life span theory of control (Heckhausen & Schulz, 1995). As the theory is exploratory, there are a variety of different models that can be examined. If numerous modifications of a model were tested, the analysis would be exploratory, and there would be an increased risk of Type I errors (Ullman, 1996). As this thesis is attempting to develop and explore the proposed model of job satisfaction and search for unexpected relationships, structural equation modelling may be problematic. Once the model is more established however, structural equation modelling may be required. A further problem with using structural equation modelling is that it requires large sample sizes. The issue of an adequate sample size continues to be debated, however Boomsma (1983) suggested that as a general rule, samples of 200 are required to give parameter estimates with any degree of confidence. As the relationship between the variables is expected to be different for academics and supermarket workers, two models would need to be conducted, thus there would need to be 200 in each occupational group.

138

A sample size of 200 is problematic due to time constraints, but also because of the particular workers selected for this study. The data collection process undertaken in study one demonstrated that workplaces, particularly those employing low autonomy workers, such as call centres, factories, or supermarkets, were reluctant to become involved in any research. The employers refused to participate in the surveys for a variety of reasons. Some mentioned that the majority of their employees were from Non-English speaking backgrounds and as such would be unable to understand the survey. Others admitted that work motivation was very low, and as such, the response rate would be poor. Still others were concerned that the employees would expect changes to be made to the workplace on the basis of their responses. These employers reactions indicate that is difficult to obtain a sample size of 400. 1.6.6.1 Summary In summary, it appears that primary control mediates the relationship between job autonomy and job satisfaction. This finding was based on multiple regression analyses however, which assumes that there is no measurement error in the mediator. Although this measurement error can be accounted for in structural equation modeling, it is concluded that such a method is not appropriate whilst the proposed model of job satisfaction is in an exploratory stage. 1.6.7 Hypothesis Seven- Occupational Differences in Job and Life Satisfaction Hypothesis seven proposes that the academics will report higher job satisfaction and higher life satisfaction than the supermarket workers. A univariate

139

analysis of variance was conducted on the global one-item measure of job satisfaction. The assumptions of normality, linearity, and homogeneity of variance were examined. Job satisfaction was normally distributed for the supermarket workers, however it was negatively skewed for the academics (-5.60). The assumption of homogeneity of variance was violated, F (1, 197) = 6.06, p < 0.05, and as such, the analysis will proceed with caution using an alpha level of 0.01. Inconsistent with hypothesis seven, there were no occupational differences in the one-item measure of job satisfaction, F (1, 197) = 3.66, p = 0.04. The levels of job satisfaction reported by the two groups are provided in Table 13. To examine whether the two groups differed on the facets of job satisfaction, a multivariate analysis of variance was conducted on the five facets of job satisfaction, namely nature of work, co-workers, pay, supervisors, and opportunities for promotion. Normality was assessed using skew/standard error < 3, Kolmogorov-Smirnof values, and normal probability plots. Although the nature of work facet (-3.49) and the co-workers facet (-4.87) were negatively skewed for the academic group, the remainder of the variables were normally distributed for both groups. Five univariate outliers were recoded to three standard deviations from the mean, and no multivariate outliers were identified. Examination of bivariate scatterplots, and correlations revealed reasonably linear relationships. There was no evidence of multicollinearity as the determinant of the within-cell correlation was >0.0001. Univariate homogeneity of variance, assessed by Levenes test, demonstrated that equality of error variance was evident for the supervision facet,

140

F (1, 197) = 0.40, p > 0.05. Equality of error variance was not found however for pay, F (1, 197) = 5.53, p < 0.05, nature of work, F (1, 197) = 27.14, p < 0.05, co-workers, F (1, 197) = 5.82, p < 0.05, and promotion, F (1, 197) = 4.20, p < 0.05. As such, the univariate tests will be examined with caution. The assumption of multivariate homogeneity of variance-covariance, as assessed through Boxs M test, was also violated, F (15, 146782) = 4.90, p < 0.001. The multivariate tests were examined using Pillais criterion. The job facets were affected by occupation, F (5, 193) = 35.10, p = 0.00. As demonstrated in Table 13, academics reported significantly higher satisfaction with nature of work, F (1, 197)= 95.59, p = 0.00, and co-workers, F (1, 197) = 32.51, p = 0.00, than supermarket workers. However, the supermarket workers reported higher opportunity for promotion than the academics, F (1, 197) = 9.21, p = 0.00. Table 13- Means and Standard Deviations of Job Satisfaction Scale for Academics and Supermarket Workers Variable Nature Co-Workers Pay Supervisors Promotion One-item measure M 85.97 82.30 51.20 60.29 38.31 66.05 Academic SD 13.74 15.75 27.55 24.47 27.40 21.09 Supermarket SD 26.95 20.01 23.56 24.49 32.18 25.69

M 56.98 67.85 54.17 67.03 51.12 59.71

Bolded variables indicate significant occupational differences It was expected that levels of job satisfaction would be related to levels of life satisfaction, and that the academics reporting higher job satisfaction than the supermarket workers would also report higher life satisfaction. To examine this

141

hypothesis, a univariate analysis of variance was conducted to examine overall life satisfaction, and a multivariate analysis of variance was conducted to examine on which domains the groups differed. To compare their overall life satisfaction, a univariate analysis of variance was conducted. Life satisfaction was normally distributed for the supermarket workers however it was mildly negatively skewed for the academics (-3.94). The assumption of homogeneity of variance, as assessed through Levenes test of equality of error variance was violated, F (1, 190) = 15.24, p < 0.05, and as such, the analysis proceeded with caution using an alpha level of 0.01. Consistent with hypothesis seven, the academics reported higher life satisfaction than the supermarket workers, F (1, 190) = 6.38, p = 0.01. A multivariate analysis of variance was conducted on the seven domains of life satisfaction to examine where these differences lay. The assumptions of normality, linearity, multicollinearity and homogeneity of variance-covariance were examined for the seven domains. The emotional well-being domain was mildly negatively skewed for the academics, and the intimacy domain was mildly negatively skewed for the supermarket workers. 12 univariate outliers were recoded to three standard deviations from the mean. Four multivariate outliers were examined and recoded to the next less extreme score. The assumption of linearity, examined through bivariate scatterplots, was met. Equality of error variance was demonstrated only for satisfaction with health, F (1, 190) = 2.28, p > 0.05, and as such, the analysis will proceed with caution. As the assumption of multivariate homogeneity of variance, examined through Boxs M test was also violated, F (28, 110823) = 2.54,

142

p < 0.001, Pillais criterion was used to examine the multivariate test. Pillais criterion was significant, F (7, 184) = 1039.56, p < 0.01. Academics reported significantly higher productivity satisfaction, F (1, 190) = 7.63, p = 0.006, and safety satisfaction, F (1, 190) = 12.62, p = 0.00, than supermarket workers. The means and standard deviations for the satisfaction domains are provided in Table 14. Table 14- Means and Standard Deviations of Life Satisfaction for Academics and Supermarket Workers Occupation Material Satisfaction Health Satisfaction Productivity Satisfaction Intimacy Satisfaction Community satisfaction Safety Satisfaction Emotional Satisfaction OVERALL SATISFACTION M 80.56 75.03 76.00 81.08 75.71 83.52 78.86 78.22 Academic SD 13.18 20.07 13.91 18.44 16.22 15.13 15.27 10.96 Supermarket M SD 78.56 19.43 69.16 23.85 69.05 21.67 76.63 24.21 70.07 20.13 74.46 20.85 76.63 22.10 73.30 15.97

Bolded variables indicates occupational differences

1.6.8

Hypothesis Eight- Predictors of Job Satisfaction In order to evaluate hypothesis eight, which proposes that primary control,

secondary control, job autonomy, personality and life satisfaction predict job satisfaction, a multiple regression analysis was conducted on both occupational groups. The correlations among the variables are displayed in Table 4. For both groups, the assumptions of normality, linearity and homoscedasticity of residuals were met, and there was no evidence of multicollinearity. R was

143

significantly different from zero after all the variables had been added for both the academics, R = 0.54, F (6,98) = 6.69, p = 0.00, and the supermarket workers, R =0.46, F (6, 73) = 3.25, p = 0.00. For the academics, the unique predictors of job satisfaction were job autonomy and primary control. As demonstrated in Table 15, primary control and job autonomy accounted for 4% and 8% of the variance in job satisfaction respectively. It must be noted however that job autonomy was not significant at the more stringent alpha level of 0.01. For the supermarket workers, there was only one unique predictor of job satisfaction, namely primary control. Primary control accounted for 8% of the variance in job satisfaction. These results suggest that hypothesis eight is partially supported as primary control and job autonomy predicted job satisfaction. However, secondary control, personality and life satisfaction were not unique predictors of job satisfaction. Furthermore, even when all the variables were included in the equation, R2 was small (R2 = 0.29, R2 = 0.21).

144

Table 15- Multiple Regression of Job Autonomy, Control Strategies, Personality, and Life Satisfaction for Academics and Supermarket Workers Group Acad IV Job autonomy Primary Control Secondary Control Neuroticism Extroversion Life Satisfaction DV JS B 0.34 0.58 0.03 -2.30 -1.02 0.10 R =0.54** Super Job autonomy Primary Control Secondary Control Neuroticism Extroversion Life Satisfaction JS 0.08 0.46 0.15 -3.63 -1.14 -0.04 R =0.46** 0.24 0.33 0.03 -0.16 -0.05 0.05 R2=0.29 0.06 0.36 0.12 -0.25 -0.06 -0.03 R2=0.21 sr2 (unique) 4.41* 8.12**

Adj2=0.25 8.35**

Adj2=0.15

** p<0.01, * p<0.05; Acad Academics; Super- Supermarket workers; JS Job satisfaction 1.6.8.1 Summary The academics reported significantly higher life satisfaction than the supermarket workers, but similar levels of job satisfaction. The major predictors of job satisfaction were job autonomy and primary control strategies.

145

1.6.9

Conclusion The major propositions of this study were that job autonomy influences the use

of the control strategies and the relationship between the control strategies and job satisfaction. As hypothesised, the academics reported higher job autonomy, higher life satisfaction and lower secondary control than the supermarket workers. Inconsistent with the hypotheses, the two groups reported similar levels of primary control and job satisfaction. However, job autonomy was positively correlated with primary control and not correlated with secondary control. In regard to the proposal that job autonomy influences the relationship between the control strategies and job satisfaction, the findings were less supportive. Primary control was the most adaptive strategy for both groups, and secondary control was not related to job satisfaction for either group. The implications of these findings will now be discussed.

146

1.7

Discussion

This study tested a new explanation for the relationship between job autonomy and job satisfaction, namely that job autonomy influences the use and adaptiveness of primary and secondary control strategies. In regard to the use, the findings demonstrated that the supermarket workers reported more secondary control than the academics, but that only primary control was related to job autonomy. In regard to the adaptiveness, primary control was the most adaptive strategy for academics and supermarket workers. These findings are discussed in terms of the life span theory of control (Heckhausen & Schulz, 1995) and the discrimination model (Thompson et al., 1998). Before these propositions are explained in detail, the basic assumptions of the study will be examined. 1.7.1 Assumption Testing The basic assumptions of the study were that job autonomy was positively related to job satisfaction, and that the study used two occupational groups that differed in their level of job autonomy. Consistently, job autonomy was positively related to job satisfaction for the academics and the supermarket workers (r = 0.41, r = 0.25, respectively). These correlations are slightly lower than those reported in other studies using Gansters (1989, cited in Dwyer & Ganster, 1991) scale. For example, Munro, Rodwell and Harding (1998) demonstrated that the correlation between job autonomy and job satisfaction was r = 0.69, whilst Fox et al., (1993) demonstrated that r = 0.46. It must be noted however that these studies relied on the original version of the scale, which included items on predictability.

147

Consistent with the second part of hypothesis one, the academics levels of job autonomy (M = 52%SM) were significantly higher than the supermarket workers (M = 32%SM). For the purpose of this study, this difference in job autonomy should be sufficient to examine the differences in primary and secondary control. It is expected that the relationship between job autonomy and the control strategies is linear, and that with increasing job autonomy, the use of primary control will increase, and the use of secondary control will decrease. As such, even if the difference between the group is not extremely large, the differences in the use of the control strategies should still exist, however they may be less extreme. In order to understand the meaning of these levels of job autonomy, it is useful to compare them with other studies. Although normative data on Gansters (1989, cited in Dwyer & Ganster, 1991) scale are not available, a few studies have relied on this scale. They have shown that nurses scored 46%SM (Ganster et al., 2001), and 57%SM (Munro et al., 1998). Furthermore, manufacturing employees scored 57%SM (Dwyer & Ganster, 1991). It is difficult to make comparisons with past studies however, as these studies have generally altered the scale in some way (e.g., Ganster et al., 2001; Munro et al., 1998). Indeed the current study made an important change to the scale, as the items on predictability were excluded. 1.7.2 Does Job Autonomy Influence the Use of the Control Strategies? Partial support was provided for the second hypothesis as there was a significant occupational difference in secondary control (M = 36%SM academics;

148

M = 46%SM supermarket workers), but not primary control (M = 71%SM academics; M = 67%SM supermarket workers). Partial support was also provided for hypothesis three as primary control was positively related to job autonomy (r = 0.46), however secondary control was not correlated (r = -0.18). These findings are somewhat inconsistent, with the former suggesting that job autonomy influences secondary control but not primary control, and the latter suggesting that job autonomy influences primary but not secondary control. More emphasis is placed on hypothesis three as it is based on the measured level of job autonomy rather than the assumed level. Thus, these findings demonstrate that as job autonomy increases, primary control increases. These findings appear to be inconsistent with Abouseries (1996) study on academics coping strategies. In this study, academics were given a list of strategies and required to indicate which ones they use to handle stress. The following coping strategies emerged as the most common; acceptance of the problem (58%), talking with others (57.7%), and trying to come to terms with each problem (55.8%). Although the most common strategy, acceptance of the problem appears to be a secondary control strategy, it is different to secondary control. Secondary control is often referred to as acceptance however it is not acceptance that the problem exists; it is acceptance that the problem cannot be overcome. Acceptance of the problem may be interpreted as recognising that the problem exists, which is not secondary control. Thus, although Abouseries (1996) results suggest that academics mostly use secondary control, this may not be the case.

149

One study partially supports the findings from the current study. Narayanan, Marian and Spector (1999) studied the coping strategies reported by academics, sales employees, and clerical workers. They used an open-ended questionnaire where participants were asked how they handled a stressful event at work. The academics tended to handle their problems at work by taking direct action (24% of sample), and talking to the chair of department (26%). The clerical workers and the sales employees, on the other hand, reported that they talked with their co-workers (22%, 29%, respectively), or friends (24%, 29%, respectively). Although Narayanan et al., (1999) did not measure job autonomy, their findings demonstrate that the employees expected to have higher job autonomy (i.e., academics) tended to rely on primary control-type strategies. The employees expected to have lower job autonomy (i.e., sales employees, clerical workers) tended to rely on secondary control-type strategies. These findings were partially consistent with the current study. The finding that job autonomy is positively related to primary control provides some support for the proposed model of job satisfaction presented in Figure 2. This model, based on the job demand-control model (Karasek & Theorell, 1990) and the life span theory of control (Heckhausen & Schulz, 1995), proposes that employees with high job autonomy are more likely to successfully change the environment using primary control. Thus, as job autonomy increases, primary control increases. However, the findings in the present study must be examined cautiously as a limitation has been identified. The primary and secondary control scale required respondents to indicate their agreement with each type of strategy, from 1 (do not

150

agree at all) to 10 (agree completely). It is now recognised that the only information this scale provides is whether the respondents have ever used the strategies, and not how often they are using the strategies. The current findings only demonstrate that as job autonomy increases, employees agreement with the primary control strategies increases, not the frequency. 1.7.3 Does Job Autonomy Influence the Relationship Between the Control Strategies and Job Satisfaction? In addition to testing whether job autonomy influences the use of the control strategies, the current study also tested whether job autonomy influenced the relationship between the control strategies and job satisfaction. Consistently, primary control (r = 0.44) was more positively related to job satisfaction than secondary control (r = 0.04) for the academics. However, secondary control (r = 0.14) was not more positively related to job satisfaction than primary control (r = 0.38) for the supermarket workers. Further analyses demonstrated that job autonomy did not moderate the relationship between the control strategies and job satisfaction. As such, it appears that primary control is more adaptive than secondary control for all employees, whether they have low or high job autonomy. These findings do not support the discrimination model (Thompson et al., 1998) which proposes that primary control is the most adaptive strategy in controllable situations, and that secondary control is the most adaptive strategy in uncontrollable situations. Rather, these findings support the primacy/back-up model (Thompson et

151

al., 1998), which proposes that primary control strategies are more adaptive than secondary control strategies for people in low-control or high-control situations. Although these results appear to support the primacy/back-up model, closer examination of the Primary and Secondary Control Scale (Heeps et al., 2000) reveals several limitations. The most notable is that some of the primary control items examined whether the employees believed that they could change their situation, rather than examining how they could change their situation. For example, the items I think I can make it happen, I believe I can achieve it and I believe I can overcome it measure whether a person believes that they can change a situation. These general and non-specific thoughts were assessed rather than specific perceived behaviours (i.e., work harder) because it was assumed that there could be an unlimited number of specific behaviours. However, it is now questioned whether believing that one can change a situation is a measure of primary control. A person may report that they can change a situation for a variety of reasons, not just if they use primary control strategies when they face difficulties. For example, a person may report that they can change their environment because they have high optimism. Alternatively, they may be using the secondary control strategy illusory optimism where they tell themselves that everything will work out okay in the end. These examples serve to illustrate that people who believe that they can change their environment may not necessarily use primary control. To overcome these limitations, the measure of primary control may need to be more specific. Rather than assessing whether people generally believe they can

152

change their environment, the primary control scale needs to assess how people change their environment using primary control strategies. Thus the scale needs to examine perceived strategies (e.g., exerting more effort, working harder) rather than beliefs. This would make the scale consistent with the secondary control scale, which assesses specific strategies. In summary, although the findings suggest that the relationship between the control strategies and job satisfaction is not influenced by job autonomy, several problems have been identified in the primary and secondary control scale. The primary and secondary control scale needs to be revised so that the primary control items refer to perceived strategies rather than beliefs, and the rating scale needs to assess frequency. 1.7.4 Do the Control Strategies Mediate the Relationship Between Job Autonomy and Job Satisfaction? Hypothesis six tested an alternative explanation to Karasek and Theorells (1990) proposal for the relationship between job autonomy and job satisfaction. This explanation, developed in chapter 1, proposes that employees with high job autonomy rely on more primary control strategies which are positively related to job satisfaction, whereas employees with low job autonomy rely on more secondary control strategies which are less positively related to job satisfaction. It must be noted however that although secondary control strategies are less positively related to job satisfaction, it is proposed that for workers with low job autonomy, secondary control strategies are

153

superior to primary control strategies. If these workers use primary control, they are expected to experience primary control failure. The results demonstrated that primary control, but not secondary control, mediated the relationship between job autonomy and job satisfaction. This provides empirical evidence supporting one mechanism by which job autonomy may influence job satisfaction. The importance of these findings must not be overemphasised however, as problems have been identified with the primary and secondary control scale. As such, the mediating role of primary and secondary control needs to be reexamined using a revised scale. 1.7.4.1 Summary The major aim of this study was to test an explanation for the relationship between job autonomy and job satisfaction. The explanation proposes that job autonomy influences the use and adaptiveness of the control strategies. The results from the current study have offered some support for job autonomy influencing the use of primary control strategies, but less support for job autonomy influencing the adaptiveness of the control strategies. However, as there are some methodological problems with the scale, the proposition requires further examination. 1.7.5 Examining Occupational Differences in Job Satisfaction The differences in job autonomy and primary and secondary control were expected to influence job satisfaction, where the academics were expected to report higher job satisfaction than the supermarket workers. This proposal was not supported however, as the academics reported similar levels of job satisfaction

154

(M = 66%SM) as the supermarket workers (M = 59%SM). In order to understand these levels of job satisfaction, past studies will be examined. 1.7.5.1 Past Studies on Job Satisfaction As few studies have examined academics or supermarket workers levels of job satisfaction, and as there does not appear to be any consensus as to what is the normative level of job satisfaction, a review was conducted. A range of studies (N=36), which examined the levels of job satisfaction reported by different occupational groups, were selected from psychology databases. These studies, displayed in Appendix G, examine several occupational groups including nurses, teachers, managers, manufacturing employees, and social workers. Although these studies relied on several different scales, including global and facet scales, they were reasonably consistent. The average level of job satisfaction was 66.75%SM, and the scores ranged from 44.75%SM (Laschinger, Finegan & Shamain, 2001) to 87%SM (Fisher, 2000). This average is similar to the academics and supermarket workers levels of job satisfaction. A few studies have specifically examined academics and supermarket workers levels of job satisfaction. For academics, researchers have reported the following levels of job satisfaction; 57%SM (Leung et al., 2000), 65%SM (Hill, 1986), 66%SM (Lahey & Vihtelic, 2000), 74% (Carson, Lanier & Carson, 2001), 82% (Olsen, 1993) and 83%SM (Niemann & Dovidio, 1998). Although these scores vary widely, it must be recognised that these studies have relied on different scales of job satisfaction. Some relied on facets scales of job

155

satisfaction (Hill, 1986; Lahey & Vihtelic, 2000) whilst others relied on global scales on job satisfaction (Carson, Lanier & Carson, 2001; Leung et al., 2000; Niemann & Dovidio, 1998; Olsen, 1993). However, the facet versus global distinction does not necessarily explain the differing levels of job satisfaction, as facet and global scales of job satisfaction have been shown to be moderately correlated (Wanous et al., 1997). Rather within the facet and global scales, there is extensive variability that may account for the inconsistent levels of job satisfaction. There are differences among the facet scales of job satisfaction. For example, Hills (1986) facet scale of job satisfaction measures satisfaction with several dimensions including economic, teaching administrative, collegial, recognitionsupport, and convenience. In contrast, Lahey and Vihtelic (2000) focussed on the work itself, pay, recognition, co-workers, and supervision. The difference between Hills (1986) facets and Lahey and Vihtelic (2000) facets may be important. Hills (1986) facets were designed to be specific to academia, however it appears that they are focussing on the areas that academics traditionally cite as a source of stress, such as recognition, finances (Leung et al., 2000), and administration (Abouserie, 1996). As such, the academics in Hills (1986) study may have a reported a lower level of job satisfaction than those in Lahey and Vihtelic (2000) study because the scale was focussed on the more negative aspects of the job. There are also differences among the global scales of job satisfaction. For example, Niemann and Dovidio (1998) relied on a 3-item measure of job satisfaction, which included the following items; I am satisfied with my job, I find fulfillment in my work and I feel free to do the work that is important to me. The level of job

156

satisfaction reported by the academics in this study may have been higher because of the inclusion of the third item, which may be confounded with job autonomy. A more valid global measure of job satisfaction was used in Olsens (1993) study. Measuring job satisfaction through one-item (i.e., All things considered, how satisfied are you with your position), they found academics reported a high level of job satisfaction (M = 82%SM). It must be noted however that this level of job satisfaction was reported by academics in their first year of appointment. Interestingly, they re-tested these academics at the end of their third year, and found that their level of job satisfaction had declined to 71.66%SM. This lower level is more consistent with other studies. In summary, it is extremely difficult to produce an average level of job satisfaction for academics. Only a few studies have examined academics job satisfaction and these have tended to rely on different scales. The level of job satisfaction found in the current study fits within the range found by past studies. It must be noted however that this range is reasonably large. In regard to supermarket workers, the only studies that can be compared with the current findings are those conducted on retail workers. These studies have generally reported a higher level of job satisfaction than that found for the supermarket workers. For example, Doran, Stone, Brief and Georges (1991) study demonstrated that retail workers given the Minnesota Satisfaction Questionnaire (Weiss et al., 1967) reported a level of job satisfaction that was 73%SM. Furthermore, Leungs (1997) study on retail workers reported similar findings using Hackman and Oldhams (1975) scale (70% SM).

157

Although these studies report higher levels of job satisfaction, it must be noted that the workers in these studies were obtained from a department store (Doran et al., 1991) and a casual apparel store (Leung, 1997), and as such, may have more job autonomy that the supermarket workers. The supermarket register operators are required to work on the cash register for the majority of their shift whereas retail assistants can often choose among different task to complete. Thus, it is difficult to compare these studies with the current findings. 1.7.5.2 Explaining the Levels of Job Satisfaction Reported by the Academics and the Supermarket Workers The finding that such two distinct occupational groups report similar levels of job satisfaction is surprising. However, there may be differences between the groups that can account for this. First of all, the nature of the work is very different for these two groups. The supermarket workers engage in repetitive work, and as such, they may face few novel difficulties. The academics, on the other hand, are expected to be involved in several complicated activities and face many varied difficulties. As such, although the current study was proposing that supermarket workers would have lower job satisfaction because they have less autonomy, they may also have fewer difficulties to overcome. Another difference between the two occupational groups that may explain their similar levels of job satisfaction concerns their different investments and expectations. Whereas the supermarket workers have invested little time into training, the academics have invested at least seven years studying at university. The number of years

158

invested in training or education may be particularly important, as it has been suggested that education is positively correlated with expectations (Clark & Oswald, 1996). For example, Clarks (1996) study of British employees demonstrated that the percentage of employees who reported that they were very satisfied with their job was greatest for the group with the lowest education (M = 78%SM). The next highest reported a level which was 74%SM, and the highest educated group reported a level of job satisfaction that was 73%SM. Although the differences between these groups are small, it is surprising that the group with the lowest level of education would report a level of job satisfaction that equalled those with a higher education, let alone surpassed it. As such, the academics may have higher job expectations than the supermarket workers. In summary, inconsistent with the hypotheses, the academics and teachers reported similar levels of job satisfaction. This finding may be partly attributed to the supermarket workers experiencing fewer difficulties than the academics, or the academics having higher job expectations than the supermarket workers. 1.7.6 Examining Occupational Differences in Life Satisfaction Consistent with hypothesis seven, the academics reported higher overall life satisfaction (M=78.22) than the supermarket workers (M=73.30). The academics levels of life satisfaction were expected to be higher because job satisfaction was expected to be positively related to life satisfaction. Although the academics did not

159

report higher job satisfaction than the supermarket workers, they did report higher life satisfaction. In regard to the normative levels of life satisfaction, Cummins (1995, 2000b) homeostatic theory of life satisfaction proposes that the mean life satisfaction across population samples lies within the 70-80%SM range. This is because people have a set-point range for their life satisfaction. This set-point range is determined by personality variables, namely neuroticism and extroversion. Together, these two variables provide an affective balance, where the mid-point for the set-point range is, on average, 75%SM. This affective balance influences the second-order buffers (i.e., optimism, self-esteem and control) so that, on average, their set-point is also 75%SM. These second-order buffers can however be influenced by the external world. Hence, the mid-point for the set-point range can range between 70-80%SM. Consistent with this prediction, both the academics (M = 78.22%SM) and the supermarket workers levels of life satisfaction (M = 73.30%SM) lay within this range. The academics level of life satisfaction was at the higher end of the normative range. According to the homeostatic theory of life satisfaction, the ceiling for population sample means is approximately 80%SM (Cummins, 2000b). This value represents the theoretical maximum for sample means grouped as data where the distribution of set-ranges is normal, and each person has achieved the upper value of their set-range. As such, the academics level of life satisfaction, in relative terms, is extremely high. The supermarket groups level of life satisfaction was at the lower end of the normative range. Cummins (2000b) proposes that when life satisfaction falls towards

160

the 70%SM mark, homeostatic devices operate to prevent it from falling further. When the sample mean approaches 70%SM however, the homeostatic machinery is defeated for a significant proportion of the sample. As this happens, the distribution collapses and the standard deviations increase. Consistent with this prediction, the standard deviation of life satisfaction for the supermarket workers group (SD = 15.97) was greater than for the academic group (SD = 10.96). Hence, a greater proportion of the supermarket workers may be experiencing homeostatic defeat. In summary, the academics reported higher life satisfaction than the supermarket workers, however both means lay within the normative range. 1.7.7 Predicting Job Satisfaction from Job Autonomy, Control Strategies, Personality, and Life Satisfaction Partial support was provided for hypothesis eight, as job autonomy and primary control predicted job satisfaction for the academics, and primary control predicted job satisfaction for the supermarket workers. Inconsistent with the proposed model of job satisfaction however, secondary control did not predict job satisfaction for either group, and job autonomy did not predict job satisfaction for the supermarket workers. The finding that secondary control did not predict job satisfaction clearly needs to be re-examined as there are several methodological problems with the secondary control scale. The finding that job autonomy did not predict job satisfaction for the supermarket workers requires further examination. The finding that job autonomy did not predict job satisfaction for the supermarket workers may reflect problems with the job autonomy scale. The job

161

autonomy scale was a multidimensional scale. The scale was thought to be superior to other scales as it prompted employees to consider several aspects of their work environment (Ganster & Fusilier, 1989). However, the scale may also be problematic, as although it ensures that respondents think of the same facets, some facets may not be appropriate for some employees. An alternative is to use a global scale of job autonomy. For example, the Job Descriptive Survey (Hackman & Oldham, 1975) measures job autonomy through assessing whether the employee has the opportunity for independence and freedom in their job. Using this scale, the respondents can just consider the areas that are important to them. They can include facets that are not specified in the facet scale, and exclude facets that are not relevant to their workplace. As such, the supermarket workers, although reporting low job autonomy on the multidimensional scale, may have higher levels of global job autonomy. As such, future studies will need to assess job autonomy using a global measure. 1.7.8 Conclusion This study has contributed to the development of the proposed model of job satisfaction (refer to Figure 2). This model, adapted from Karaseks (1979) job demand-control model, proposes that job autonomy relates to job satisfaction through influencing the way employees manage their work difficulties. The findings demonstrated that workers with higher job autonomy do manage their work difficulties differently from workers with lower job autonomy. Specifically, as job autonomy increases, primary control increases.

162

In addition to examining how job autonomy influences the use of control strategies, this study also proposed that job autonomy influences the adaptiveness of the control strategies. Primary control strategies were, as predicted, the most adaptive strategies for the academics, however secondary control strategies were not the most adaptive strategies for the supermarket workers. These findings supported the primacy/back-up model, suggesting that all employees, whether they have low or high job autonomy, should rely on primary control strategies when they face a difficulty at work. However, as problems have now been identified with the primary and secondary control scale and job autonomy scale, further research needs to re-examine these hypotheses.

163

2 Chapter 3 - Study Two

164

2.1

Abstract

This study aims to re-test the proposal that job autonomy influences the amount of control strategies that employees use, the relationship between the control strategies and job satisfaction. This study attempted to overcome the major limitations identified in study one, concerning the primary and secondary control scale and the job autonomy scale. Furthermore, this study examined the influence of two new variables, namely need for job autonomy and social support at work. Two occupational groups that were expected to differ in their levels of job autonomy (i.e., secondary school teachers and academics) were compared. It was expected that the academics would report higher job autonomy, higher primary control, and lower secondary control than the teachers. Furthermore, it was expected that primary control would be more adaptive for the academics, whereas secondary control would be more adaptive for the teachers. These hypotheses were not supported however, as both groups reported equally high levels of primary and secondary control, and primary and secondary control were not related to job satisfaction. These inconsistent results prompted a review of the underlying assumptions of the study. Some methodological limitations were identified in the hypotheses examining job autonomy and the control strategies. Despite this, support for the remaining hypotheses highlighted the importance of social support at work in predicting job satisfaction.

165

2.2

Proposal for Study Two

This study re-examines the proposal that job autonomy influences the use and the adaptiveness of primary and secondary control strategies. It attempts to overcome the limitations identified in study one. This study uses: a) a revised version of the Primary and Secondary Control Scale; b) a new measure of job autonomy; and c) different occupational groups for comparison. Furthermore, this study incorporates recent research suggesting that the need for job autonomy mediates the relationship between job autonomy and job satisfaction, and examines how social support at work influences job satisfaction. These changes will now be discussed. 2.2.1 a) The Primary and Secondary Control Scale The Primary and Secondary Control Scale, developed by Heeps et al., (2000) was implemented in study one because it was one of the best scales that concurred with Rothbaum et als., (1982) and Heckhausen and Schulzs (1995) definition of control. However, the scale was exploratory, and study one highlighted some problems with the scale. As such, a review was conducted on the scale in collaboration with RoseAnne Misajon. This review, which was based on factor analyses of the scale, highlighted several problems with the scale. These problems involved: i) the stem of the item; ii) the content of the item; and iii) the rating scale. From this review, a third and fourth edition of the Primary and Secondary Control Scale was developed. Factor analyses conducted on the first and second edition of the Primary Control and Secondary Control Scale were reviewed (e.g., Cahill, 1998; Cousins,

166

2001; Maher & Cummins, 2001, Misajon, 2002; Misajon & Cummins, in press; Spokes, 1998). These analyses were based on a variety of samples, including elderly people, people with arthritis, people with multiple sclerosis, and academics. Each researcher tended to make minor changes to the scale, where they may have excluded some items, or changed the wording of others, to make the scale more suitable to their sample. These researchers then conducted exploratory factor analyses on the scale, and found that the items initially loaded on 3, 4 or 5 factors. As they were often unable to explain these factors, they then requested two factors. The resulting analyses are displayed in Table 16. In this table, items that were excluded from that particular version of the scale are represented by NA. Items that did not load on any factors, or alternatively loaded on both factors are represented by a dotted line. Items that loaded on the primary control factor are bolded, whilst items that loaded on the secondary control factor are not bolded. This table demonstrates that the primary control items generally factored well, however the secondary control items often loaded on both factors. The primary control scale will be discussed first.

167

Table 16- Factor Analysis of Primary and Secondary Control Scale Study Primary Control Items New ways to achieve goal Persistence Remove obstacles Invest time Learn skills Ask for help or advice Effort to make it happen Secondary Control Items Positive Re-interpretation Wisdom Illusory optimism Downward social comparison Past success Vicarious Positive approach Goal disengagement Present success Predictive negative Attribution Behavioural avoidance Active avoidance Sour grapes Support Give up 1 0.56 0.75 0.71 0.60 0.59 0.37 0.31 0.46 0.47 0.54 0.65 0.71 0.76 0.63 ---0.57 ---0.43 0.30 ---0.39 ---NA 2 0.80 0.73 0.70 0.61 0.74 0.56 0.53 0.32 0.58 ---0.50 0.52 0.60 NA NA 0.73 0.66 0.76 ---NA NA 0.32 0.59 3 0.60 0.64 0.66 0.72 0.76 0.36 ---0.60 0.49 0.52 0.68 0.71 0.41 ---NA 0.62 0.37 0.41 ---NA NA ----0.49 4 0.53 0.62 0.56 0.66 0.65 0.61 0.54 0.57 0.51 0.67 0.43 0.31 0.98 0.65 NA 0.72 ---0.43 ------NA 0.57 -0.58 5 0.56 0.84 0.74 0.81 0.49 0.68 0.82 0.44 0.41 0.37 0.77 0.81 0.82 0.54 0.82 0.58 0.74 0.74 0.49 0.67 NA 0.77 NA 6 0.70 0.71 0.51 0.51 NA NA 0.39 ---------------------0.72 0.74 0.61 0.40 0.69 0.74 0.61 ------7 0.79 0.81 NA 0.72 0.67 NA 0.65 0.42 0.47 ---0.46 ---------0.76 ---0.54 ---0.76 0.76 0.54 -------

Studies 1 = Maher (2001); 2 = Misajon (2000); 3 = Spokes (1998); 4 = Cahill (1998); 5 = Misajon (2001); 6 = Study one; 7 = Cousins (2001) Bolded factor loadings refer to the primary control factor.

168

2.2.1.1

Primary Control Scale

2.2.1.2

i) Stem of Primary Control Items The factor analyses in Table 16 demonstrate that the primary control items

generally factor well. However, one reason why the primary control items may have loaded on a different factor to the secondary control items is that the primary and secondary control items were presented separately in the scale. The primary control items began with when I find a goal that is difficult to reach, when I really want something and when something gets in the way of a goal, whereas the secondary control items all began with when something bad happens that I cannot change. The reason the two strategies have different stems is that it was originally assumed that primary control strategies were only used when a person faced a difficulty that they could change, and that secondary control strategies were only used when the difficulty could not be changed. This assumption may be incorrect however, as it is possible for people to use secondary control when they face a situation that they can change. For example, an employee may be upset that a co-worker is always late. They may know that if they use primary control and talk to their supervisor about the problem, the co-worker will be reprimanded, and as such begin to arrive on time. However, they may choose not to use primary control as they may then lose their friendship with the co-worker. Rather, they may implement secondary control, and tell themselves that the problem doesnt matter. Similarly, it is possible that people use primary control when they face situations that they cannot change. For example, an employee may dislike their work

169

times, yet be aware that the work times cannot be changed. Even so, they may attempt to change their working times through using primary control, and discussing solutions with their supervisor. The supervisor would presumably reject their proposal, and the primary control strategy would have failed. Despite knowing the possibility of primary control failure however, the employee may have decided to take a risk. As it is possible for primary and secondary control to be used in controllable and uncontrollable situations, the scale was changed so that the stems of the items are the same. The revised scale includes the primary and secondary control items together, with the following introductory sentence; Here are ways people deal with difficult situations in their lives. How often have you had these thoughts when facing a difficulty over the past week? Examples of these thoughts are it will work out okay in the end and I knew it would happen. The other control items which involved actions rather than thoughts had an alternative introductory paragraph; How often have you done these things when facing a difficulty over the past week? (i.e., I kept trying, I told someone about it, I worked to overcome it). 2.2.1.3 ii) Primary control Item Content As demonstrated in Table 16, all studies found that the items assessing new ways to achieve goals, persistence, remove obstacles, learn skills and invest time, loaded on the primary control factor. The items measuring effort to make it happen, and ask others for help or advice occasionally loaded on the secondary control factor. These two items were deleted as they were criticised for being similar to secondary control strategies. Specifically, the item referring to effort to make it happen,

170

generally worded as I think I can make it happen does not actually refer to the person putting in effort, and rather is similar to the secondary control strategy of illusory optimism (i.e., I know it will work out okay in the end). The other item referring to asking for help or advice was also deleted as it difficult to separate it from secondary control. Indeed asking the boss or someone who has some power over the problem for help or advice may be a means of changing the environment. However, discussions with people who have less power over the situation, such as friends, may only serve to make the person accept the problem. Although the remainder of the primary control items loaded on the primary control factor, there were still conceptual problems with the items. For example, the item referring to learning skills was deleted from the scale, as it is only relevant if the person is attempting to achieve something, and cannot be applied to the new stem, namely difficult situations. Furthermore, the item assessing investing time was omitted, as it was not necessarily indicative of primary control. A person may spend lots of time on a problem, yet not attempt to change the environment. The remainder of the items were criticised as they examined whether the employees believed that they could change their situation, rather than examining how they change their situation. For example, the items I think I can make it happen, I believe I can achieve it and I believe I can overcome it measure whether a person believes that they can change a situation. As discussed in chapter 2, it is questioned whether believing that one can change a situation is a measure of primary control. A person may report that they can change a situation for a variety of reasons, not just if they use primary control strategies when they face difficulties. For example, a person

171

may report that they can change their environment because they have high optimism. Alternatively, they may be using the secondary control strategy illusory optimism, where they tell themselves that everything will work out okay in the end. As such, the revised primary control scale, displayed in Table 17, is changed to examine perceived strategies (e.g., exerting more effort, working harder) rather than beliefs. Table 17- Original and Revised Primary Control Items Primary Control Strategy New ways to achieve goal Effort to make it happen Invest time Persistence Remove obstacles 2.2.1.4 2nd Edition (Heeps et al., 2000) I think about different ways to achieve it I think I can make it happen I think about it a lot I believe I can achieve it I believe I can overcome it 4th Edition (Maher et al., 2001) I looked for different ways to overcome it I worked to overcome it NA I kept trying I worked out how to remove obstacles

iii) Rating Scale The primary control items were originally rated on a 10-point scale ranging

from 1 (do not agree at all) to 10 (agree completely). This rating scale indicates whether an individual agrees that they have used a strategy, not how much they have used a strategy. Two people may report that they agree completely that they have used a strategy, however one may use it 10 times a day, whilst the other may use it once a week. As the scale did not differentiate between these people, the primary control rating scale was changed to assess frequency. In order to reduce inaccuracies, the scale was changed from measuring the control strategies that people generally use when they face a difficulty to examining

172

the strategies people have used over the past week. As such, the rating scale ranged from 0 (never) to 10 (every time). 2.2.1.5 Secondary Control

2.2.1.6

i) Stem of Secondary Control Items As previously discussed, it was thought that the primary control items may

have loaded on a different factor to the secondary control items because the stems of the items were different. In order to overcome this, the secondary control items were placed with the primary control items. The stem of the item was changed from when something bad happens that I cannot change to how often have you done these things when facing a difficulty over the past week. 2.2.1.7 ii) Item Content As demonstrated in Table 16, a few secondary control items loaded on both the primary control factor and the secondary control factor. There did not appear to be a consistent pattern in these studies however, with some studies finding that an item loaded on a secondary control factor, whilst others found that it loaded on a primary control factor. It was originally expected that the secondary control items would form one factor, however it is now proposed that each item measures a different strategy and that these strategies are independent. One person may use one secondary control strategy in all situations, and so not use any of the others. This proposal has implications for the scoring of the secondary control scale, and also for factor analyses of the scale.

173

In regard to scoring, the proposal that respondents scores on secondary control items may not be consistent suggests that the secondary control items cannot be aggregated. However, secondary control can still be measured by using the highest scoring item. This scoring procedure will be explained in detail later. In regard to factor analyses, the proposal that respondents scores on secondary control items are not consistent may explain why the secondary control items loaded on more than one factor. Respondents may report different scores on all the secondary control items, and thus they would not be expected to cluster together. As such, rather than eliminating any items which loaded on a primary control factor, the items were examined in terms of their theoretical usefulness. Many of the items were similar to others, such as past success and present success, and positive approach and behavioral avoidance. Present success (I think about my success in other areas) encompasses past success (I remember I have accomplished a lot in life). Furthermore, positive approach (I do something nice to take my mind off things) and behavioral avoidance (I do some physical exercise or try to relax) could be combined to measure active avoidance (I do something to take my mind off things). Two items that had been deleted from the first edition of the scale, namely denial and support, were reinstated. Denial, measured by the item I ignored it was deleted from previous versions of the scale as it was thought to be similar to the item for goal disengagement (i.e., It doesnt matter). However, telling oneself that a problem is not important is clearly different from denying that the problem exists. Intuitively, goal disengagement may be more adaptive than denial.

174

Support, measured by the item told someone about it was also added to the scale. It was originally deleted from the first edition of the scale as it was a behavioural strategy. It was assumed that all secondary control strategies had to be cognitive strategies. This is not the case however, and Heckhausen and Schulz (1995) recommend that the distinction between primary and secondary control should not be based on behavioural versus cognitive, rather whether it involves changing the environment versus changing the self. Support allows the person to change themselves and become more likely to accept a situation. After this theoretical analysis, 12 secondary control strategies remained (refer to Table 18). These strategies were grouped according to their purpose. All of the strategies are designed to make the person feel better about their situation, however they may do this by reducing negative feelings (i.e., self-protective) or by increasing positive feelings (i.e., self-affirmative). As demonstrated in Table 19, people may reduce negative feelings by telling themselves that a difficult situation is not their fault, that they knew it would happen, or that it doesnt matter. People may increase positive feelings however by thinking that they are better off than many other people, and thinking about areas of their life in which they have been successful.

175

Table 18- Original and Revised Secondary Control Items Secondary control strategy Positive reinterpretation Wisdom Illusory-optimism Downward social Omparison Past success Vicarious Positive approach Goal disengagement Predictivenegative Attribution Active avoidance Sour grapes Present success Denial Support Second Edition I can see that something good will come if it I remember you cant always get what you want I know thing will work out OK in the end I remember I am better off than many other people I remember I have already accomplished a lot in life I remember the success of my family and friends I think nice thoughts to take my mind off it I tell myself it doesnt matter I dont feel disappointed because I knew it might happen I can see it is not my fault I ignore it by thinking about other things I realise I didnt need to control it anyway I think about my success in other areas NA NA 4th Edition I looked for something else that was positive in the situation I cant always get what I want It will work out okay in the end I am better off than many other people NA I thought of the success of my family or friends NA It doesnt matter I knew it would happen It was not my fault I did something different, like going for a walk NA I thought about my success in other areas. I ignored it Told someone about it

Table 19- Functions of the Secondary Control Strategies Use Selfprotective Selfaffirmation Definition of use Reduces the negative impact of the situation Increases positive feelings about self Secondary Control Strategy Illusory optimism, goal disengagement, predictive negative, attribution, denial, wisdom Downward social comparison, vicarious, present success, support, positive reinterpretation, active avoidance

176

2.2.1.8

iii) Rating Scale As with the primary control items, the secondary control items were changed to

measure frequency. Each strategy was rated on an 11-point scale ranging from 0 (never) to 10 (every time). 2.2.1.9 Summary Following a review of the factor analyses conducted on the primary and secondary control scale, and an investigation of the item stem, the item content, and the rating scale, a revised scale was developed. This scale, presented in Appendix H will be implemented in the second study. One further point that requires discussion however, is the scoring of the control scale. 2.2.1.10 Scoring the primary and secondary control scale

Previous versions of the Primary and Secondary Control Scale (Heeps et al., 2000) averaged across the strategies to obtain an overall score for primary control score and an overall score for secondary control. The problem with this method however, is that a person may report that they use one secondary control strategy every time (10) and report never (0) for the remaining strategies. Calculating the average level of secondary control in this situation would result in a low score. As they used a secondary control strategy every time they faced a difficulty in the previous week, a low score is not representative of their secondary control use. One solution to this problem is to take the highest score for primary control and the highest score for secondary control. Using this method, a person who reports 10

177

for one secondary control strategy and 0 for the rest would receive a score of 10 for secondary control. Another person may report different scores for all secondary control strategies, including a 5, 7, 8, 10, 2, 3, 4. This person would also receive a score of 10, as it is the highest score. The fact that the second person has higher scores on other secondary control items does not mean that the person uses more secondary control strategies, only that they use a greater variety of secondary control strategies. 2.2.2 b) Job Autonomy Scale The next limitation that was identified in study one concerns the measurement of job autonomy. The autonomy scale implemented in study one was a multidimensional scale. The scale examined specific facets of the work, such as variety of work, pace of work, scheduling of rest breaks, and interaction with others. This multidimensional scale was advantageous as it prompted the employees to consider all aspects of their work. This is important as employees may fail to consider some facets of their work. They may have accepted for example that they cannot change their pay, policies and amount of interruptions, and thus no longer expect to be able to make choices in these areas. The multidimensional scale ensures that all workers think about the same job facets. However, it is now recognised that the multidimensional scale is also problematic. Although the scale prompts employees to consider all aspects of their work, some of the facets may not be appropriate for them, or important to them. With a global scale, the respondent can include facets that are not specified in the facet scale, and exclude facets that are not relevant to their workplace. As such, their

178

response is only based on facets that they think are important. They may exclude some facets because they have lowered their expectations, however if they have accepted them, then they are not expected to influence their levels of job satisfaction. 2.2.3 c) Occupational Groups This study will compare two occupational groups that have different levels of job autonomy, low and high. As in study one, university academic staff have been selected for the high job autonomy group. Academics traditionally have flexibility in their work and freedom to pursue their own research interests (Winefield, 2000). They can often choose among a variety of tasks, including research, teaching, and administration (Fisher, 1994). Whether this theoretical expectation existed in practice was tested in study one. The results demonstrated that the academics reported a level of job autonomy which was 53%SM. It could not be ascertained whether this score was high however, as there was few comparative studies for Gansters (1989, cited in Dwyer & Ganster, 1991) multidimensional scale of job autonomy. Study two will overcome this problem by relying on a scale, which has been used more extensively. Secondary school teachers have been selected for the low job autonomy group. Teachers have been selected rather than supermarket workers because this study is attempting to minimise the differences between the groups. In study one, it was demonstrated that although the supermarket operators reported higher job autonomy, and higher secondary control than the academics, the two groups reported similar levels of job satisfaction. However, there were differences between the two occupational groups that may have accounted for the similar levels of job satisfaction.

179

The supermarket workers would have experienced fewer difficulties at work than the academics, and may have had lower job expectations than the academics. Study two attempts to examine two groups which have similar experiences at work, but which have differing levels of job autonomy, namely secondary school teachers and academics. Although both occupational groups deliver education to students and have similar roles, it is expected that teachers will report lower job autonomy. Although few studies have examined Australian teachers levels of job autonomy, a recent report proposes that although the Government attempted to empower schools and teachers through providing schools with more responsibility, teachers are experiencing reduced autonomy (Senate Employment, Education and Training References Committee, 1998). Teachers are reporting that they want to have more involvement in decision-making. One study which interviewed 956 Australian teachers about the changes they felt were necessary to reduce stress (Teacher Stress in Victoria, 1990) found that the most common change (80%) was to increase staff collaboration and communications. They also mentioned increasing consultations before major decisions are made. The type of decisions that the teachers want to be consulted on concern curriculum selection, development and implementation (Senate Employment, Education and Training References Committee, 1998). It is particularly important that the teachers are involved in curriculum selection so that they can have control over the means of producing the results by which they will then be judged (Cole, 1989).

180

In summary, this study will test the major hypotheses by comparing academics and teachers. Although little research has examined these two groups, it is expected that the academics will report higher job autonomy than the teachers. 2.2.4 d) Need for Job Autonomy In study one, it was assumed that high job autonomy was beneficial for all employees. This assumption was based on Karasek and Theorells (1990, p. 12) proposal that if jobs were redesigned with high job decision latitudedemands would be seen as challenges and would be associated with increased learning and motivation, with more effective performance and less risk of illness. However, it must be noted that other researchers have suggested that people may differ in the extent to which they like to exercise control over their environment (Burger & Cooper, 1979; Parkes, 1989). This difference in need for autonomy may influence the relationship between job autonomy and job satisfaction, where job autonomy may have greater influence on job satisfaction when need for job autonomy is high. Only a few studies have examined the moderating role of need for job autonomy on the relationship between job autonomy and job related outcomes (e.g., de Jonge, Landeweerd & Breukelen, 1994, cited in de Rijk, Le Blanc, Schaufeli, & de Jonge, 1998; Nicolle, 1994). These studies have tended to produce inconsistent findings. For example, de Rijk et al., (1998) cite de Jonge et als., (1994) study as providing evidence that the need for autonomy moderated the relationship between job autonomy and emotional exhaustion and health complaints. When de Rijk et al.,

181

(1998) replicated the study however, they failed to find support for the moderating role. One other study, conducted by Nicolle (1994) provides some support for the moderating role of need for autonomy. This study demonstrated that for nurses with a low need for autonomy, job autonomy was positively related to absenteeism, however for nurses with a high need for autonomy, job autonomy was not related to absenteeism. These results must be interpreted with caution however, as only 3 of the 36 analyses were significant. One further study has been reported to provide evidence for the moderating role of need for autonomy. De Rijk et al., (1998) cited Gaziels (1989) study on school administrators as being supportive of the hypothesis. According to De Rijk et al., (1998) this study demonstrated that for administrators who had a low need for autonomy, job autonomy was not related to job satisfaction, whereas for administrators who had a high need for autonomy, there was a positive relationship between job autonomy and job satisfaction. Examination of the study demonstrates that this is not the case however. Gaziels (1989) study did not examine the relationship between autonomy and job satisfaction for workers with differing levels of job autonomy. Rather, the study examined the major predictors of a perceived deficiency in autonomy. In summary, it has been suggested that employees may differ in their need for autonomy and that the relationship between job autonomy and job satisfaction may differ depending on this need. As only a few studies have examined this proposed

182

moderating effect, and as the studies tend to be inconsistent, clearly further research is needed. 2.2.5 e) Addition of Social Support As mentioned in chapter 1, the job demand-control model was extended to include social support (Johnson & Hall, 1988; 1994; Johnson, Hall & Theorell, 1989; Karasek & Theorell, 1990). Social support at work refers to overall levels of helpful social interaction available in the job from both co-workers and supervisors (Karasek & Theorell, 1990, p.69). Two major types of social support have been identified, namely emotional support and instrumental support. According to Karasek and Theorell (1990, p. 70), emotional support refers to the degree of social and emotional integration and trust between co-workers, supervisors and others, whereas instrumental support refers to extra resources or assistance with work tasks given by co-workers or supervisors. The job demand-control-support model proposes that social support at work predicts job satisfaction. Study one did not examine social support at work as it focussed on understanding how job autonomy influences the control strategies, and on personality and life satisfaction. However, after examining research on the relationship between social support and job satisfaction further, social support appears to be an extremely important predictor, and as such, study two will examine social support at work in more detail. Social support at work has been shown to directly and indirectly increase job satisfaction. In regard to the direct effects, several researchers have demonstrated that

183

social support at work is positively related to job satisfaction (r = 0.52; Dollard et al., r = 0.66, Munro et al., 1998), and negatively related to job dissatisfaction (r = -0.29, r = -0.28; LaRocco, House & French, 1980). These studies suggest that workers who report higher social support tend to be more satisfied with their jobs. One possible explanation for the positive relationship between social support and job satisfaction is that social support reduces the negative effects of work demands. This explanation, known as the buffering hypothesis, proposes that social support at work buffers the potentially negative effects of high demands on job satisfaction. Only a few studies have examined the buffering hypothesis for job satisfaction. A review of these studies, conducted by Van Der Doef and Maes (1999) demonstrated that only two (i.e., Karasek, Triantis & Chaudry, 1982; Landsbergis, Schnall, Dietz, Friedman & Pickering, 1992) of the six studies (Chay, 1993; de Jonge & Landeweerd, 1993, cited in Van der Doef & Maes, 1999; Melamed, Kushnir & Meir, 1991; Parkes & von Rabenau, 1993) that examined the buffering hypothesis were supportive. Their review found no major differences among the studies to account for the inconsistent findings except that both supportive studies used male samples and the others used mixed or female samples. One difference among the studies that may explain the findings is the operationalisation of social support at work. For example, Karasek et al., (1982) measured tolerance of supervisor, attentiveness of supervisor, instrumental support of supervisor, demands of supervisor, number of co-workers, instrumental co-worker support, and emotional co-worker support. Alternatively, Chay (1993) relied on the

184

Interpersonal Support Evaluation List (Cohen, Kamarack, Mermelstein & Hoberman, 1985) which measures appraisal support, belonging support, tangible support and esteem support. A similar and briefer scale was employed by Landsbergis et al., (1992), who relied on Karasek and Theorells (1990) scale. This scale measures emotional and instrumental support from co-workers and supervisors. There is certainly no agreed upon way of measuring social support at work (Unden, 1996), and as such, it is unclear if the operationalisation of social support influenced the results. What is clear is that the buffering role of social support requires more investigation. In summary, although it is intuitively expected that social support at work would reduce the negative effects of job demands or job stressors, the results are far from consistent. As there are such few studies however, more research is required.

185

2.3

Model of Job Satisfaction

A revised model of job satisfaction, displayed in Figure 6, will be tested. This model is similar to that presented in Figure 2, as the major proposal of the model is that primary and secondary control mediate the relationship between job autonomy and job satisfaction. However, this model includes new sections on social support and need for job autonomy. In Figure 6, these changes are represented by bolded variables and arrows. These new proposals will be discussed. It is now proposed that the relationship between job autonomy and job satisfaction is moderated by need for job autonomy. It should not be assumed that all employees desire high autonomy. Indeed, some workers may have low job autonomy yet still report high job satisfaction because they do not desire freedom and independence in their job. Need for job autonomy and job autonomy predict the interaction term (i.e., need for job autonomy x job autonomy), which in turn predicts job satisfaction. It is also proposed that social support at work influences job satisfaction directly and indirectly. It is expected to be positively correlated with job satisfaction, and to also moderate the effect of work difficulties on job satisfaction. In Figure 6, this is represented by the interaction term. Difficulties at work and social support at work together predict the interaction term (i.e., difficulties x social support), which in turn predicts job satisfaction.

186

Figure 6 -Revised Model of Job Satisfaction for Study 2

Difficultie s at work

Social Support at work

Job Autonomy x Primary Control

Difficulties x Social Support

Primary Control

Job Autonomy Secondary Control

Job Satisfaction

Job Autonomy x Secondary Control

Personality

Life Satisfaction

Job Autonomy x Need for job autonomy

Need for job autonomy

187

2.4

Aims and Hypotheses

This study will compare levels of job autonomy, control strategies, and job satisfaction reported by university academic staff and secondary school teachers. It will also test the extent to which job autonomy mediates the relationship between primary and secondary control strategies using the revised measures of job autonomy and control strategies. The hypotheses are as follows:

1) Job autonomy will be positively related to job satisfaction, and the academics will report higher job autonomy than the teachers. This hypothesis tests the basic assumptions of the study. It needs to be demonstrated that job autonomy is related to job satisfaction, and that comparisons made between the two occupational groups are valid.

2) The academics will report more primary control, and less secondary control than the teachers. As the academics have higher job autonomy, they are expected to be more likely to successfully implement primary control strategies than the teachers. As secondary control is used to compensate for, and avoid future primary control failure, it is expected that the teachers will report more secondary control than the academics

3) Job autonomy will be positively related to primary control, and negatively related to secondary control.

188

As in hypothesis two, this hypothesis is examining whether job autonomy influences the use of primary and secondary control. However, unlike hypothesis two, it is based on the measured level of autonomy rather than the expected occupational level

4) Primary control will be more positively related to job satisfaction than secondary control for the academics, and secondary control will be more positively related to job satisfaction than primary control for the teachers. This study proposes that job satisfaction results from a match between job autonomy and control strategies. Based on the discrimination model, it is proposed that primary control is most adaptive for employees who can control their work environment (i.e., high job autonomy), and that secondary control is most adaptive for employees who have little control over their environment (i.e., low job autonomy). Although primary control is generally more adaptive than secondary control, the teachers have a high probability of experiencing primary control failure when implementing primary control strategies, and thus it is expected that, for them, secondary control strategies will be more adaptive.

5) The relationship between the control strategies and job satisfaction is moderated by perceived job autonomy. This hypothesis, like hypothesis four, is testing whether job autonomy influences the relationship between the control strategies and job satisfaction. Unlike

189

hypothesis four however, it is based on the measured level of job autonomy rather than the assumed level of autonomy based on the occupation.

6) The relationship between job autonomy and job satisfaction is mediated by primary and secondary control strategies. This hypothesis is testing an explanation for the relationship between job autonomy and job satisfaction. This explanation proposes that people who have high job autonomy have high job satisfaction because of their use of the control strategies. These workers use more primary control and less secondary control, and are thus able to overcome their difficulties.

7) The academics will report higher job satisfaction and higher life satisfaction than the teachers. The academics are expected to report higher job satisfaction than the teachers as they have higher job autonomy, and use more primary control and less secondary control. This level of job satisfaction is expected to influence their level of life satisfaction.

8) The influence of work difficulties on job satisfaction is moderated by levels of social support at work. This hypothesis is based on the job demand-control-support model (Karasek, 1979) which proposes that social support can reduce the effect of demands at work.

190

9) The relationship between perceived job autonomy and job satisfaction is moderated by need for autonomy. People may differ in their need for autonomy, and this will influence the relationship between job autonomy and job satisfaction.

10) Job autonomy, control strategies, life satisfaction, personality, difficulties at work, and social support at work, predict job satisfaction. These are all of the variables included in Figure 6. These are the major predictors of job satisfaction.

191

2.5

Method

2.5.1

Participants The sample consisted of 108 university academic staff, and 97 secondary

school teachers. The academics were obtained from one university, whereas the secondary school teachers were obtained from 20 Government schools. For the academics, the response rate was 21%. The response rate of the teachers could not be calculated as the questionnaires were collected from the staff room only if the teachers were interested in completing the survey. The demographic characteristics of the sample are displayed in Table 20. The bolded values demonstrate where the largest proportion of the sample lies, which tends to be fairly consistent across the groups.

192

Table 20- Demographics of the Academics and Teachers Variable Gender Age % Academic 49 51 0 8.3 32.4 44.4 14.8 13 27.8 22.2 7.4 29.6 6.5 7.4 47.2 32.4 6.5 % Teachers 47 53 2.1 15.5 37.1 39.2 6.2 11.3 11.3 13.4 18.6 45.4 4.1 12.4 45.4 27.8 10.3

Years in Occupation

Hours worked per week

Male Female 18-25 26-35 36-45 46-55 56+ 0-5 6-10 11-15 16-20 20+ 21-30 31-40 41-50 51-60 61+

2.5.2

Materials Both the academics and the teachers received a plain language statement (refer

to Appendix I) and an anonymous questionnaire. The questionnaire consisted of several scales, which measured job autonomy, need for job autonomy, primary and secondary control, work difficulties, job satisfaction, life satisfaction, personality and social support at work. 2.5.2.1 Job Autonomy As discussed in the rationale for study two, a global measure of job autonomy was administered. This scale developed by Hackman and Oldham (1975) is part of a larger scale, the Job Diagnostic Survey. This scale is the most commonly used instrument for measuring job autonomy (Spector, 1986). It consists of three items that

193

assess overall perceived job autonomy, such as in my job, I can decide on my own how to go about doing my work (refer to Appendix J). Although the psychometric properties of the scale have been questioned in the past (Fried, 1991; Fried and Ferris, 1991), a major review which examined 15 years of empirical research on the psychometric properties of the scale provided some support. Taber and Taylor (1990) demonstrated that the average test-retest correlations for the scale were moderate (r = 0.63), internal consistency was moderate (0.69), and there was good discriminant validity. Although these psychometric statistics are not exceptional, the use of the scale has been supported in a recent review conducted by Boonzaier, Flicker and Rust (2001). Furthermore, it must be noted that as mentioned by Breaugh (1989), a better alternative is not available. Breaugh (1989, Breaugh, 1998) actually developed a new measure of job autonomy, however this scale was deemed not to be appropriate for this study as like Gansters (1989, cited in Dwyer & Ganster, 1991) scale, it is multidimensional. As such, this study used the autonomy items of the Job Diagnostic Survey (Hackman & Oldham, 1975). In this study, Cronbachs Alpha was 0.83. 2.5.2.2 Need for Job Autonomy As there are only a few studies that have examined need for job autonomy, the measures of need for job autonomy were reviewed. First, Fung-Kam (1998) tested preference for job autonomy using Edwards (1959) Personal Preference Schedule. This scale consists of 28 sets of paired statements representing different personality traits and a score is given to the respondent who chooses the statement representing the

194

personality trait of need for autonomy. The major problem with this scale is that it refers to general autonomy, and not specifically to autonomy at work. Another need for job autonomy scale is Algeras (1981, cited in Landeweerd & Boumans, 1994) scale. This scale asks the respondents to rate the attractiveness of various work situations. Although this scale may have been adequate, to date it has only been published in Dutch, and as such was not viable. An exploratory scale was developed by de Rijk et al., (1998). This scale consists of four items which examine how important it is for the person to set the pace of their tasks, have control over what they do at work and the way that they do it, doing their own planning at work, and giving orders instead of receiving them. This scale was selected for the current study even though psychometric statistics have not been produced, as the items have face validity. These items were rated on a 10-point scale, ranging from 1 (not at all important) to 10 (could not be more important; refer to Appendix K). As this scale is exploratory, a factor analysis was conducted on the scale to ensure that the items were measuring need for job autonomy. The assumptions were met, where Bartletts test of sphericity was large and significant, and Kaiser-MeyerOlkin (KMO) measure of sampling adequacy exceeded 0.6. A principal components analysis with direct oblimin rotation yielded one factor. Examination of the eigenvalues however demonstrated that the second factor had an eigenvalue of 0.999, and as such a two-factor solution was tested. This analysis, displayed in Table 21, demonstrates that item four (i.e., How important is it for you to give orders instead of receiving them) loaded on the second factor. Item four is different to the other three

195

items as it may also measure need for authority. As a result, item four was deleted from the scale. When all four items were included in the scale, Cronbachs alpha was low (0.56), however when item four was deleted, Cronbachs alpha was adequate (0.77). Table 21- Factor Analysis of the Need for Job Autonomy Scale No. 1 2 3 4 Item How important is it for you to set the pace of your tasks at work. How important is it for you to have control over what you do at work and the way that you do it. How important is it for you to do your own planning at work. How important is it for you to give orders to work instead of receiving them. Eigenvalues % of variance Cumulative variance Cronbach's Alpha (for revised scale) F1 0.83 0.87 0.80 0.99 2.098 52.45 52.45 0.77 0.999 24.98 77.43 F2

Loadings less than 0.40 are excluded; Bolded items are included in the scale 2.5.2.3 Primary control and Secondary Control As discussed in the rationale, the Primary and Secondary Control Scale developed by Heeps et al., (2000) was revised for this study (Maher et al., 2001). The scale now includes four primary control items and 12 secondary control items (refer to Appendix H). These items are rated on a 11-point scale ranging from 0 (never) to 10 (every time). Although the control strategies were aggregated in study one, it now appears that this scoring method is flawed. The items cannot be aggregated as people may use one strategy all the time, and never use the others. Using the average, they would receive a score that is not representative of the frequency of secondary control

196

strategies (i.e., every time). As such, an alternative solution used here is to record the highest frequency for primary control strategies and the highest frequency for secondary control strategies. 2.5.2.4 Work Difficulties Work difficulties were measured in the Primary and Secondary Control Scale (Maher et al., 2001; refer to Appendix H). Prior to assessing how the employees deal with their work difficulties, the scale assesses the frequency of work difficulties. Specifically, the item is how often do you have difficulty doing something at work. The rating scale ranges from 1 (never) to 10 (all the time). 2.5.2.5 Job Satisfaction Two scales of job satisfaction were administered; a facet scale and a global scale. The facet scale was changed from the Job Descriptive Index (JDI) in study one to the Minnesota Satisfaction Questionnaire (MSQ; Weiss et al., 1967; refer to Appendix L). The MSQ was used because unlike the JDI, which only examines five facets of the job, the MSQ examines 20 facets. It was thought that a greater understanding of the groups could be obtained by using the MSQ. Furthermore, the items in the MSQ can be aggregated to measure intrinsic and extrinsic job satisfaction. Intrinsic job satisfaction refers to how people feel about the nature of the tasks, whereas extrinsic job satisfaction refers to how people feel about aspects of the work situation that are external to the work itself (Spector, 1997). This scale has adequate reliability where Cronbachs alpha ranges from 0.82 to 0.88 and discriminant validity has been demonstrated (Hirschfeld, 2000).

197

The facet measure is only useful to gain insight into the teachers and academics level of job satisfaction. It cannot be used as the dependent variable however as a facet scale cannot be aggregated (Ironson et al., 1989). It may exclude areas that are important to the respondent, or include areas that are unimportant to the respondent. As such, a one-item measure of job satisfaction was used as the dependent variable. Although internal consistency cannot be established with a single-item measure, the single item measure of job satisfaction has been shown to correlate with other measures of job satisfaction, where r = 0.63 (Wanous et al., 1997). 2.5.2.6 Life Satisfaction As in study one, the subjective dimension of the Comprehensive Quality of Life Scale (Com-QOL) developed by Cummins (1997) was used to assess satisfaction with seven domains of life, including material well-being, health, productivity, intimacy, safety, community and emotional well-being (refer to Appendix E). An 11point scale was utilised, ranging from 0 (completely dissatisfied) to 10 (completely satisfied) 2.5.2.7 Personality The extroversion and neuroticism subscales of the NEO Five Factor Inventory, developed by Costa and McCrae (1992) were used to measure personality. This scale, discussed in study one, contains 12 items to measure extroversion and 12 items to measure neuroticism (refer to Appendix F). Convergent and discriminant validity of both of these personality factors has been established (Costa & McCrae, 1992, Leong & Dollinger, 1991; Tinsley, 1994).

198

2.5.2.8

Social Support at Work Social support at work was measured by Karasek and Theorells (1990) scale

which has two components; supervisor support and co-worker support (refer to Appendix M). Each component is measured by 4 items, and rated on a scale from 1 (not true at all) to 10 (could not be more true). Two items measure emotional support, and two measure instrumental support. Emotional support measures the degree of social cohesion in the work group, whilst instrumental support measures the amount of assistance given with work tasks. Although the scale measures emotional and instrumental support, the four items are summed to provide an overall support score. The items in the scale were changed slightly to ensure that they referred to the employee. Some of them were quite ambiguous, such as my supervisor shows concern and my supervisor pays attention. As these items could be interpreted in regard to work tasks or other employees, they were changed to my supervisor shows concern for me and my supervisor pays attention to me. Past studies using the original scale have demonstrated that the scale has adequate reliability with Cronbachs alpha ranging from 0.69 to 0.89 (Karasek et al., 1998), and 0.81 to 0.87 (Pelfrene, Vlerick, Mak, De Smets, Kornitzer & De Backer, 2001). Furthermore, factor analyses have demonstrated that the supervisor support items load on a different factor to the co-worker support items (Pelfrene et al., 2001).

199

2.5.3

Procedure Ethics approval was obtained from Deakin University, and the Department of

Education, Employment and Training (DEET). Consent was obtained from the Heads of School to recruit the academics, and from the Principals for the teachers. The recruiting procedure differed depending on the group. Five hundred academics within one University were sent a questionnaire package. If they chose to participate in the study, they completed the questionnaire and returned it using a reply paid envelope. For the teachers, each Principal that agreed to assist with the study was sent 10-15 questionnaires. They then discussed the questionnaires in their staff meetings, and left them in the staff room for the teachers to collect. On occasion, the Principals chose to distribute the questionnaires to a selection of staff members. These questionnaires were sent back to Deakin University using a reply-paid envelope. At the conclusion of the study, the participating Heads of School and the Principals received a summary of the results.

200

2.6

Results

2.6.1

Data Screening and Checking of Assumptions The data set for each occupational group was initially examined for missing

values, acquiescence, outliers, normality and linearity. Less than 5% of the values for academics and teachers were missing for any one item. As there was no pattern to these missing values, they were, as in study one, replaced with the group mean. Univariate outliers were identified in the primary and secondary control scale (5), the job autonomy scale (1), the facet job satisfaction scale (2), and the life satisfaction scale (18). These values were recoded to lie within three standard deviations of the mean. Normality was assessed using the skew/standard error<3, Kolmogorov-Smirnof values, frequency histograms, and normal probability plots. For the academics, job autonomy (-5.76), and co-worker support (-3.24) were mildly negatively skewed. For the teachers, job satisfaction (-3.79), supervisor support (-3.44), and co-worker support (-5.16) were mildly negatively skewed. As in study one, these variables were not transformed as transformations are not recommended for data that are mildly and naturally skewed (Tabachnick & Fidell, 1996). Rather, these were examined using the more conservative alpha level of 0.01. Linearity was assessed through bivariate scatterplots, and these appeared to demonstrate reasonable linear relationships.

201

2.6.2

Descriptive Statistics and Inter-Correlations Table 22 contains the means and standard deviations for the major variables in

the study for each occupational group. Whilst the teachers reported lower job autonomy than the academics, they reported similar levels of job satisfaction, and primary and secondary control. Table 23 displays the correlations among all of the major variables for the academics and the teachers. This table demonstrates that although job autonomy is correlated with job satisfaction, primary and secondary control strategies are not. Table 22- Means and Standard Deviations of Major Variables for Academics and Teachers Variable Job Satisfaction-One item Intrinsic Job Satisfaction Extrinsic Job Satisfaction Job Autonomy Primary Control Secondary Control Life Satisfaction Neuroticism Extroversion Supervisor Support Co-worker Support Difficulties at work Academics M SD 64.09 21.72 72.76 13.24 44.46 20.45 74.93 16.81 81.58 14.42 83.33 13.37 74.20 11.34 35.07 16.60 61.62 13.06 44.57 29.44 71.11 18.90 49.59 24.08 Teachers M SD 68.79 20.23 77.07 14.80 56.15 20.89 66.32 17.79 80.64 15.82 82.93 11.90 75.61 14.48 33.33 16.04 63.92 15.69 64.31 26.56 77.49 15.57 46.05 23.01

All scores have been converted to a percentage of scale maximum (%SM) which ranges from 0-100. The formula is (mean score for the original domain-1) x 100/ (number of scale points 1).

202

Table 23- Inter-Correlations for the Academics and Teachers

JS JS JA PC SC LS N E SS CS Di 0.37** -0.07 -0.06 0.38** -0.38** 0.22* 0.46** 0.42** -0.31** JA 0.51** 0.12 0.00 0.20* -0.38** 0.16 0.26** 0.15 -0.20* PC 0.14 0.06 0.40** 0.09 -0.07 0.20* -0.15 0.07 -0.07 SC 0.14 0.06 0.56** 0.03 0.03 0.13 -0.09 0.05 0.03 LS 0.46** 0.38** 0.15 0.27** -0.48** 0.26** 0.19* 0.23** -0.33** N -0.29** -0.14 -0.19 -0.08 -0.39** -0.44** -0.06 -0.31** 0.25** E 0.23* 0.17 0.01 0.19* 0.52** -0.32** 0.08 0.38** -0.09 SS 0.64** 0.39** 0.01 -0.05 0.36** -0.15 0.16 0.39** -0.13 CS 0.39** 0.34** -0.14 -0.06 0.26** 0.03 0.18 0.55** -0.26**

* p<0.05 , ** p>0.01; Correlations for teachers are bolded. JS = Job satisfaction; JA = Job autonomy; PC = Primary control; SC = Secondary control; LS = Life satisfaction; N = Neuroticism; E = Extroversion; SS = Supervisor support; CS = Co-worker support; Di = Difficulties at work 2.6.3 Preliminary Examination of the Primary Control and Secondary Control Scale As the primary and secondary control scale is exploratory, it will be examined here before the hypotheses are tested. The descriptive statistics displayed in Table 22 and 23 indicate that primary and secondary control did not behave as expected. Both the academics and the teachers reported high levels of primary control (M = 82%SM, M = 81%SM), and secondary control (M = 83%SM, M = 83%SM). Furthermore, the control strategies did not correlate with job satisfaction. One interesting finding however is that primary control was positively correlated with secondary control for both groups (r = 0.40, r = 0.56). Overall however, these statistics are inconsistent with study one, where the supermarket workers (M = 46%SM) reported significantly higher levels of secondary

203

control than the academics (M = 36%SM). Furthermore, primary control was moderately correlated with job satisfaction (r = 0.31, r = 0.44). A major difference between these two studies is the edition of the Primary and Secondary Control Scale. The scale was changed for study two, where a new scoring procedure was implemented. In the first study, the items were simply aggregated for each control strategy, however in the current study, the highest frequency for primary and secondary control was recorded. This method does not appear to have been successful however in differentiating the respondents. The frequency distribution, displayed in Table 24, demonstrates that 76% of the subjects reported a level of primary control between 77%SM and 100%SM, and that 84% of the subjects reported a level of secondary control between 77%SM and 100%SM. This range is concerning, suggesting that there may have been a ceiling effect. Table 24- Frequency of Primary and Secondary Control Value 33.33 44.44 55.56 66.67 77.78 88.89 100.00 Primary Control Frequency % 2.0 2.4 4.4 15.1 25.4 32.2 18.5 Secondary Control Frequency % 0.5 1.5 3.4 9.3 33.7 32.7 19.0

In order to examine how much the scoring procedure in this study influenced the resulting levels of primary and secondary control, the Primary and Secondary Control Scale was also examined as in study one, where the average was calculated. Before the items were aggregated, a factor analysis was conducted on the scale.

204

The assumptions were met, where Bartletts test of sphericity was large and significant, and Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy exceeded 0.60. A principal components analysis with direct oblimin rotation yielded four factors. These four factors accounted for 57% of the variance. Only 9 of the 16 items loaded on only one factor. Two of the primary control items loaded on the first factor, however a secondary control item also loaded on this factor. The remaining secondary control items (i.e., 6) were equally distributed among the second, third and fourth factor. There was no pattern to these items, and the four-factor solution could not be interpreted. The analysis was repeated requesting three factors to determine if a threefactor solution could be useful. This analysis was no clearer however, with the primary control and secondary control items loading on all three factors. The secondary control items that loaded on different factors did not appear to be measuring different functions of secondary control. In response to these analyses, a two-factor solution was requested. This analysis provided a much clearer solution, with the four primary control items loading on the first factor, and six of the 12 secondary control items loading on the second factor. It must be noted however, that this two-factor solution only accounted for 40% of the variance. Another factor analysis was conducted with only the bolded items in Table 25. This analysis demonstrated that the two factors accounted for 49% of the variance.

205

Table 25- Factor Analysis of the Revised Primary and Secondary Control Scale No. pc1 pc2 pc3 pc4 sc1 sc2 sc3 sc4 sc5 sc6 sc7 sc8 sc9 sc10 sc11 sc12 Item I looked for different ways to overcome it. I kept trying. I worked to overcome it. I worked out how to remove obstacles. It will work out okay in the end. I knew it would happen. I cant always get what I want. It doesnt matter. I am better off than many other people. It was not my fault. I told someone about it. I thought of the success of my family or friends. I thought about my success in other areas. I did something different, like going for a walk. I ignored it. I looked for something else that was positive in the situation. Eigenvalues % of variance Cumulative variance Cronbach's Alpha (for revised scale) F1 0.67 0.77 0.82 0.59 0.41 F2

0.40 0.61 0.58

0.53 0.71 0.68 0.64 0.40 0.59

0.41 0.54 0.71 4.41 26.98 27.59 0.82 2.11 13.19 40.78

Loadings less than 0.40 are excluded; Bolded items are included in the scale When the control items were aggregated rather than separated into the highest score, both groups still reported similar levels of the control strategies. For primary control, the academics reported a mean of 79.09, whilst the teachers reported 80.18. For secondary control, the academics reported 46.21, and the teachers reported 46.56. Furthermore, the control strategies were not strongly related to job satisfaction. Primary control was not related at all to job satisfaction, whilst secondary control was slightly negatively related to job satisfaction (r = -0.24). These descriptive statistics suggest that even if the scales were aggregated, the results would still not be significant.

206

As the scoring procedure does not appear to have drastically altered the results, the hypotheses will be tested using the intended scoring procedure (i.e., highest number). This scoring method, although problematic because of its small range, is theoretically superior to the aggregated measure. However, preliminary analyses using this scoring method clearly demonstrate that primary and secondary control strategies are not behaving as expected, and as such, many of the hypotheses will not be supported. In order to reduce the repetitiveness of these findings, the hypotheses examining primary and secondary control will be examined collectively. Specifically, this refers to hypotheses 2, 3, 4, 5, and 6.

207

2.7

Hypothesis Testing

In order to test the proposed model of job satisfaction, multivariate analyses of variance were conducted to compare the levels of control strategies, job autonomy, job satisfaction and life satisfaction reported by the academics and the teachers. Multiple regression analyses were also conducted to examine the major predictors of job satisfaction, and the moderating role of need for autonomy and social support at work. As in study one, the alpha level was reduced to 0.01 in order to reduce the risk of Type I errors. 2.7.1 Hypothesis One: Levels of Job Autonomy and Job Satisfaction In order to test the first part of hypothesis one, proposing that job autonomy is positively related to job satisfaction, the correlation coefficients for each occupational group were examined. Consistently, job autonomy was positively related to job satisfaction for both the academics (r = 0.37) and the teachers (r = 0.51). In order to test the second part of hypothesis one, proposing that the academics would report higher job autonomy than the teachers, an analysis of variance was employed. The assumption of univariate homogeneity of variance, using Levenes test, was met, F (1, 203)= 2.58, p > 0.05. The univariate test of significance demonstrated that, as hypothesised, the academics (M = 75.13, SD = 16.12) reported significantly higher job autonomy than the teachers, (M = 66.34, SD = 17.74), F (1, 203) = 13.82, p = 0.00.

208

2.7.2

Hypotheses Two and Three: Examining how Job Autonomy Influences the Amount of Primary and Secondary Control Strategies Hypotheses two and three examine how job autonomy influences the amount

of control strategies that employees use. In order to test hypothesis two, that the academics would report more primary control and less secondary control than the teachers, a multivariate analysis of variance was employed. The variables were normally distributed, and reasonably linear relationships were evident. There was no evidence of multicollinearity as the determinant of the within-cell correlation > 0.0001 (i.e., 0.774). Univariate homogeneity of variance, as assessed through Levenes test was met for primary control, F (1, 203) = 0.50, p > 0.05, and secondary control, F (1, 203) = 1.68, p > 0.05. The assumption of multivariate homogeneity of variance was also met using Boxs M test. The multivariate test using Pillais criterion was not significant, F (2, 202) = 0.09, p = 0.91. Inconsistent with hypothesis two, the academics (M = 81.58, SD = 14.42) reported similar level of primary control as the teachers (M = 80.64, SD = 15.81), F (1, 203) = 0.20, p = 0.66. Furthermore, the academics (M = 83.33, SD = 13.37) reported similar levels of secondary control as the teachers (M = 82.93, SD = 11.90), F (1, 203) = 0.05, p = 0.82. In order to test hypothesis three that job autonomy is positively related to primary control and negatively related to secondary control, job autonomy was correlated with primary and secondary control. This analysis demonstrated that, inconsistent with hypothesis three, primary control (r = 0.09) and secondary control (r = 0.03) were not significantly related to job autonomy.

209

2.7.3

Hypotheses Four and Five: Examining how Job Autonomy Influences the Relationship Between the Control Strategies and Job Satisfaction Hypothesis four and five test the proposal that job autonomy influences the

relationship between job autonomy and job satisfaction. In order to test hypothesis four, proposing that primary control will be more positively related to job satisfaction than secondary control for the academics, and secondary control will be more positively related to job satisfaction than primary control for the teachers, two standard multiple regression analyses were conducted. The assumptions of normality, linearity, and homoscedasticity of residuals, as assessed through examination of the residual scatterplots, were met for both groups. As demonstrated in Table 26, R was not significantly different from zero for the academics, R = 0.08, F (2, 105) = 0.32, p = 0.73, or for the teachers, R = 0.16, F (2, 94) = 1.16, p = 0.32. Inconsistent with hypothesis four, primary and secondary control were not related to job satisfaction for either group.

210

Table 26- Multiple Regression of Primary and Secondary Control on Job Satisfaction for Academics and Teachers Group Acad IV PC SC Teach PC SC 0.14 0.14 0.56 0.12 0.15 R =0.16 0.09 0.09 R2=0.02 0.55 0.50 AdjR2=-0.003 JS -0.07 -0.06 PC 0.40 B -0.08 -0.07 R =0.08 -0.05 -0.04 R2=0.006 sr2 (unique) 0.22 0.14 AdjR2=-0.013

Acad Academics; Teach Teachers; PC - Primary control; SC Secondary control; JS Job satisfaction For the academics, R is composed of 0.36% unique variance and 99.64% shared variance. For the teachers, R is composed of 1.05% unique variance and 98.95% shared variance. Hypothesis five, similar to hypothesis four, examines how job autonomy influences the adaptiveness of the control strategies, however it is based on the measured level of job autonomy rather than the presumed level. Hypothesis five proposes that the relationship between the control strategies and job satisfaction is moderated by job autonomy. As discussed in study one, job autonomy is a moderator rather than a mediator as it specifies when certain effects will hold. That is, when job autonomy is high, primary control will be more strongly correlated with job satisfaction, and when job autonomy is low, secondary control will be more strongly correlated with job satisfaction. In order to test the moderating effect of job autonomy on primary control and secondary control, two hierarchical multiple regression analyses were conducted. In

211

these analyses the control strategy was entered first, then job autonomy, and then the interaction term. The assumptions of normality, linearity, and homoscedasticity of residuals, as assessed through examination of the residual scatterplots were met. As demonstrated in Table 27, for primary control, R was not significantly different from zero after the first step (i.e., primary control), R = 0.03, F (1, 203)= 0.15, p = 0.70. The addition of job autonomy did result in an increase in R, where R = 0.39, Finc (1, 202) = 36.29, p = 0.00. However, the addition of the interaction term was not significant, R = 0.39, Finc (1, 201) = 0.12, p = 0.73. For secondary control, R was not significantly different from zero after the first step, R = 0.02, F (1, 203)= 0.09, p = 0.77. After job autonomy was entered, the value of R increased, R = 0.39, F (1, 202) = 36.35, p = 0.00, however the addition of the interaction term in step three was not significant, R = 0.39, F (1, 201) = 0.21, p = 0.65. Thus, hypothesis five was not supported.

212

Table 27- Hierarchical Multiple Regression testing the Moderating role of the Control Strategies on the Relationship Between Job Autonomy and Job Satisfaction Step IV DV B sr
2

( u ni q u e) 1 2 3 Primary control Primary control Job Autonomy Primary control Job autonomy Primary control x job autonomy JS 0.04 -0.01 0.48 0.12 0.62 -0.001 R =0.39 Step 1 2 3 IV Secondary control Secondary control Job Autonomy Secondary control Job autonomy Secondary control x job autonomy DV JS 0.03 0.01 0.47 0.23 0.72 -0.002 R =0.39 *p<0.05, ** p<0.01; JS Job satisfaction 0.03 -0.009 0.39 0.08 0.51 -0.16 R2=0.15 0.02 0.009 0.39 0.14 0.59 -0.24 R2=0.15 15.21**

AdjR2=0.14 sr2 (unique) 15.21**

AdjR2=0.14

213

2.7.4

Hypothesis Six: Examining the Proposed Explanation for the Relationship Between Job Autonomy and Job Satisfaction In order to test hypothesis six, which proposes that the relationship between job

autonomy and job satisfaction is mediated by the control strategies, a hierarchical multiple regression analysis was conducted. The assumptions of normality, linearity and homoscedasticity of residuals were met, and there was no evidence of multicollinearity. As demonstrated in Table 28, R was not significantly different from zero after primary and secondary control were entered, R = 0.03, F (2, 202) = 0.08, p = 0.92. R did significantly increase after job autonomy was added to the equation, R = 0.39, Finc (3, 201)= 36.16, p = 0.00. Only job autonomy predicted job satisfaction accounting for 15% of the variance in job satisfaction. As such, when primary and secondary control were controlled for, job autonomy still predicted job satisfaction. Inconsistent with hypothesis six, primary and secondary control did not mediate the relationship between job autonomy and job satisfaction. Table 28- Hierarchical Multiple Regression Testing the Mediating Role of the Control Strategies Step 1 IV Primary control Secondary control DV JS B 0.03 0.02 R =0.03 2 Primary control Secondary control Job autonomy JS -0.03 0.03 0.48 0.02 0.01 R2=0.001 -0.02 0.02 0.39 sr2(unique)

AdjR2=-0.009

15.21**

214

R =0.39** **p>0.01; JS Job satisfaction

R2=0.15

AdjR2=0.14

2.7.5

Hypothesis Seven: Occupational Differences in Job Satisfaction and Life Satisfaction In order to test hypothesis seven, that academics would report higher job

satisfaction than the teachers, an analysis of variance was conducted on the one-item measure of job satisfaction. Years working in occupation and age were entered as covariates in this analysis as the academics tended to be older, and had worked less years than the teachers (i.e., refer to Table 20). Life satisfaction was normally distributed, and reasonably linear relationships were evident. Univariate homogeneity of variance, as assessed through Levenes test was met, F (1, 203)= 0.33, p > 0.05. The univariate test demonstrated that the academics (M = 64.09, SD = 21.72) reported similar levels of job satisfaction as the teachers (M = 68.79, SD = 20.23), F (1, 201) = 3.39, p = 0.07. Although the one-item measure of job satisfaction is the dependent variable in this study, the facet measure of job satisfaction was also explored to gain a greater understanding of the two occupational groups. A multivariate analysis of variance was conducted on the intrinsic and extrinsic facets of job satisfaction. The variables were normally distributed, and reasonably linear relationships were evident. There was no evidence of multicollinearity as the determinant of the within-cell correlation > 0.0001 (i.e., 0.526). Univariate homogeneity of variance, as assessed through Levenes test was met for intrinsic job satisfaction, F (1, 203) = 1.63, p > 0.05, and extrinsic job

215

satisfaction, F (1, 203)= 0.03, p > 0.05. The assumption of multivariate homogeneity of variance was also met using Boxs M test. The multivariate test, using Pillais criterion was significant, F (1, 199) = 5.96, p = 0.00. Examination of the univariate tests demonstrated that the teachers (M = 56.15, SD = 20.89) reported higher extrinsic job satisfaction than the academics (M = 44.46, SD = 20.45), F (1, 203)= 16.35, p = 0.00. The teachers (M = 77.07, SD = 14.80) also reported higher intrinsic job satisfaction than the academics (M = 72.76, SD = 13.24), F (1, 203)= 4.84, p = 0.03, however this finding was not significant as the more stringent alpha level of 0.01. The means and standard deviation for the intrinsic and extrinsic items for academics and teachers are provided in Table 29.

216

Table 29- Means and Standard Deviations of the Intrinsic and Extrinsic Job Satisfaction Items for Academics and Teachers Item Being able to keep busy all the time The chance to work alone on the job The chance to do different things from time to time The change to be somebody in the community The way my boss handles his/her work The competence of my supervisor in making decisions Being able to do things that dont go against my conscience The way my job provides for steady employment The chance to do things for other people The chance to tell people what to do The chance to do something that makes use of my abilities The way company policies are put into practice My pay and the amount of work that I do The chance for advancement on the job The freedom to use my own judgement The chance to try my own methods of doing the job The praise I get for doing a good job The feeling of accomplishment I get from the job Academics M SD 71.70 25.35 80.45 16.49 77.77 21.69 65.74 47.32 51.95 68.72 77.47 77.77 59.16 77.05 30.34 49.59 43.41 74.18 73.25 44.14 69.86 24.87 30.62 30.60 25.01 25.21 18.10 22.52 20.05 24.28 26.46 28.64 20.42 20.50 30.96 22.73 Teachers M SD 81.21 19.60 73.88 22.51 78.24 21.75 69.64 67.35 70.10 78.92 87.51 87.05 67.12 80.06 48.68 45.02 53.15 72.62 76.51 52.57 72.05 21.12 25.85 24.23 21.90 21.35 14.85 20.34 20.15 24.91 28.79 29.88 22.22 21.51 28.31 21.76

Bolded items measure extrinsic job satisfaction. Non-bolded items measure intrinsic job satisfaction. In order to test the second part of hypothesis seven, that the academics would report higher life satisfaction than the teachers, a univariate analysis of variance was employed. Life satisfaction was normally distributed using the skewness/standard

217

error < 3 criterion. The assumption of univariate homogeneity of variance, as assessed by Levenes test, was not met, F (1, 203) = 0.96, p < 0.05, and as such, this analysis proceeded with caution using an alpha level of 0.05. The univariate test of significance demonstrated that, inconsistent with hypothesis seven, there were no occupational differences in levels of life satisfaction, F (1, 203) = 0.51, p = 0.11. As demonstrated in Table 30, the teachers average level of life satisfaction was 74.21 (SD = 11.34) and the academics average level was 75.61 (SD = 14.48). The means and standard deviations for the seven domains of life satisfaction are also presented to demonstrate that the two groups appear to be more satisfied with safety, intimacy and material well-being, and less satisfied with health and community. Table 30- Means and Standard Deviations of the Domains of Life Satisfaction for Academics and Teachers Domain Material Satisfaction Health Satisfaction Productivity Satisfaction Intimacy Satisfaction Safety Satisfaction Community Satisfaction Emotional Satisfaction Overall life satisfaction Academics M SD 76.87 17.91 68.73 21.37 75.03 14.58 76.77 19.98 81.80 16.87 70.24 18.72 72.91 18.79 74.35 11.16 Teachers M SD 76.08 16.92 67.35 23.22 73.03 18.84 78.97 20.33 84.56 16.67 72.98 20.57 76.59 19.62 75.64 14.45

218

2.7.6

Hypothesis Eight: Examining how Social Support at Work Moderates the Relationship between Difficulties at Work and Job Satisfaction Hypothesis eight proposes that social support at work moderates the effect of

work difficulties on job satisfaction. Social support at work is proposed to be a moderator, that is, a variable that affects the direction and/or strength of the relationship between an independent variable (i.e., work difficulties) and a dependent variable (i.e., job satisfaction). It is a moderator rather than a mediator because it affects the relationship between work difficulties and job satisfaction, however it does not explain why work difficulties and job satisfaction are related. In order to test this moderation effect, a hierarchical multiple regression is required. In the first step the independent variable is entered (i.e., work difficulties). In the second step the moderator variable (i.e., social support) is entered. Finally, in the third step the interaction term is entered (i.e., independent variable multiplied by the moderator variable). Moderator effects are evident if the interaction term predicts the dependent variable after the independent variable and the moderator variables have been entered in steps one and two. Two hierarchical multiple regression analyses were conducted for supervisor support, and co-worker support. For both analyses, the assumptions of normality, linearity, and homoscedasticity of residuals were met, and there was no evidence of multicollinearity. The moderating role of supervisor support on the relationship between work difficulties and job satisfaction was tested first. R was significantly different from zero at the end of the first step (i.e., work difficulties), R = 0.26, F (1, 203) = 14.94,

219

p = 0.00. The addition of supervisor support resulted in a significant increment in R2, where R = 0.57, Finc (1, 203) = 78.86, p = 0.00. Supervisor support accounted for 26% of the variance in job satisfaction. As demonstrated in Table 31, the addition of the interaction term (difficulties x supervisor support) did result in a significant increment in R2, where R =0.59, Finc (1, 201) = 4.29, p = 0.04. It must be noted that this finding was not significant using the more stringent alpha level of 0.01. Table 31- Hierarchical Multiple Regression Analysis Examining if Supervisor Support Moderates the Relationship between Work Difficulties and Job Satisfaction Step 1 IV Freq. of Diff DV JS B -0.23 R =0.26** 2 Freq. of Diff Supervisor JS -0.18** 0.37** R =0.57** 3 Freq. of Diff Supervisor Diff x Sup JS -0.36 0.16 0.04 R =0.59* -0.26 R2=0.07 -0.20 0.52 R2=0.33 -0.40 0.22 0.36 R2=0.34 sr2 (unique) 6.86** Adj R2=0.06 3.84 26.11 Adj R2=0.32 4.04** 1.39* Adj R2=0.33

p<0.05 , ** p>0.01; JS Job satisfaction Although not significant at 0.01, the interaction between difficulties and supervisor support will be examined further for two reasons. First, only a few studies have examined the moderating role of social support at work, and as such, the pattern of the interaction requires investigation. Second, it is difficult to achieve statistical significance in moderation analyses as the power is low (Bobko, 2001). By having the

220

independent variable, the moderator and the interaction term (independent variable x moderator), there is an increased chance of multicollinearity (Bobko, 2001). As the correlation between predictors increases, the standard deviation of the regression weights increases, and it becomes less likely that the null hypothesis will be rejected. In order to reach significance therefore, the analysis needs to have large effects of large sample sizes (Bobko, 2001). Thus, as the analysis was significant at 0.05, it will be examined further. Work difficulties were regressed on job satisfaction separately for those with low supervisor support, and those with high supervisor support. As proposed by Cohen and Cohen (1983), the low and high distinction was defined as scores that fell one standard deviation below or above the mean for supervisor support. As demonstrated in Figure 7, the regression lines were consistent with the hypothesis, where the slope of the regression line of work difficulties on job satisfaction was steeper for high supervisor support than for low supervisor support.

221

Figure 7 - Relationship Between Work Difficulties and Job Satisfaction for Employees with Low/High Supervisor Support

222

In order to test the moderating role of co-worker support, another hierarchical multiple regression was conducted. R was significantly different from zero at the end of the first step (i.e., work difficulties), R = 0.26, F (1, 202)= 14.91, p = 0.00. The addition of co-worker support resulted in a significant increment in R2, where R = 0.46, Finc (1, 202)= 36.78, p = 0.00. As demonstrated in Table 32, co-worker support accounted for 7% of the variance in job satisfaction. The interaction term (difficulties x co-worker support) did not result in a significant increment in R2, R = 0.46, Finc (1, 202) = 0.06, p = 0.81. As such, co-worker support does not appear to moderate the effect of work difficulties on job satisfaction. Table 32- Hierarchical Regression Analyses examining whether Co-worker Support Moderates the Relationship between Work Difficulties and Job Satisfaction. Step 1 IV Freq. of Diff DV JS B -0.23 R =0.26** 2 Freq. of Diff Co-worker JS -0.18 0.46 R =0.46** 3 Freq. of Diff Co-worker Diff x Co-worker JS -0.13 0.50 -0.008 R =0.46 p<0.05 , ** p>0.01; JS Job satisfaction -0.26 R2=0.07 -0.20 0.38 R2=0.21 -0.14 0.42 -0.07 R2=0.21 sr2 (unique) 6.86** Adj R2=0.06 4.00** 14.36** Adj R2=0.20 0.12 2.25* 0.023 Adj R2=0.20

223

2.7.7

Hypothesis Nine: The Moderating Role of Need for Autonomy on the Relationship Between Job Autonomy and Job Satisfaction In order to test hypothesis nine that need for autonomy moderates the

relationship between job autonomy and job satisfaction, a hierarchical multiple regression analysis was conducted. In this case, need for autonomy is a moderator variable, as it specifies when the relationship between job autonomy and job satisfaction will be strong or weak. The assumptions of normality, linearity, and homoscedasticity of residuals were met, and there was no evidence of multicollinearity. In the first step, job autonomy was entered, followed by need for autonomy in the second step. Finally, the interaction term (i.e., autonomy x need for autonomy) was entered. R was significantly different from zero at the end of the first step, R = 0.39, F (1, 203)= 36.61, p = 0.00. The addition of need for job autonomy did not result in a significant increment in R2, where R = 0.39, Finc (1, 203) = 0.02, p = 0.89. The addition of the interaction term did not result in a significant increment in R2, where R = 0.39, Finc (1, 201) = 0.23, p = 0.63. As demonstrated in Table 33, need for job autonomy does not moderate the relationship between job autonomy and job satisfaction.

224

Table 33- Hierarchical Regression Analyses examining whether Need for Job Autonomy Moderates the Relationship between Job Autonomy and Job Satisfaction. Step 1 IV Job autonomy DV JS B 0.47 R =0.39** 2 Job Autonomy Need for job autonomy JS 0.48 -0.19 R =0.39 3 Job Autonomy Need for autonomy Job Autonomy x Need for autonomy JS 0.78 2.35 -0.03 R =0.39 ** p<0.01; JS Job satisfaction 2.7.8 Hypothesis Ten: Major Predictors of Job Satisfaction In order to test hypothesis ten, which examines several major predictors of job satisfaction, two standard multiple regression analyses were conducted for the academics and the teachers. The following predictors were included: primary and secondary control; job autonomy; personality (neuroticism and extroversion); life satisfaction; social support at work (supervisors and co-workers); and difficulties at work. For both analyses, the assumptions of normality, linearity, and homoscedasticity of residuals were met, and there was no evidence of multicollinearity. 0.39 R2=0.15 0.39 -0.01 R2=0.15 0.64 0.12 -0.31 R2=0.15 Adj R2=0.14 sr2(unique) 15.29** Adj R2=0.15 14.51** Adj R2=0.15

225

R was significantly different from zero for both the academics, R = 0.65, F (9, 98) = 7.92, p = 0.00, and the teachers, R = 0.75, F (9, 87) = 12.49, p = 0.00. The major predictors of job satisfaction were the same for both occupational groups, namely job autonomy and supervisor support at work. As demonstrated in Table 34, job autonomy accounted for approximately 2% and 5% of the variance in job satisfaction for the academics and the teachers respectively. The finding for the academics must be examined cautiously however, as it was not significant at the more stringent alpha level of 0.01. Supervisor support at work accounted for 5% and 13% of the variance in job satisfaction for the academics and teachers respectively.

226

Table 34- Standard multiple Regression Predicting Job Satisfaction for Employees with Low Autonomy and Employees with High Autonomy Group Acad Variable Primary Control Secondary Control Job Autonomy Life Satisfaction Neuroticism Extroversion Co-worker Support Supervisor Support Difficulties at Work Teach Primary Control Secondary Control Job Autonomy Life Satisfaction Neuroticism Extroversion Co-Worker Support Supervisor Support Difficulties at Work 0.03 0.19 0.28 0.15 -0.14 -0.01 0.04 0.36 -0.11 R =0.75** p<0.05 , ** p>0.01; Acad Academics; Teach- Teachers For academics, R is composed of 14.46% unique variance and 85.54% shared variance. For teachers, R is composed of 21.97% unique variance and 78.03% shared variance 0.02 0.11 0.25 0.11 -0.11 -0.008 0.03 0.47 -0.12 R2=0.56 4.70** B -0.13 -0.01 0.23 0.29 -0.19 0.02 0.20 0.21 -0.10 R =0.65** -0.09 -0.01 0.17 0.15 -0.15 0.01 0.18 0.28 -0.16 R2=0.42 sr2 (unique)

2.28*

5.61** AdjR2=0.37

13.54** AdjR2=0.52

As supervisor support appeared to account for the largest proportion of the variance in job satisfaction for both occupational groups, a further regression analysis was conducted to examine the value of R with only supervisor support. R was significantly different from zero for the academics, R = 0.46, F (1, 106) = 27.81, p =

227

0.00, and the teachers, R = 0.64, F (1, 95) = 67.29, p = 0.00. Supervisor support at work accounted for 21% and 41% of the variance in job satisfaction for the academics and the teachers respectively. 2.7.9 Conclusion This study tested whether job autonomy influenced the use and adaptiveness of primary and secondary control strategies. Inconsistent with the hypotheses, the teachers and academics reported similar levels of primary and secondary control. Furthermore, primary and secondary control strategies were not correlated with job satisfaction. Although the control strategies did not mediate the relationship between job autonomy and job satisfaction, the findings highlighted the importance of supervisor support in predicting job satisfaction. These findings will now be discussed.

228

S E M

=
2.8

Discussion

1 rx x

This study was designed to extend the job demand-control model (Karasek & Theorell, 1990), testing an explanation for the positive relationship between job autonomy and job satisfaction. This explanation proposed that job autonomy influences the use and adaptiveness of the control strategies. Employees who reported higher job autonomy were expected to successfully implement primary control. The hypotheses were generally not supported however, and as such, a thorough review of the assumptions and the hypotheses is required. 2.8.1 Assumption- The Academics Represent a High Job Autonomy Group and the Teachers Represent a Low Job Autonomy Group The major assumption underlying this study is that the academics were expected to report higher job autonomy than the teachers as they have more freedom and choice in many aspects of their work. Consistent with this expectation, the academics (M = 75%SM) reported significantly higher job autonomy than the teachers (M = 66%SM). Although this difference was significant, it must be demonstrated that the difference is meaningful. One way to determine if the difference is meaningful is to calculate the standard error of measurement (SEM; Wyrwich, Nienaber, Tierney & Wolinsky, 1999). In the past, researchers have used the SEM to determine clinically meaningful standards. It is estimated by multiplying the standard deviation of the scale by the square root of one minus the reliability coefficient, or Although there is no consensus about how many SEMs an individuals score must change for it to be considered significant, Wyrwich et al., (1999) suggest that a

229

2.77 SEM criterion is the safest. In their study however, they demonstrated that a one SEM criterion reflected a minimal clinically important difference. In the current study, the SEM is estimated to be 7.18. Thus, on average there appears to be one SEM difference between the levels of job autonomy reported by academics and teachers. This may not necessarily be meaningful as Wyrwich et al (1999) stress that their results should not be generalised to other populations or tests. Another way to examine if the difference is meaningful is to compare the current levels of job autonomy with that reported by other occupational groups (Hackman & Oldham, 1980). As demonstrated in Table 35, the means range from 58%SM to 73%SM. Compared to these occupational groups, the academics are in the higher range and the teachers are in the lower range. However, it must be noted that these data are relatively old. More recent studies have administered Hackman and Oldhams (1975) scale to different occupational groups, however they do not separate the occupational groups (e.g., Renn & Vandenberg, 1995; Tiegs et al., 1992).

230

Table 35- Normative Data for Hackman and Oldhams (1980) Autonomy Scale Occupation Professional Management Clerical Sales Service Processing Machine Trades Bench Works Structural Works Normative Data (%SM) 73.33 73.33 58.33 63.33 66.66 58.33 65.00 60.00 66.66

Although past studies do not shed much light on whether the differences in levels of job autonomy reported by the academics and teachers are meaningful, it is clear that the academics are reporting significantly higher job autonomy than the teachers. These two groups may not represent the extremes of job autonomy, however the difference should be great enough to demonstrate the expected differences in the control strategies. It is assumed that the use of the control strategies varies linearly with job autonomy over this range of values. As such, even if the academics and teachers do not represent extremes of job autonomy, the expected findings should be evident, albeit weaker. In summary, it appears that the academics report higher job autonomy than the teachers. It is difficult to ascertain whether this difference is meaningful, however it is concluded that the difference should be great enough to demonstrate the expected effects.

231

2.8.2

Hypothesis Testing The major hypotheses tested in the study were that job autonomy influences the

use of primary and secondary control strategies, and also the relationship between the control strategies and job satisfaction (i.e., adaptiveness). These hypotheses were not supported, and as such possible explanations for the findings will be considered, and the methodology will be re-examined. 2.8.3 Job Autonomy Influences the Amount of the Control Strategies It was hypothesised that job autonomy would influence the amount of primary and secondary control strategies. Academics, who have higher job autonomy than the teachers, were expected to report using more primary control and less secondary control. Furthermore, it was expected that job autonomy would be positively correlated with primary control and negatively correlated with secondary control. Inconsistent with these hypotheses however, the two occupational groups reported similar levels of control strategies, and job autonomy was not related to the control strategies. These findings do not support the model of job satisfaction presented in Figure 6. Specifically, the findings do not support the arrow from job autonomy to primary and secondary control. This proposal was based on an extension of the life span theory of control (Heckhausen & Schulz, 1995), which essentially proposes that if a person can control a situation, they will attempt to change it. If they cannot control the situation however, it is more likely that their attempts to change it would fail, and thus they will seek to accept the situation. It is most surprising therefore that the academics

232

reported high secondary control, and that the teachers reported high primary control. Explanations for these unexpected findings will be discussed. 2.8.3.1 Why did the Academics Report High Secondary Control? The academics reported equally high levels of primary and secondary control as the teachers. This means that when they face a difficulty at work, they use both primary and secondary control strategies. It is interesting that they rely on secondary control however, because theoretically, they should have less need for it than the teachers. As they have reasonably high control over their environment, they are expected to successfully implement primary control most of the time and rarely rely on secondary control strategies. As this is clearly not the case, the use of secondary control may need to be re-examined. In addition to using secondary control to compensate for primary control failure, secondary control may be used as a means of temporarily avoiding primary control. If employees were faced with a large number of difficulties at work, they may initially use secondary control. For example, workers may tell themselves that it will work out okay in the end, or that it doesnt matter. This may be necessary for workers, such as academics, who face many difficulties, and must delay dealing with some of them. Once they can deal with them however, it is expected that they do so using primary control. Thus, secondary control may be used prior to primary control. The explanation that secondary control can be used prior to primary control may explain why the academics reported high levels of secondary control. However, this explanation does not account for the lack of occupational differences in secondary

233

control. Even if both groups use secondary control prior to dealing with their difficulties, the teachers would be expected to rely on more secondary control than the academics. After initially delaying dealing with a problem using secondary control, it is expected that the academics would then use primary control, but that the teachers would continue using secondary control. 2.8.3.2 Why did the teachers report high primary control? Although the teachers were expected to rely mostly on secondary control, they reported equally high levels of primary and secondary control. One explanation for this finding is that the teachers may have avoided repeated primary control failure. If they implemented primary control and failed, they were expected to rely mostly on secondary control. Through relying on secondary control, and accepting their situation rather than trying to change it however, they can then maintain their perceptions of primary control. Thus, the teachers levels of primary control may be explained, in part, by their reliance on secondary control strategies. 2.8.3.3 Summary Both the academics and the teachers reported similar levels of primary and secondary control. The academics reported higher secondary control than expected, and the teachers reported higher primary control than expected. The academics level of secondary control may be explained by the proposal that secondary control may also be used prior to primary control. The teachers levels of primary control may be explained by the proposal that they can avoid primary control failure through relying on secondary control strategies.

234

2.8.4

Job Autonomy Influences the Relationship Between the Control Strategies and Job Satisfaction It was hypothesised that job autonomy influences the adaptiveness of the

control strategies, such that primary control would be more positively related to job satisfaction than secondary control for the academics, and that secondary control would be more positively related to job satisfaction than primary control for the teachers. Furthermore, it was hypothesised that the relationship between the control strategies and job satisfaction would be moderated by job autonomy. Inconsistent with both of these hypotheses however, primary and secondary control strategies were not related to job satisfaction. This suggests that workers can use primary or secondary control strategies to deal with their difficulties. These findings are inconsistent with the primacy/back-up model and the discrimination model (Thompson et al., 1998). The discrimination model proposes that primary control is the most adaptive strategy in controllable situations, and that secondary control is the most adaptive strategy in uncontrollable situations. The primacy/back-up model proposes that primary control is the most adaptive strategy in low-control and high-control situations. The current results demonstrated that primary and secondary control strategies were not related to job satisfaction. However, a major limitation has now been identified in the study that may render these findings invalid.

235

2.8.5

Limitations in the Hypotheses Examining Job Autonomy and Control Strategies There was a methodological problem in this study that may have limited the

findings examining how job autonomy influences the use and the adaptiveness of the control strategies. This problem concerns the specificity of the hypotheses. The hypotheses tested whether job autonomy influenced the relationship between the control strategies and job satisfaction at a general level, however the primacy/back-up model and the discrimination model actually only refer to one situation. These models propose that the controllability of a situation influences the control strategies used to handle that situation. Thus, the hypotheses need to be measured at a situational level, rather than at an occupational level. This lack of consistency between the definition of the discrimination model and the testing of the discrimination model is not limited to this study. All of the studies that Thompson et al., (1998) claimed to test the discrimination model actually fail to test it as specified in the definition. To test the discrimination model, researchers need to correlate a measure of perceived control over one situation with the control strategies used in that situation. Past researchers have relied on specific measures of the controllability of the situation, and the control strategies, however they aggregated them rather than examining them separately. For example, Thomson et al., (1996, 1994) developed a list of specific facets relevant to living with HIV, such as progression of HIV infection, family relationships, and quality of medical care. For each of these facets, respondents rated how much

236

control they had over them, and the extent to which they used primary and secondary control to handle them. If they correlated each facet with the control strategies used to handle that facet, they would be testing the discrimination model. However, they aggregated the items to obtain an overall measure of perceived control, an overall measure of primary control and an overall measure of secondary control. Another study measured specific controllability and control strategies, yet failed to use this information to test the discrimination model. Thompson et al., (1998) examined control over physical appearance, measuring how much control people had over the attractiveness of their hair, body strength and agility, weight and body shape, skin and overall physical appearance. They also measured the primary and secondary control strategies in relation to age-related changes over physical appearance. Although they measured these specific variables however, they added the perceived control scale to the primary control items to measure primary control. Thus, they failed to examine whether the controllability of a situation influenced the control strategies used in that situation. It thus appears as though the current study and previous studies have failed to adequately test the discrimination model. In order to do so, future studies need to examine the controllability of the situation and the control strategies at a situational level. It should then be tested whether the amount of autonomy an employee has over a situation predicts the use and the adaptiveness of the control strategies in that situation. As such, it is not necessary to examine the control strategies that employees with high/low job autonomy are using, rather to examine which control strategies all workers use in low-control and high-control situations.

237

2.8.5.1

Summary Job autonomy did not influence the use or adaptiveness of the control

strategies. These findings were inconsistent with the proposed model of job satisfaction, which attempted to explain the relationship between job autonomy and job satisfaction. However, a major limitation was identified in the study, where it appears as though the study has failed to test the discrimination model. 2.8.6 Other Predictors of Job Satisfaction The remainder of the hypotheses will now be examined. These hypotheses examine occupational differences in job and life satisfaction, the buffering role of social support at work, and the moderating role of need for autonomy. Additionally, the major predictors of job satisfaction that are included in the proposed model of job satisfaction are examined. 2.8.7 Occupational Differences in Job Satisfaction and Life Satisfaction Inconsistently, the academics did not report higher job satisfaction or life satisfaction than the teachers. The finding that there was no difference in life satisfaction is not surprising given that there were no occupational differences in job satisfaction. Furthermore, the levels of life satisfaction reported by both groups were within the normative range according to the homeostatic theory of life satisfaction (Cummins, 1995, 2000b). It is surprising however that the academics (M = 64.09) and the teachers (M = 68.79) reported similar levels of job satisfaction. The levels of job satisfaction reported by both groups will firstly be compared with past studies.

238

2.8.7.1

Comparisons with Past Studies Past research has reported varying levels of job satisfaction for academics.

Researchers have reported the following levels of job satisfaction; 57%SM (Leung et al., 2000), 65%SM (Hill, 1986), 66%SM (Lahey & Vihtelic, 2000), 74%SM (Carson et al., 2001), 82%SM (Olsen, 1993) and 83%SM (Niemann & Dovidio, 1998). Furthermore, study one reported a level of 66%SM. The scales used in some of these studies were criticised in study one, however a normative level of job satisfaction was not established. All that can be concluded is that the academics in this study, as with those in study one, report a level of job satisfaction that is within the range found by other researchers. In regards to teachers, researchers have reported several different levels of job satisfaction. For example, Klecker and Loadman (1999) found a similar level of job satisfaction (M = 68%SM) as the present study, however others have reported a higher level of job satisfaction of 80%SM (Ma & Macmillan, 1999; Schonfeld, 2000). All of these studies are flawed however as they relied on poor measures of job satisfaction. For example, Ma and MacMillan (1999) included the following items to measure job satisfaction; I find my professional role satisfying, I look forward to each day, I am committed to making our school one of the best in the province and If I could start over, I would become a teacher again. The item I look forward to each day may actually be dependent on personality and quality of life, as well as job satisfaction. Furthermore, the item If I could start over, I would become a teacher again, is likely to be dependent on the teachers age and how much they have invested into becoming a teacher. A teacher who has spent 20 years teaching may

239

agree with this item because they are satisfied with their job, or because they want to justify why they are still in the profession. Until psychometric data are obtained for this scale, the results are questionable. Other researchers have relied on facet scales of job satisfaction. For example, Klecker and Loadman (1999) measured satisfaction with salary, professional advancement, professional challenge, autonomy, work conditions, interactions with colleagues, and interactions with students. Facet scales are criticised however for excluding facets that are important to the individual, or including facets that are not important. Furthermore, it is expected that the teachers level of job satisfaction would be lower in this scale, as it is dependent on their level of job autonomy. This is problematic because although job autonomy is expected to be related to job satisfaction, job autonomy may not be a domain of job satisfaction. In general, it appears that the academics level of job satisfaction is reasonably consistent with past studies. The teachers level of job satisfaction tends to be slightly lower than previous studies, however these studies have relied on inadequate measures of job satisfaction. As few studies have examined academics and teachers levels of job satisfaction, the findings can also be compared to the review conducted in study one, which demonstrated an average level of job satisfaction of 66%SM. This average level is consistent with both the academics and the teachers. Possible explanation for the similar levels of job satisfaction will be presented.

240

2.8.7.2

Explaining the Similar Levels of Job Satisfaction There are a number of explanations for the academics and teachers similar

levels of job satisfaction. First, the recruitment process was different for the teachers and the academics. The academics were sent their questionnaire through internal mail, whereas the teachers were, in some cases, given their questionnaire by the Principal of the school. This may be problematic for the teachers, as the Principal, in order to obtain positive results, may have given the questionnaires to happier workers. Alternatively, even if social desirability was not important, the Principals may have given the questionnaires to teachers that were more likely to agree, or were more organised. Thus, it must be questioned whether the teachers included in the sample are representative of the average teacher. It is possible however that the teachers average level of job autonomy is representative of the average teacher. Although teachers report a lower level of job autonomy than the academics, there are certainly many other determinants of job satisfaction. One major factor that was highlighted in this research was social support at work. Examination of the descriptive statistics demonstrates that the teachers report higher satisfaction with their supervisor support, and that supervisor support was strongly correlated with job satisfaction. 2.8.8 The Influence of Social Support at Work on the Relationship Between Work Difficulties and Job Satisfaction Supervisor support, but not co-worker support moderated the relationship between work difficulties and job satisfaction. The relationship between work

241

difficulties and job satisfaction was weaker when supervisor support was high. It must be noted however that this hypothesis was only significant at an alpha level of 0.05, and not 0.01, suggesting that the effect may not be large. However, even if the effect is not large, the findings suggest that supervisors should ensure that they provide emotional and instrumental support to their employees. To do this, the supervisor needs to show concern for their employees, and provide tangible assistance (Karasek & Theorell, 1990). The finding that supervisor support plays a greater role than co-worker support has been reported by previous researchers (e.g., Beehr, 1985; Fenlason & Beehr, 1994; Russell, Altmaier & Van Velzen, 1987). Co-workers have less influence at work, and as such may have less influence over difficulties at work (Fenlason & Beehr, 1994). Consistent with the current findings, a few studies have demonstrated that social support at work has positive moderating effects on job satisfaction (i.e., Karasek et al., 1982; Landsbergis et al., 1992). However, there are studies that have failed to demonstrate the moderating role of social support (Chay, 1993; de Jonge & Landeweerd, 1993, cited in Van Der Doef & Maes, 1999; Melamed at al., 1991; Parkes & Von Rabenau, 1993). One difference between these supportive and non-supportive studies is in the measure of social support. Two of the supportive studies (i.e., current study and Landsbergis et al., 1992) relied on Karasek and Theorells (1990) scale, whereas the non-supportive studies relied on several different scales. However, there are too few studies to draw conclusions about the influence of the scales. It is clear that further research is needed to examine the moderating role of social support.

242

2.8.9

The Influence that Need for Job Autonomy has on the Relationship Between Job Autonomy and Job Satisfaction Need for job autonomy did not moderate the relationship between job

autonomy and job satisfaction. Although past empirical studies were equivocal, it seemed intuitive that differences in need for autonomy would influence the relationship between job autonomy and job satisfaction. The non-supportive finding is consistent with a few past studies. For example, de Rijk et al., (1998) failed to demonstrate that need for autonomy moderated the relationship between job autonomy and emotional exhaustion and health complaints. Furthermore, Nicolle (1994) found need for autonomy moderated the relationship between job autonomy and absenteeism in only 3 of 36 analyses. As the current study and previous studies generally fail to demonstrate that need for job autonomy moderates the relationship between job autonomy and job satisfaction/job stress, this hypothesis will no longer be investigated. Too few studies have been conducted to simply conclude the effect does not exist, however it seems that more testing is required to develop a valid measure of need for job autonomy. Studies are relying on exploratory measures, and as such may not be adequately measuring the need for job autonomy construct. As such, it is recommended that researchers firstly attempt to develop a need for job autonomy scale that is psychometrically sound.

243

2.8.10

Major predictors of Job Satisfaction Inconsistent with the proposed model of job satisfaction, only job autonomy

and supervisor support uniquely predicted job satisfaction for both groups. The relationship between these variables and job satisfaction was consistent with past research. For job autonomy, r = 0.43 (Tiegs et al., 1992), and for social support at work, r = 0.52 (Dollard et al., 2000), and r = 0.66 (Munro et al., 1998). The control strategies, personality, life satisfaction, co-worker support, and difficulties at work were moderately correlated with job satisfaction, however they did not uniquely predict job satisfaction. As such, several changes need to be made to the variables included in the model. The control strategies will be retained in the model, however as discussed earlier, several changes will be made to the Primary and Secondary Control Scale. Personality was a poor predictor of job satisfaction, and as such, it will be excluded from study three. Life satisfaction was also a poor predictor of job satisfaction, however it will be retained in the model as it is acting as both an independent variable and a dependent variable. Supervisor support explained much of the variance in job satisfaction, and as such both types of social support (i.e., coworker and supervisor) will be re-examined. Difficulties at work will also be retained in the model, as it is necessary to demonstrate what the primary and secondary control strategies are used for. In summary, as a result of the model of job satisfaction only being partially supported, several changes will be made in study three. These changes will be explained further in chapter 4.

244

2.8.11

Conclusion Although the findings demonstrated that job autonomy did not predict the use

or adaptiveness of the control strategies, one major limitation was identified in this study. The hypotheses were criticised for being too general, and it was suggested that study three should examine job autonomy and the control strategies at the situational level rather than at the occupational level. Furthermore, this study highlighted the importance of supervisor support at work, which will be examined further in study three.

245

3 Chapter 4 - Study Three

246

3.1

Abstract

The major proposal of this study is that the controllability of a work difficulty influences the use and adaptiveness of the control strategies used to handle that difficulty. It was expected, based on the discrimination model, that in controllable situations, employees would use more primary control than secondary control, and that primary control would be the most adaptive. In uncontrollable situations however, it was expected that employees would use more secondary than primary control, and that secondary control would be the most adaptive strategy. These proposals were not supported as employees reported using similar strategies for controllable and uncontrollable difficulties. Furthermore, primary control strategies were more adaptive than secondary control strategies for both types of difficulties. These findings challenge the belief that control strategies are influenced by situational variables and also question the assumption that primary control failure negatively affects job satisfaction. The implications of these findings for the proposed model of job satisfaction are discussed.

247

3.2

Proposal for Study Three

This study continues to test the proposal that job autonomy influences the use and the adaptiveness of primary and secondary control strategies, however unlike previous studies, it will be examined at a situational level, rather than an occupational level. As such, changes are made to the specificity of the hypotheses and the primary and secondary control scale. Further changes are made to the model of job satisfaction, where it is proposed that the control strategies and social support at work moderate the relationship between controllable and uncontrollable work difficulties and job satisfaction. 3.2.1 Specificity of Hypotheses Testing the Proposal that Job Autonomy Influences the Control Strategies In this study, the proposal that job autonomy influences the use and adaptiveness of the control strategies is examined at a more specific level. As discussed in chapter 3, studies one and two were criticised as they were not consistent with the definition of the discrimination model. The discrimination model proposes that when the situation is controllable, primary control is more adaptive, and when the situation is uncontrollable, secondary control is more adaptive. However, empirical tests of the model, including studies one and two, have examined perceived control and control strategies at a general level (i.e., Thompson et al., 1996; Thompson et al., 1994; Thompson et al., 1998). The difference between the definition and empirical tests of the discrimination model may be important. If the discrimination model is tested at a more specific level,

248

the relationship between the two variables may be stronger. All employees, whether they be low autonomy or high autonomy, are expected to have high control over some aspects of their job and less control over other aspects. Furthermore, all employees, whether they be low or high job autonomy, are expected to use primary control in some situations and secondary control in others. By correlating their general level of job autonomy with their general level of control strategies, the results become less extreme, the low and high autonomy groups become more similar, and the correlations become weaker. In order to accurately test the discrimination model, study three will examine whether the controllability of a situation influences the use and adaptiveness of the control strategies used in that situation. Past research examining these proposals will be examined. 3.2.2 Examining how the Controllability of a Difficulty Influences the Use of the Control Strategies As proposed by the life span theory of control, it is expected that all individuals will implement primary and secondary control strategies (Heckhausen & Schulz, 1995). However, the ratio of these strategies is expected to be influenced by the controllability of the situation. If the situation is appraised as being controllable, it is expected that people will try to change it using primary control. The situation is amenable to change, and as such, it is expected that attempts to change the environment using primary control would be successful.

249

If however the situation is appraised as being uncontrollable, it is expected that people will attempt to change themselves using secondary control. If the person tried to change the environment using primary control, they would be likely to experience primary control failure. In order to avoid primary control failure therefore, it is expected that people would rely on more secondary control. Hence, it is proposed that the controllability of the situation is inversely related to the probability of primary control failure, which in turn, influences the use of secondary control strategies. 3.2.3 Empirical Studies Examining if the Controllability of a Situation Influences the Use of Control Strategies Although chapter 1 identified some studies that examined the amount of general primary and secondary control reported by employees to handle general work difficulties, no studies have been located which report the amount of primary and secondary control people use in controllable and uncontrollable situations. One study has examined the control strategies reported by people only in low-control situations (i.e., HIV-positive men in prison). According to the discrimination model, it would be expected that these men would rely on more secondary control than primary control. This was not the case however, as Thompson et al., (1996) demonstrated that the men reported slightly more primary control (M = 48.5%SM) than secondary control (M = 45%SM). Although there are no studies examining the use of the control strategies in controllable and uncontrollable situations, there are some studies that have been conducted on coping strategies. These studies generally examine the amount of

250

problem-focussed and emotion-focussed coping strategies reported by people in controllable and uncontrollable situations (e.g., Bowman & Stern, 1995; Folkman et al., 1986; Forsythe & Compas, 1987; Terry & Hynes,1998; Valentiner, Holahan & Moos, 1994; Vitaliano, DeWolfe, Maurio, Russo & Katon, 1990). One study has specifically examined coping strategies at work. Bowman and Stern (1995) asked nurses to describe two stressful events, one that they could control, could change or could do something about and one that was difficult to control, that you had to accept or had to get used to. Participants then rated the controllability of the situation, and completed Lazarus and Folkmans (1984) coping scale. Unfortunately however, Bowman and Stern (1995) did not examine the mean coping strategies separately for each stressful situation. Instead they aggregated them, providing mean scores for avoidance coping, problem-reappraisal coping, and problem solving coping. It must be noted however that even if the means were provided, the validity of the research design is questioned. There are problem with using the terms change and do something about for controllable situations and accept and get used to for uncontrollable situations. These terms may bias the employees to respond in ways that are consistent with the discrimination model. By their nature, situations that have been changed are those where primary control strategies have been used, and situations that have been accepted are those where secondary control strategies have been used. Other studies have examined coping strategies in controllable and uncontrollable non-work situations. These studies have provided somewhat mixed support. For example, Forsythe and Compas (1987) demonstrated that people used

251

more problem-focussed coping for events appraised as controllable (M = 13.82) than uncontrollable (M = 10.81), however there were no differences in emotion-focussed coping. Furthermore, Folkman et al., (1986) found that married couples tend to use more problem-focussed coping in situations perceived as changeable, and more emotion-focussed coping in situations perceived as having to be accepted. However, Valentiner et al., (1994) demonstrated that college students did not report more problem-focussed type coping (M = 55.39) than emotion-focussed type coping (M = 57.78) in a controllable event. Furthermore, they did not report more emotionfocussed type coping (M = 53.50) than problem-focussed type coping in an uncontrollable event (M = 54.34). Instead of reporting the average level of coping strategies in controllable and uncontrollable situations, other studies have examined the correlations between perceived control and coping strategies. In this case, it would be expected that perceived control would be positively correlated with problem-focussed coping and negatively correlated with emotion-focussed coping. Overall however, these studies have tended to be inconsistent. For example, Osowiecki and Compas (1999) demonstrated that problemfocussed and emotion-focussed coping were not significantly related to perceived control. A similar result was found by Conway and Terry (1992) where problemfocussed coping, self-denigration and escapism did not correlate with the controllability of an event. However, Park, Folkman and Bostrom (2001) demonstrated that controllability appraisal was positively correlated with problem-

252

focussed strategies (i.e., planful problem solving, r = 0.23), and negatively correlated with emotion-focussed strategies (distancing, r = -0.29). When examining these studies on coping strategies, the flaws in the conceptualisation of problem-focussed and emotion-focussed coping must be considered. As discussed in chapter 1, the theory underlying problem-focussed and emotion-focussed coping and the questionnaire designed to assess these strategies (i.e., Ways of Coping Questionnaire; WCQ) has methodological limitations (Edwards & ONeill, 1998). The most concerning problem is that there is overlap among the coping dimensions, where some problem-focussed coping strategies resemble emotion-focussed coping strategies (Edwards & ONeill, 1998). The conceptualisation of primary and secondary control is superior to problemfocussed and emotion-focussed coping because it maintains the distinction between changing the environment (i.e., primary control), and changing the self (i.e., secondary control). Thus, the proposal that the controllability of the situation influences the amount of control strategies will be tested in this study. 3.2.3.1 Summary Based on the proposals of the life span theory of control, it is expected that when employees have a controllable difficulty, they use more primary control than secondary control. When they have an uncontrollable difficulty, it is expected that they will attempt to avoid primary control failure, and thus report more secondary control than primary control. Although no studies have examined the control

253

strategies reported by people in controllable and uncontrollable situations, studies using coping strategies have offered, at best, mixed support. 3.2.4 Examining how Controllability Influences the Adaptiveness of the Control Strategies Based on the discrimination model (Thompson et al., 1998), it is expected that primary control is the most adaptive strategy in controllable situations, and that secondary control is the most adaptive strategy in uncontrollable situations. This is consistent with the life span theory of control which proposes that although primary control is the more adaptive strategy, primary control failure can have negative consequences (Heckhausen et al., 1997). It is postulated that the controllability of the situation is inversely related to the probability of primary control failure. As discussed in chapter 1, only a few studies have examined the most adaptive control strategies in low-control situations (Thompson et al., 1996; 1994; 1993; 1998). These studies suggest that primary control strategies are more adaptive in controllable and uncontrollable situations, however these studies were criticised for their measurement of perceived control, and primary and secondary control strategies. It must be noted however that a similar hypothesis was being developed in the coping literature. Several researchers have tested this proposition, referred to as the goodness of fit hypothesis (Carver, Scheier & Weintraub, 1989; Conway & Terry, 1992; Folkman et al., 1986; Roberts, 1995; Vitaliano et al., 1990). They recognise that it is not the coping response per se that is the key to reduce emotional distress, but

254

rather how well the coping strategy fits the perceived situation (Osowiecki & Compas, 1998, p. 485). Although there is significant overlap among the coping studies and control strategy studies, researchers are yet to integrate the results. As there are few such studies, this integration is essential to gain a greater understanding about whether the controllability of the situation influences the use and adaptiveness of the control/coping strategies in that situation. It must be noted however that the majority of studies have relied on Lazarus and Folkmans (1984) problem-focussed coping and emotion-focussed coping, which was criticised in chapter 1. 3.2.4.1 Integrating Empirical Studies on the Discrimination Model and the Goodness of Fit Model Empirical studies examining whether the controllability of the situation influences the coping/control scales used in that situation provide mixed support. Generally, these studies demonstrate the primary control-type strategies are more adaptive than secondary control-type strategies in controllable situations. However, some of the studies fail to demonstrate that secondary control-type strategies are more adaptive than primary control-type strategies in uncontrollable situations (e.g., Bowman & Stern, 1995; Conway & Terry, 1992; Osowieki & Compas, 1998, 1999; Park, Folkman & Bostrom, 2001; Vitaliano et al., 1990). For example, Thompson et als., (1996) study on HIV-positive men demonstrated that primary control was negatively related to distress and secondary control was positively related to distress. However, other studies have demonstrated

255

that secondary control is more adaptive than primary control. For example, Terry and Hynes (1998) demonstrated that for women coping with in vitro fertilization, problem management coping (i.e., trying to solve the problem) was related to more distress. Secondary control-type strategies (i.e., problem appraisal, and emotional approach) were related to less distress. Integration of the studies testing the goodness of fit model and the discrimination model highlights the inconsistencies in the area. Generally, it appears as though primary control is adaptive in controllable situations, but that secondary control may not be the most adaptive strategy in uncontrollable situations. These results may be limited, as problems have been identified in the design of the studies. 3.2.4.2 Research Design Researchers have typically relied on two major types of designs to measure the goodness of fit hypothesis and the discrimination model. The first design assesses how people handle one stressful situation (e.g., Carver et al., 1989; Conway & Terry, 1992; Folkman et al., 1986; Roberts, 1995; Vitaliano et al., 1990). Typically, the person reports on the most stressful encounter they had during the previous week, indicating how much they could control the situation, and what they did to handle the situation. The researcher then correlates the controllability of the situation with the coping strategies. This measure is problematic however as the respondent chooses whether they report a controllable or uncontrollable situation and the researcher cannot influence this variable.

256

The second type of design, much like studies one and two, examines the control strategies used by people in low-control situations. Researchers have studied various groups such as cancer patients (Osowiecki & Compas, 1998, 1999), HIV- positive men (Thompson et al., 1996; Thompson et al., 1994), people experiencing age-related physical changes (Thompson et al., 1998), and children experiencing homesickness (Thurber & Weisz, 1997). These studies generally assess how much perceived control the person has over the situation (e.g., cancer) and then assesses which control/coping strategies they used to handle the situation (e.g., Osowiecki & Compas, 1998). This design is criticised however as it is inconsistent with the proposed models. Both the discrimination model and the goodness of fit model refer to one situation. In order to test whether the controllability of a situation influences the control strategies used in that situation, the scale needs to be more specific. 3.2.4.3 Summary Research examining the discrimination model and the goodness of fit hypothesis is equivocal. It appears however that the majority of studies find that primary control is adaptive in controllable situations, but less demonstrate that secondary control is adaptive in uncontrollable situations. The validity of these findings are questioned however, as the research designs are criticised. In order to accurately test the discrimination model, a more specific scale is required which assesses how people handle controllable and uncontrollable difficulties.

257

3.2.5

Developing a Situation Specific Primary and Secondary Control Scale In order to test the proposal that the controllability of a situation influences the

use and adaptiveness of the control strategies in that situation, a situation specific primary and secondary control scale is developed. This scale overcomes many of the limitations identified in previous scales, as it: a) assesses how people react in controllable and uncontrollable situations; b) can be used by workers in any occupation; and c) contains few items. The Situation Specific Primary and Secondary Control Scale (Maher & Cummins (2002) is an extension of the 4th edition of the Primary and Secondary Control Scale (Maher et al., 2001). The scale includes four primary control strategies and 12 secondary control strategies. Respondents are asked to indicate how often, during the past week, they have used various strategies when facing a difficulty at work. The major change made to this scale is that rather than thinking about any difficulty at work, the respondents now think about one difficulty that they can control and one difficulty that they cannot control. The major issue in designing a scale to measure low-control and high-control situations was deciding on the wording of the situation. Only one study was located that tested respondents in controllable situation and uncontrollable situations at work (Bowman & Stern, 1995). For the controllable situation, the employee was told to consider a situation that they could control, could change, or do something about. For the uncontrollable situation, the employee was told to consider a situation that was difficult to control, that you had to accept or get used to.

258

Other researchers examining one situation have asked how much control do you have over (Thompson et al., 1994; Thurber & Weisz, 1997), how much influence do you have over (Conway & Terry, 1992) and how much can you change (Carver et al., 1989, Folkman et al., 1986). As mentioned previously, there are problem with using the terms change, do something about, and influence for controllable situations and accept and get used to for uncontrollable situations. These terms may bias the employees to respond in ways that are consistent with the discrimination model. A viable alternative that does not imply that the situation has been changed is control. The employees could be asked to think of a difficulty that they can control and a difficulty that they cannot control. Control is superior to the other constructs, as it does not bias the respondents to nominate primary control strategies in a highcontrol situation. Changes were also made to the primary control items. The primary control items were designed to be general, assessing whether the person looks for different ways to overcome difficulties, persists, puts in effort, and works out how to remove obstacles. Closer examination of these items however revealed several problems. For example, the item, looked for different ways to overcome it may not actually represent primary control. A person who looks for different ways to overcome their difficulties does not necessarily attempt to change the environment to suit their needs. They may think about the different ways, decide that they are all too risky, and then resort to secondary control strategies. To demonstrate primary control, a person needs to do more than just think of different ways to overcome the difficulty, rather they

259

need to act on their environment. As such, this item was changed to a strategy identified by Folkman and Lazarus (1980) termed choose and act on a potential solution. The item, work hard to overcome it is also criticised as it implies that the strategy must be successful for it to represent primary control. According to this item, the person must not only work harder, but must also overcome the difficulty. Primary control does not necessarily involve overcoming the problem, only that the person attempts to change the environment. As such, this item was changed to work harder. Another problematic item is work out how to remove obstacles. This item, adapted from Heckhausen et als., (1997) scale, refers to goals rather than difficulties. This item is appropriate for goals as, if a goal is not obtained, there must be obstacles in the way of it. However, there may not necessarily be obstacles in the way of a difficulty. This item does not appear to fit with the control scale, which focuses on difficulties at work, and as such was deleted from the scale. Based on these analyses, the following three items were included in the primary control scale; choose and act on a potential solution; keep trying; and work harder. One extra item was developed to account for the fact that other people, such as management staff, often control many problems in the workplace. A major way that a person may change difficulties in the workplace is through discussions or confrontations. An item developed in Latacks (1986) coping scale to measure this is discussing the problem with the people involved. As discussing the problem may not necessarily mean that the environment is changed, the item was revised to discuss solutions with the people involved.

260

The addition of this item led to a review of the support item in the secondary control scale to ensure that the two were different. Indeed, it is difficult to separate support as a primary control strategy and support as a secondary control strategy. The main distinction between the two, however, is that support as a primary control strategy involves the person changing the environment, whereas support as a secondary control strategy involves the person changing themselves to accept the problem. The secondary control strategy of support was measured by the item told someone about it in previous editions of the scale. In order to ensure that this item is distinct from the primary control strategy, it was changed to clearly demonstrate that it involves changing the self (i.e., I told someone about the difficulty to make me feel better). One final change to the primary and secondary control scale concerns the scoring. In study two, the highest score for primary and secondary control was recorded. The items were not aggregated because the resulting score was deemed to be unrepresentative of the control strategies. For example, using an aggregated score, a person who reported that they use one secondary control strategy every time (10), and reported never (0) for the remaining strategies would receive a low score. In order to demonstrate that this person is using one secondary control strategy all the time, the highest score for secondary control was recorded (10), and the person received a high score. However, as demonstrated in study two, using the highest score does not appear to be the answer. This method did not differentiate the respondents, with 76% of the subjects reporting a level of primary control between 77%SM and 100%SM, and 84%

261

of the subjects reporting a level of secondary control between 77%SM and 100%SM. This range is concerning, suggesting that there may have been a ceiling effect. As a result, the aggregated scoring procedure implemented in study one will be used in the current study. However, it must be noted that the aggregated scoring method, although used by the majority of researchers in the field, is biased towards people who use a greater variety of strategies. 3.2.5.1 Summary In order to test the proposal that the controllability of a situation influences the use and adaptiveness of the control strategies in that situation, a situation specific primary and secondary control scale was developed. The major change made to the scale is that rather than thinking about any difficulty at work, the employees are now required to think about one difficulty that they can control and one difficulty that they cannot control. Changes were also made to the wording of the primary control items and the scoring procedure. 3.2.6 Examining the Moderating Role of Primary and Secondary Control Strategies In addition to examining the amount and adaptiveness of the control strategies in controllable and uncontrollable situations, this study also tests whether the control strategies moderate the relationship between work difficulties and job satisfaction. Although no other studies have examined the moderating role of the control strategies, several researchers in the coping literature have suggested that it is not the stressor that predicts job satisfaction, but rather how the person deals with the stressor (Aldwin &

262

Revenson, 1987, Ashford, 1988; Parkes, 1990, 1994; Perrewe & Zellars, 1999; Osipow, Doty & Spokane, 1985). For example, Ashford (1988) demonstrated that coping moderated the effect of organisation transitions on job stress, where employees who shared emotions experienced less stress after organisational change. Parkes (1990) also demonstrated that coping moderated the effect of work demands on general health, however they found that employees who reported more direct coping (i.e., problem-focussed coping) had better health. As only a few studies have examined this proposal in the workplace, and as they have relied on varying measures of coping, this proposal will be examined further. It is proposed that primary control strategies are only useful in reducing stress when the situation is controllable. In these situations, primary control can be implemented successfully, and the negative effects of the difficulty can be reduced. When the situation is uncontrollable however, secondary control strategies may be useful in helping the person adjust to the situation. If they use primary control strategies, they are likely to experience primary control failure, which may increase their stress. However, if they use secondary control, they can reduce their stress by accepting the situation. These exploratory proposals will be tested. 3.2.7 Examining the Moderating Role of Social Support at Work In addition to primary and secondary control, social support at work may moderate the relationship between controllable and uncontrollable work difficulties and job satisfaction. In study two, the moderating roles of co-worker and supervisor

263

support were examined. This study will also examine the major types of social support, namely instrumental and emotional support. These two types of social support are expected to play different roles (Ducharme & Martin, 2000; Wong, Cheuk & Rosen, 2000). Instrumental support is expected to buffer work difficulties because it helps workers to cope effectively with problems, whereas emotional support is not expected to buffer work difficulties as it does not directly alter the stressor (Wong et al., 2000). Some support has been provided for these proposals, where instrumental supervisor support, but not emotional supervisor support, has been shown to buffer the effects of job stress on job satisfaction (Wong et al., 2000). A more specific explanation is developed for this study, where it is proposed that both instrumental and emotional support buffer the effects of work difficulties. Specifically, it is proposed that instrumental support buffers the effects of controllable difficulties and emotional support buffers the effects of uncontrollable difficulties. Instrumental support is useful if the difficulty is controllable as other people may help the person to overcome the problem. However, when the difficulty is uncontrollable, instrumental social support may not be useful as there is nothing that can be done to overcome the difficulty. Rather, in these situations, emotional social support may help the person to accept these difficulties. As this study is examining both controllable and uncontrollable difficulties, it must be considered whether emotional and instrumental social support moderate both types of difficulties.

264

3.2.7.1

Summary This study proposes that the control strategies and social support at work

moderate the relationship between work difficulties and job satisfaction. Specifically, it is expected that primary control and instrumental support moderate the relationship between controllable work difficulties and job satisfaction. Furthermore, it is expected that secondary control and emotional support moderate the relationship between uncontrollable work difficulties and job satisfaction.

265

3.3

Revised Model of Job Satisfaction

This study aims to test the model of job satisfaction presented in Figure 8, which is fundamentally different to the previous two models. The model proposes that employees experience controllable and uncontrollable difficulties, which are negatively related to job satisfaction. In response to these difficulties, employees can implement primary and secondary control strategies. It is expected that the ratio of control strategies will vary depending on whether the difficulty is controllable or uncontrollable. For controllable difficulties, it is expected that workers will rely on primary more than secondary control. For uncontrollable difficulties, it is expected that workers will rely on secondary more than primary control. The adaptiveness of the control strategies is also expected to vary depending on whether the difficulty is controllable or uncontrollable. For controllable difficulties, it is expected that primary control will be more adaptive and therefore more positively related to job satisfaction than secondary control. As the situation is controllable, it is likely that a person can change it. For uncontrollable difficulties, secondary control will be more positively related to job satisfaction than primary control. As the situation is uncontrollable, it is unlikely that a person can change the situation using primary control, and thus secondary control would be more adaptive than primary control failure. Both types of difficulties (controllable and uncontrollable) are expected to be directly and indirectly related to job satisfaction. Employees who report more difficulties at work are expected to report lower job satisfaction. However two

266

variables that may moderate the relationship between work difficulties and job satisfaction are primary and secondary control and social support at work. These moderation effects are represented by the interaction terms in Figure 8. In regard to primary and secondary control, it is proposed that primary control strategies are useful in reducing the effects of work difficulties when the situation is controllable. When the situation is uncontrollable, secondary control strategies may reduce the effects of work difficulties on job satisfaction. In regard to social support at work, it is expected that instrumental support will buffer the effects of controllable difficulties on job satisfaction. Emotional support is expected to buffer the effects of uncontrollable difficulties on job satisfaction. In addition to work difficulties, job autonomy and life satisfaction are expected to directly predict job satisfaction. Both of these relationships have been demonstrated in studies one and two. In summary, controllable and uncontrollable work difficulties, the primary and secondary control strategies used to handle such difficulties, and social support at work, are expected to determine job satisfaction. Job satisfaction is, in turn, expected to influence, and be influenced by, life satisfaction.

267

Figure 8- Revised Model of Job Satisfaction

Secondary Control Instrumental Support

Primary Control

Controllable Diff x Primary Control

Controllable Difficulties

Control Diff x Instrumental Support

Job
Life Satisfaction

Satisfaction

Job Autonomy

Uncontrollable Diff x Secondary Control

Uncontrollable Difficulties

Uncontrol Diff x Emotional Support

Secondary Control Emotional Support

268
Primary Control

269

3.4

Hypotheses

1) When workers face controllable difficulties, they are expected to use primary control more than secondary control. Conversely, when workers face uncontrollable difficulties, they are expected to use secondary control more than primary control. Workers should match their control strategies to the controllability of the situation. When the situation is controllable, the workers will be likely to change it using primary control. When the situation is uncontrollable, the workers will be likely to accept the situation using secondary control.

2) When workers face controllable difficulties, primary control is expected to be more positively related to job satisfaction than secondary control. Conversely, when workers face uncontrollable difficulties, secondary control is expected to be more positively related to job satisfaction than primary control. This hypothesis is based on the discrimination model, which proposes that primary control is more adaptive in controllable situations, and that secondary control is more adaptive in uncontrollable situations.

3) Primary control will moderate the effect of controllable difficulties on job satisfaction and secondary control will moderate the effect of uncontrollable difficulties on job satisfaction. This hypothesis proposes that primary control strategies are useful in reducing the influence of work difficulties on job satisfaction when the situation is controllable.

270

When the situation is uncontrollable however, secondary control strategies are expected to reduce the effects of work difficulties on job satisfaction.

4) Instrumental social support will moderate the effects of controllable work difficulties on job satisfaction. When the difficulty is controllable, it is proposed that other people may help the person to overcome the problem.

5) Emotional social support will moderate the effects of uncontrollable work difficulties on job satisfaction. Emotional social support is expected to help reduce the influence of uncontrollable difficulties on job satisfaction. In these cases, instrumental social support may not be useful as the situation cannot be overcome, however emotional social support may help reduce the severity of these difficulties.

6) Work difficulties, the control strategies used to handle work difficulties, social support at work, job autonomy and life satisfaction, will predict job satisfaction. These variables are expected to be major predictors of job satisfaction, as demonstrated in Figure 8.

271

3.5

Method

3.5.1

Participants The sample consisted of 214 general employees, obtained from a database and

through convenience sampling. The age of the participants ranged from 21-73 years, with the average being 44.78 years (SD=15.18). The demographic characteristics of the sample are displayed in Table 36. Table 36- Demographic Characteristics of the Sample Variable Gender Male Female Occupation Professional Business Trade Clerical Retail Labourer Other Hours worked per week 1-20 21-30 31-40 41-50 51-60 61+ Bolded values indicate the largest proportion. 3.5.2 Materials All of the respondents received a plain language statement (refer to Appendix N) and a questionnaire. The questionnaire consisted of scales measuring controllable 13.6 14.0 27.6 27.1 12.1 4.7 46.7 13.1 10.7 13.1 6.1 1.9 3.3 % sample 47.7 47.7

272

and uncontrollable difficulties, primary and secondary control for such difficulties, job autonomy, job satisfaction, life satisfaction and social support at work. 3.5.2.1 Controllable and Uncontrollable Difficulties Controllable and uncontrollable difficulties were measured in the Situation Specific Primary and Secondary Control Scale (Maher & Cummins, 2002; refer to Appendix O). Although the most direct way to measure this variable would be to ask, how often do you experience difficulties that you can control/cannot control, this item was deemed to be too cognitively taxing and prone to errors. Rather the employees were given a list of potential difficulties that they may face at work such as supervisors, co-workers, kind of work, pay, work-place rules, promotion, time management and others. They indicated which difficulties they experienced that they could control. In order to determine how frequently they experienced controllable difficulties, they were asked to consider the difficulty that they experienced most often and could control, and indicate how often they experienced it. This process was repeated for uncontrollable difficulties. Although this question only refers to one difficulty, it was deemed to be the best method. One alternative is to ask them on average how often they experience the difficulties they ticked. However, it is extremely difficult for employees to mentally calculate how often they experience each difficulty they ticked and then calculate the average. Another alternative would be to get them to indicate how often they experienced each difficulty. This was problematic however, as there are an unlimited

273

number of work difficulties, particularly as this study is relying on a general sample of employees. 3.5.2.2 Primary Control and Secondary Control The Situation Specific Primary and Secondary Control Scale (Maher & Cummins, 2002; refer to Appendix O) was used in this study. The main difference between this scale and earlier scales is that the respondents now indicate which control strategies they use for controllable difficulties and uncontrollable difficulties. There are four primary control, and 12 secondary control strategies from which to choose. There is also the option to list other strategies that they use. Each strategy is rated on a five-point scale ranging from 0 (never) to 4 (always). 3.5.2.3 Job Autonomy Job autonomy was measured by the job autonomy items in the Job Diagnostic Survey (Hackman & Oldham, 1975). This scale consists of three items that assess overall perceived job autonomy, such as In my job, I can decide on my own how to go about doing my work (refer to Appendix B). The items were rated on a 10-point scale ranging from 1 (do not agree at all) to 10 (agree completely). As discussed in study two, the psychometric statistics are adequate. 3.5.2.4 Job Satisfaction Job satisfaction was measured by a one-item measure. The item taking into consideration all the things about your job, how satisfied are you with it was rated on an 11-point scale. Although internal consistency cannot be established with a single-

274

item measure, this measure of job satisfaction has been shown to correlate with other measures of job satisfaction, where r = 0.63 (Wanous et al., 1997). In studies one and two, a facet measure of job satisfaction was also used. As this facet measure was only used to gain insight into the particular occupational groups, it is not necessary in this study which is relying on a general sample of employees. 3.5.2.5 Life Satisfaction Life satisfaction was measured by the Personal Well-being Index developed recently by Cummins et al., (2001). This scale attempts to overcome some methodological problems that were identified in the Comprehensive Quality of Life Scale, which was used in studies one and two. Most of these problems involve the objective scale, or the importance scale, rather than the satisfaction scale. However, some problems were identified with the seven domains of life satisfaction, such as the domain of emotional well-being or happiness refers to an affective state rather than a domain of life (Cummins, 2002). Furthermore, the wording of some of the items were not optimal (Cummins, 2002). As such, rather than happiness, future security is included as a domain of life satisfaction. Thus, the Personal Well-being Index consists of seven domains of satisfaction which are rated on an 11-point scale (refer to Appendix P). 3.5.2.6 Social Support at Work In study two, social support at work was measured by Karasek and Theorells (1990) scale. This scale consisted of both instrumental and emotional support,

275

however the items were designed to be summed to provide an overall support score. These items were examined more closely in this study to ensure that they were appropriate for a range of occupations. Some problems were identified with the instrumental support scale. For example, the item, my supervisor creates a good teamwork environment for me may not measure instrumental support. It is not necessary that an employees works in a teamwork environment for them to receive instrumental support. To provide instrumental support, the employer only needs to offer some material assistance. A further problem with the scale concerns the item my co-workers are competent. This item, although intended to measure instrumental support, only assesses the co-workers competence. Co-workers may indeed be competent, however this does not mean that they offer assistance when required. These problems, although they were not recognised in study two, will be rectified in this study. A review of the other major measures of social support at work was undertaken. This review demonstrated that although there are many scales that claim to measure social support at work, few adequately measure emotional and instrumental support. Some researchers rely on a one item measure, such as how true is it that your supervisors are warm/friendly when you have problems and how true is it that your supervisor helps you complete a given task (Himle & Jayaratne, 1991; Wong et al., 2000). These scales are criticised however for failing to capture the different ways that emotional or instrumental support can be offered. Some studies have relied on scales which only focus on emotional support and fail to measure instrumental support (Dollard et al., 2000, Rodriguez, Bravo, Peiro &

276

Schaufeli, 2001). Others do not claim to measure either component (Caplan et al., 1975; Van der Doef & Maes, 1999), whilst others still claim to measure five types of social support (Unden, 1996). One exploratory scale recently developed by Ducharme and Martin (2000) does not appear to suffer from any of these problems. Their scale, developed only to assess co-worker support, includes five items assessing instrumental support and five items assessing emotional support. Consistently, a factor analysis on the scale demonstrated that two factors emerged. To make this scale appropriate for the current study, it was posited that the co-worker items could also be applied to supervisors. If this were done, the scale would consist of 20 items. As this may be unnecessarily long, some items were deleted. Specifically, only the three highest loading items were selected each for instrumental and emotional support. For emotional support, these were your co-workers really care about you, you feel close to your co-workers and your co-workers take a personal interest in you. These three items were also changed to be applicable to supervisors. For instrumental support, the items were your co-workers would fill in while youre absent, your co-workers are helpful in getting the job done and your co-workers give useful advice on job problems. These three items also needed to be applicable to supervisors. As one of them was not (i.e., your supervisors would fill in while you were absent), the next highest loading item in the factor analysis was selected. This was your co-workers assist with unusual work problems. The resulting scale is a six-item scale for co-worker support and a six-item scale for

277

supervisor support, that both assess instrumental and emotional support (refer to Appendix Q). 3.5.3 Procedure Ethics approval was obtained from Deakin University. The majority of the sample was obtained from a database developed by Australian Unity and Deakin University. This database contains information for 900 people that have been randomly selected from the population, and have agreed to participate in a survey. The employment status of these people was unknown, and as such two questionnaires were sent to them, one if they were employed, and one if they were unemployed. A total of 250 (27%) questionnaires were returned however only 130 (14.44%) of these were completed by people that were employed. The remainder of the sample was obtained through convenience, and snowballing.

278

3.6

Results

3.6.1

Data Screening and Checking of Assumptions The data set was examined for missing values, outliers, normality and linearity.

There were very few missing values for measures of life satisfaction, job satisfaction, job autonomy, job demands, and co-worker support (i.e., < 5%). There was a higher rate of missing values for supervisor support (19%) and primary and secondary control (6%, 11%, respectively). The treatment of these values depended on their context. If the participant had completed the majority of the scale, the missing values were replaced with the group mean. If however, the person had failed to complete any of the scale, they were excluded from analyses using that scale. Overall, this treatment resulted in less than 5% of the missing values being replaced with the group mean. Univariate outliers were identified in the measures of life satisfaction (2 cases), job satisfaction (3 cases), job autonomy (1 case), co-worker support (4 cases), and primary and secondary control (3 cases). These values were re-coded to lie within three standard deviations of the mean. Many of the scales were negatively skewed, where the skew/standard error > 3. These include life satisfaction (-4.44), job satisfaction (-5.09), job autonomy (-7.07), and co-worker and supervisor support (-4.56, -5.15, respectively). As transformations are not recommended for variables that are mildly and naturally skewed (Tabachnick & Fidell, 1997), these variables were not transformed. Reasonably linear relationships were evident among the variables.

279

3.6.2

Descriptive Statistics and Inter-Correlations Table 37 contains the means and standard deviations of the major variables.

This table demonstrates that employees are using primary more than secondary control in controllable and uncontrollable situations. Other interesting findings are that employees report that their co-workers offer more instrumental and emotional support than their employers. Additionally, the level of life satisfaction is in the expected range. Table 38 displays the correlations among the major variables. Several variables correlate well with job satisfaction including job autonomy, life satisfaction, social support at work, and difficulties at work. Strong correlations are also observed among primary control in controllable situations and primary control in uncontrollable situations. Similarly, secondary control in controllable situations is strongly correlated to secondary control in uncontrollable situations.

280

Table 37- Means and Standard Deviations of the Major Variables Variable Job Satisfaction Job Autonomy Life Satisfaction-domain Co-worker emotional support Co-worker instrumental support Supervisor emotional support Supervisor instrumental support Frequency of controllable difficulties Frequency of uncontrollable difficulties Primary control for controllable difficulty Secondary control for controllable difficulty Primary control for uncontrollable difficulty Secondary control for uncontrollable difficulty M 71.67 78.18 73.68 74.63 78.03 65.86 70.55 58.70 57.07 71.98 53.74 65.56 53.10 SD 18.41 20.34 13.76 20.81 21.30 26.66 26.32 20.33 22.25 13.05 12.12 16.73 12.68

All scores have been converted to a percentage of scale maximum (%SM) which ranges from 0-100. The formula is (mean score for the original domain-1) x 100/ (number of scale points 1) Table 38- Inter-Correlations among Major Variables JS JS JA Cdif Udif PcC ScC PcU ScU LS Sup Cow 0.57 -0.37 -0.35 0.27 -0.12 0.23 -0.04 0.46 0.37 0.48 JA -0.22 -0.32 0.34 -0.25 0.18 -0.16 0.34 0.40 0.41 Cdif Udif PcC ScC PcU ScU LS Sup

0.42 -0.13 -0.03 -0.12 -0.05 -0.14 -0.21 -0.14

-0.14 0.10 -0.11 0.10 -0.13 -0.24 -0.14

-0.03 0.69 0.07 0.13 0.10 0.07

0.06 0.64 -0.13 0.04 -0.06

0.20 0.19 0.08 0.09

0.08 0.04

0.26 0.31

0.57

Bolded items p<0.01 JS - job satisfaction; JA - job autonomy; Cdif - controllable difficulties; Udif -uncontrollable difficulties; PcC - primary control for controllable difficulties; ScC - secondary control for controllable difficulties; PcU - primary control for uncontrollable difficulties; ScU - secondary control for uncontrollable difficulties; LS - life satisfaction; Sup - supervisor support; Cow - co-worker support.

281

3.6.3

Factor Analyses Factor analyses were conducted on the two exploratory scales in the study

measuring primary and secondary control strategies and social support at work. 3.6.4 Primary and Secondary Control Scale Two factor analyses were required to examine the primary and secondary control items for a controllable difficulty and an uncontrollable difficulty. For both of these analyses, the assumptions were met where Bartletts test of sphericity was significant, and Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy exceeded 0.60. For controllable difficulties, a principle component analysis with direct oblimin rotation yielded five factors. These five factors accounted for 53.8% of the variance, however only 6 of the 16 items loaded on only one factor, and as such, this five-factor solution could not be interpreted. The analysis was repeated using a four-factor solution. Ten items loaded on only one factor, and the primary control items loaded on a separate factor to the secondary control items. However, there was no pattern to the secondary control items that loaded on the remaining three factors. In response to this, a three-factor analysis was conducted. This analysis was much clearer, demonstrating that all of the primary control items loaded on Factor 1, and the secondary control items were mainly divided among Factor 2 and Factor 3. As demonstrated in Table 39, Factor 1 consists of all four primary control items. Two of the secondary control items negatively loaded on this factor as well, and as such, they will be excluded from the analyses. Factors 2 and 3 contain the

282

remaining secondary control items. The items that loaded on Factor 2 were support, vicarious, past success, behavioural avoidance and positive re-interpretation, whilst the items that loaded on Factor 3 were predictive-negative, wisdom, and attribution. There is a theoretical distinction among these items that was introduced in chapter 3. As demonstrated in Table 19, two functions of secondary control are posited, namely selfprotective and self-affirmative. Self-protective secondary control strategies reduce the negative impact of the situation, whilst self-affirmative secondary control strategies increase positive feelings about the self. All of the items that loaded on Factor 2 involve self-affirmation, whilst the items that loaded on Factor 3 involve self-protection. It must be noted however that some strategies that were expected to load on the two factors did not. The self-affirmative strategy of downward social comparison did not load on Factor 2. Furthermore, the self-protective strategies of goal disengagement, illusory optimism, and denial did not load on Factor 3. Despite this however, overall the factor analysis supports the two types of secondary control.

283

Table 39- Factor Analysis of Primary and Secondary Control Item in Controllable Situations F1 F2 F3 Discuss solutions with the people involved. 0.56 Choose a solution and act on it. 0.70 Work harder. 0.42 Keep trying. 0.60 Think that the difficulty doesnt matter. -0.61 Think that this difficulty will work out okay in the end. Sc5 Think that I knew this difficulty would 0.68 happen. Sc6 Think that I cant always get what I want. 0.73 Sc8 Think that I am better off than many other people. Sc9 Think that this difficulty is not my fault. 0.42 Sc11 Tell someone about this difficulty to make me 0.54 feel better. Sc12 Think of the success of my family/friends. 0.61 Sc13 Think about my success in other areas. 0.74 Sc14 Do something different, like going for a walk. 0.66 Sc15 Ignore this difficulty. -0.63 Sc16 Look for something else that is positive in the 0.65 situation. Eigenvalues 2.40 2.25 1.46 % Variance 15.01 14.04 9.12 Cumulative variance 15.01 29.05 36.17 Items with loadings less than 0.30 are not shown. Self-protective secondary control items are bolded. Self-affirmative secondary control items are italicised. Factor 1- Primary Control, Factor 2- Self-affirmative, Factor 3- Self-protective In addition to the controllable difficulties, a principal component analysis with direct oblimin rotation was conducted on the strategies used for uncontrollable difficulties. This analysis yielded six factors, however as only one item loaded on one factor, a five-factor solution, and a four-factor solution were requested. Both analyses could not be interpreted, as there was no pattern to the secondary control items. As such, a three-factor solution was conducted. This solution was remarkably similar to Item Pc1 Pc4 Pc7 Pc10 Sc2 Sc3

284

the analysis of controllable situations. As demonstrated in Table 40, all four primary control items loaded on Factor 1, and the secondary control items loaded on their respective types of secondary control (i.e., self-protective and self-affirmative). As with the analysis for controllable situations however, there are a few exceptions, where goal disengagement and denial loaded on Factor 1. Unlike the controllable analysis, Factor 2 included the self-affirmative strategy of downward social comparison, however it also included illusory optimism, which is a selfprotective strategy. Finally, Factor 3 was the same in both analyses where it excluded goal disengagement, illusory optimism, and denial.

285

Table 40- Factor Analysis of Primary and Secondary Control Items in Uncontrollable Situation F1 F2 F3 Discuss solutions with the people involved. 0.66 Choose a solution and act on it. 0.70 Work harder. 0.59 Keep trying. 0.61 Think that the difficulty doesnt matter. -0.68 Think that this difficulty will work out okay 0.41 in the end. Sc5 Think that I knew this difficulty would 0.63 happen. Sc6 Think that I cant always get what I want. 0.73 Sc8 Think that I am better off than many other people. 0.58 Sc9 Think that this difficulty is not my 0.54 fault. Sc11 Tell someone about this difficulty to make me 0.39 feel better. Sc12 Think of the success of my family/friends. 0.57 Sc13 Think about my success in other areas. 0.70 Sc14 Do something different, like going for a walk. 0.63 Sc15 Ignore this difficulty. -0.65 Sc16 Look for something else that is positive in the 0.66 situation. Eigenvalues 2.90 2.29 1.60 % Variance 18.16 14.33 10.02 Cumulative variance 18.16 32.49 42.51 Items with loadings less than 0.30 are not shown. Self-protective secondary control items are bolded. Self-affirmative secondary control items are italicised. Factor 1- Primary Control, Factor 2- Self-affirmative, Factor 3- Self-protective Item Pc1 Pc4 Pc7 Pc10 Sc2 Sc3

Commonalties among the two factor analyses were examined. This study aims to compare the primary and secondary control strategies for controllable with uncontrollable difficulties, and as such, the comparisons should be based on the same items. For primary control, all four items loaded on Factor 1 in both analyses, and as such they will all be included. For self-affirmative secondary control, the common

286

items were sc11, sc12, sc13, sc14, and sc16. As such, self-protective secondary control will be measured by these items (i.e., support, vicarious, present success, active avoidance, and positive re-interpretation). For self-protective secondary control, the common items were sc5, sc6, and sc9, which are predictive-negative, wisdom, and attribution. Table 41 demonstrates which self-protective and self-affirmative strategies were included in the analyses. Table 41-Secondary Control Items included in Analyses Type of Strategy Self-protective Self-protective Self-protective Self-protective Self-protective Self-protective Self-affirmative Self-affirmative Self-affirmative Self-affirmative Self-affirmative Self-affirmative 3.6.5 Strategy Item Attribution Predictive-Negative Wisdom Goal Disengagement Illusory Optimism Denial Support Vicarious Present Success Active Avoidance Positive Re-interpretation Downward Social Comparison Current Study Self-protective Self-protective Self-protective ---Self-affirmative Self-affirmative Self-affirmative Self-affirmative Self-affirmative --

Social Support at Work A principal components factor analysis with direct oblimin rotation was

conducted on the social support scale to ensure that the items were measuring four types of support, namely co-worker instrumental support, co-worker emotional support, supervisor instrumental support and supervisor emotional support. The assumptions were met where Bartletts test of sphericity was significant, and Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy exceeded 0.60.

287

As demonstrated in Table 42, all of the supervisor items loaded on Factor 1, and all of the co-worker items loaded on Factor 2. The instrumental support and the emotional support items for supervisors and co-workers did not load on different factors. As these factor loadings are very high, the scale should only be divided into co-worker support and emotional support. However, as the hypotheses refer to the specific types of support, the instrumental and emotional support items will remain separate. Table 42- Factor Analysis of the Social Support at Work Scale Item My co-workers really care about me. I feel close to my co-workers. My co-workers take a personal interest in me. My co-workers assist with unusual work problems. My co-workers are helpful in getting the job done. My co-workers give useful advice on job problems. My supervisor really cares about me. I feel close to my supervisor. My supervisor takes a personal interest in me. My supervisor assists with unusual work problems. My supervisor is helpful in getting the job done. My supervisor gives useful advice on job problems. Eigenvalues % Variance Cumulative % variance Items with loadings less than 0.30 are not shown. F1 F2 0.82 0.88 0.93 0.88 0.91 0.77

0.86 0.91 0.88 0.91 0.93 0.93 7.34 61.97 61.97

2.05 17.05 79.02

288

3.7

Hypothesis Testing

In order to test the hypotheses, repeated measures analyses of variance were conducted to compare the use of control strategies in different situations, and multiple regression analyses were used to examine the adaptiveness of the control strategies. Hierarchical regression analyses were conducted to examine the moderating role of the control strategies and social support on the relationship between work difficulties and job satisfaction. Finally, a standard multiple regression analysis was conducted to examine the major predictors of job satisfaction. As in study two, the alpha level was reduced to 0.01 in order to reduce the risk of Type I errors. 3.7.1 Hypothesis One- Use of Control Strategies for Controllable and Uncontrollable Difficulties Hypothesis one proposes that when workers face controllable difficulties, they use more primary control than secondary control, and when workers face uncontrollable difficulties, they use more secondary control than primary control. Before examining this hypothesis, the types of difficulties that employees are reporting as being controllable or uncontrollable are examined. As demonstrated in Table 43, the most common controllable difficulties were with time management, motivation and co-workers. The most uncontrollable difficulties were with pay, amount of work, work-place rules, and promotion.

Table 43 - Controllable and Uncontrollable Difficulties Reported by Employees

289

Difficulty Difficulties with supervisors Difficulties with co-workers Difficulties with kind of work Difficulties with pay Difficulties with work-place rules Difficulties with promotion Difficulties with time management Difficulties with motivation Difficulties with work times Difficulties with amount of work Other

% yes and controllable 38.8 60.7 52.8 23.4 30.4 12.1 62.1 62.6 49.5 42.1 4.7

% yes and uncontrollable 35.5 17.8 29.9 46.3 44.4 40.2 19.6 11.2 23.4 44.9 4.2

To test whether workers use more primary control when they face controllable difficulties and more secondary control when they face uncontrollable difficulties, two repeated-measures analyses of variance were conducted. The variables were normally distributed, and the homogeneity of variance was met. Keppel (1991) proposes that the adequacy of the sphericity tests has been questioned and as such, researchers should assume that sphericity does not hold, and make the adjustments. For the controllable difficulties, Mauchlys test of sphericity was not violated. However, Greenhouse-Geisser epsilon was greater than 0.75, and as such, the degrees of freedom were calculated using the Huynh-Feldt epsilon. The difference was significant, F (1.984, 400.712) = 99.64, p = 0.00. As demonstrated in Table 44, employees reported significantly more primary control than affirmative secondary control, F (1, 202) = 187.94, p = 0.00, and self-protective secondary control, F (1, 202) = 32.27, p = 0.00. Employees did not report more protective secondary control than affirmative secondary, F (1, 202) = 1.20, p = 0.27. Thus, consistent with the first part of hypothesis one, employees reported more primary control than secondary control for a controllable difficulty.

290

For the uncontrollable difficulties, Greenhouse-Geisser epsilon was greater than 0.75, and as such, Huynh-Feldt epsilon was used. This demonstrated that the difference was significant, F (2, 386) = 38.57, p = 0.00. As demonstrated in Table 44, employees reported more primary control than self-affirmative secondary control, F (1, 193) = 80.27, p = 0.00, and self-protective secondary control, F (1, 193) = 29.94, p = 0.00. Employees also reported more self-protective secondary control than self-affirmative secondary control, F (1, 193) = 8.78, p = 0.003. Overall however, inconsistent with the second part of hypothesis one, employees reported more primary than secondary control for uncontrollable difficulties. Table 44- Employees Use of Primary and Secondary Control in Controllable and Uncontrollable Situations Situation Controllable Strategy Primary Control Secondary Control- Self affirmation Secondary Control- Self protective Primary Control Secondary Control-Self-affirmation Secondary Control- Self-protective M 71.98 52.88 54.60 64.56 50.74 55.45 SD 13.05 16.70 16.21 16.73 16.76 16.90

Uncontrollable

To determine the most commonly used control strategies to handle controllable and uncontrollable difficulties, the means for each individual strategy were examined. As demonstrated in Table 45, the most commonly used primary control strategy for controllable and uncontrollable difficulties is keep trying and the least common strategy is discuss solutions with the people involved. The most common secondary

291

control strategy for controllable and uncontrollable difficulties is think that I am better off than many other people. The least common secondary control strategy for controllable difficulties is ignore this difficulty and for uncontrollable difficulties is think that the difficulty doesnt matter.

292

Table 45- Means and Standard Deviations of Individual Control Strategies Strategy Discuss solutions with the people involved. (pc) Choose a solution and act on it. (pc) Work harder. (pc) Keep trying. (pc) Think that the difficulty doesnt matter. (sc) Think that this difficulty will work out okay in the end. (sc) Think that I knew this difficulty would happen. (sc) Think that I cant always get what I want. (sc) Think that I am better off than many other people. (sc) Think that this difficulty is not my fault. (sc) Tell someone about this difficulty to make me feel better. (sc) Think of the success of my family/friends. (sc) Think about my success in other areas. (sc) Do something different, like going for a walk. (sc) Ignore this difficulty. (sc) Look for something else that is positive in the Situation. (sc) Bolded strategies have highest frequencies 3.7.2 Hypothesis Two- Adaptiveness of the Control Strategies for Controllable and Uncontrollable Difficulties Hypothesis two proposes that primary control is more positively related to job satisfaction than secondary control for controllable difficulties and that secondary control is more positively related to job satisfaction than primary control for uncontrollable difficulties. In order to test this hypothesis, two standard multiple regression analyses were conducted. Controllable M SD 66.00 23.64 73.28 19.35 67.49 20.56 77.58 18.52 33.99 26.72 53.33 27.78 55.67 51.42 67.49 56.16 56.40 46.43 54.56 45.81 24.87 61.21 25.19 24.83 22.98 23.44 27.51 27.30 23.48 26.69 22.59 23.08 Uncontrollable M SD 55.03 27.00 61.73 25.21 63.92 24.40 77.58 18.52 30.15 26.19 49.36 25.63 55.67 51.42 67.49 56.16 56.44 41.23 53.22 44.20 30.54 58.63 25.19 24.83 22.98 23.44 26.39 28.00 24.53 28.14 26.78 22.61

293

The assumptions of normality, linearity, and homoscedasticity of residuals were met for both analyses, and there was no evidence of multicollinearity. For the controllable situation, R was significantly different from zero, R = 0.26, F (3, 199) = 4.63, p = 0.004. As demonstrated in Table 46, only primary control predicted job satisfaction, accounting for 5% of the variance. Thus, consistent with the first part of hypothesis two, primary control was more positively correlated with job satisfaction than secondary control for controllable difficulties. For the uncontrollable difficulties, R was also significantly different from zero, R = 0.29, F (3, 190) = 5.87, p = 0.001. As with the controllable difficulty, only primary control predicted job satisfaction, accounting for almost 6% of the variance. Thus, inconsistent with the second part of hypothesis two, primary control was more positively correlated with job satisfaction than secondary control for uncontrollable difficulties.

294

Table 46- Standard Multiple Regression Analysis Predicting Job Satisfaction From Primary and Secondary Control
Difficulty Control Var PC SC- A SC- P JS 0.24** 0.03 -0.01 PC 0.12 -0.01 SC-A B 0.33 0.007 -0.08 R =0.26** Uncontrol PC SC-A SC-P 0.26** 0.13 -0.08 0.18 0.05 0.14 0.28 0.11 -0.11 R =0.29** 0.25 0.01 0.08 R2=0.09 5.95** 0.24 0.006 -0.08 R2=0.07 sr2(unique) 5.38**

0.09

AdjR2=0.05

AdjR2=0.07

**p<0.01 Control - controllable difficulty; Uncontrol - uncontrollable difficulty; JS- job satisfaction; PC - primary control; SC-A - self-affirmative secondary control; SC-P- self-protective secondary control. In order to determine the most adaptive control strategies for controllable and uncontrollable difficulties, the correlations between each individual strategy and job satisfaction are presented in Table 47. This table demonstrates that the most adaptive primary control strategy for both types of difficulties is keep trying and the least adaptive is work harder. The most adaptive secondary control strategy for a controllable difficulty is think that I am better off than many other people and in an uncontrollable situation is look for something else that is positive in the situation. The least adaptive secondary control strategy in both situations is ignore this difficulty.

295

Table 47- Correlations between Individual Control Strategies and Job Satisfaction for Controllable and Uncontrollable Difficulties Strategy Discuss solutions with the people involved. (pc) Choose a solution and act on it. (pc) Work harder. (pc) Keep trying. (pc) Think that the difficulty doesnt matter. (sc) Think that this difficulty will work out okay in the end. (sc) Think that I knew this difficulty would happen. (sc) Think that I cant always get what I want. (sc) Think that I am better off than many other people. (sc) Think that this difficulty is not my fault. (sc) Tell someone about this difficulty to make me feel better. (sc) Think of the success of my family/friends. (sc) Think about my success in other areas. (sc) Do something different, like going for a walk. (sc) Ignore this difficulty. (sc) Look for something else that is positive in the situation. (sc) **p<0.01 3.7.3 Hypothesis Three- The Moderating Role of Primary and Secondary Control In order to test hypothesis three proposing that primary control moderates the effect of controllable difficulties on job satisfaction and secondary control moderates the effect of uncontrollable difficulties on job satisfaction, two hierarchical multiple regression analyses were conducted. Controllable 0.11 0.23** 0.09 0.24** -0.22** 0.13 -0.09 -0.02 0.22** -0.10 -0.08 -0.04 0.12 -0.02 -0.25** 0.15 Uncontrollable 0.19** 0.15 0.14 0.29** -0.03 0.12 -0.08 0.006 0.17 -0.09 -0.03 0.004 0.19** -0.009 -0.19 0.31**

296

The assumptions of normality, linearity, and homoscedasticity of residuals were met for both analyses, and there was no evidence of multicollinearity. As demonstrated in Figure 9, difficulties (i.e., controllable or uncontrollable) were entered in the first step. In the second step the moderator variable was entered (i.e., primary or secondary control), and in the third step the interaction term was entered.

297

Figure 9 Primary and Secondary Control Moderate the Relationship between Work Difficulties and Job Satisfaction Order of Variable Entry a) Step 1 Step 2 Step 3 Controllable difficulties Primary control Controllable difficulties x primary control Job Satisfaction

b) Step 1 Step 2 Step 3 Uncontrollable difficulties Secondary control Uncontrollable Difficulties x secondary control Job Satisfaction

In the primary control analysis, R was significantly different from zero after controllable difficulties were entered, R = 0.32, F (1, 199) = 22.91, p = 0.00. As demonstrated in Table 48, R increased after primary control was added, R = 0.39, Finc (1, 198) = 12.05, p = 0.001. The addition of the interaction term however did not increase R, where R = 0.41, Finc (1, 197) = 2.99, p = 0.09. Thus, inconsistent with hypothesis three, primary control did not moderate the effect of controllable work difficulties on job satisfaction. In the secondary control analysis, R was significantly different from zero after uncontrollable difficulties were entered, R = 0.35, F (1, 189) = 26.68, p = 0.00. As

298

demonstrated in Table 48, R did not increase when secondary control was added, R = 0.36, Finc (1, 188) = 1.73, p = 0.19, or when the interaction term was added, R = 0.37, F (1, 187) = 1.35, p = 0.25. As such, inconsistent with hypothesis three, secondary control did not moderate the effect of uncontrollable work difficulties on job satisfaction.

299

Table 48- Hierarchical Multiple Regression testing the Moderating Role of Control Strategies on the Relationship between Work Difficulties and Job Satisfaction Step 1 IV Controllable difficulties DV JS B -0.29 R =0.32** 2 Controllable difficulties Primary control JS -0.28 0.32 R =0.39** 3 Controllable difficulties Primary Control Controllable difficulties x Primary control JS -0.81 -0.08 0.007 R =0.41 1 Uncontrollable difficulties JS -0.29 R =0.35** 2 Uncontrollable difficulties Secondary Control JS -0.30 R =0.36 3 Uncontrollable difficulties Secondary control Uncontrollable difficulties x Secondary control JS -0.56 -0.13 0.004 R =0.37 **p<0.01; JS Job satisfaction -0.32 R2=0.10 -0.31 0.23 R2=0.16 -0.89 -0.06 0.65 R2=0.17 -0.35 R2=0.12 -0.36 R2=0.13 -0.68 -0.09 0.39 R2=0.14 AdjR2=0.12 sr2(unique) 10.30** AdjR2=0.10 9.24** 5.15** AdjR2=0.15 2.82**

AdjR2=0.16 12.39** AdjR2=0.12 12.89** AdjR2=0.12

300

3.7.3.1

Summary Employees reported using more primary than secondary control strategies for

both controllable and uncontrollable difficulties. Primary control was more positively related to job satisfaction than secondary control for both types of difficulties. Although primary control was positively related to job satisfaction, it did not moderate the effect of controllable difficulties on job satisfaction. Similarly, secondary control did not moderate the effect of uncontrollable work difficulties on job satisfaction. 3.7.4 Hypothesis Four - Moderating Role of Instrumental Support To test hypothesis four, proposing that instrumental support moderates the relationship between controllable difficulties and job satisfaction, hierarchical regression analyses were conducted for co-workers and supervisors. The assumptions of normality, linearity, and homoscedasticity of residuals were met for both analyses, and there was no evidence of multicollinearity. In the first step difficulties were entered (i.e., controllable or uncontrollable). In the second step the moderator variable was entered (i.e., co-worker instrumental or supervisor instrumental), and in the third step the interaction term was entered. For co-workers instrumental support, R was significantly different from zero after step one, R = 0.32, F (1, 198) = 22.74, p = 0.00, where controllable difficulties accounted for 10% of the variance in job satisfaction. R increased after step two, R = 0.49, Finc (1, 197) = 30.72, p = 0.00, where difficulties and co-worker instrumental support accounted for 8% and 14% of the variance, respectively. R did significantly increase when the interaction term was added in step 3, R = 0.50,

301

Finc (1, 196) = 4.30, p = 0.03, where the interaction term accounted for 1.64% of the variance. This analysis, displayed in Table 49, is consistent with hypothesis four. Although not significant at 0.01, this analysis, as in study two, will be examined further as only a few studies have examined the moderating role of social support. Furthermore, as discussed in study two, it is difficult to achieve statistical significance in moderation analyses as the power is low (Bobko, 2001). Controllable work difficulties were regressed on job satisfaction separately for those with low co-worker support, and those with high co-worker support. As proposed by Cohen and Cohen (1983), the low and high distinction was defined as scores that fell one standard deviation below or above the mean for supervisor support. As demonstrated in Figure 10, the regression lines were consistent with the hypothesis, where the slope of the regression line of controllable work difficulties on job satisfaction was steeper for high co-worker instrumental support than for low coworker instrumental support.

Figure 10 - Regression of Controllable Work Difficulties on Job Satisfaction for Employees with Low Instrumental Co-Worker Support and Employees with High Instrumental Co-Worker Support

302

100 90 80 70 Job Satisfaction 60 50 40 30 20 10 0 0 10 20 30 40 50 60 70 80 90 100 Controllable Difficulites y = -0.0058x + 64.518 y = 0.0231x + 76.84

Low co-worker instrumental support High co-worker instrumental support

For supervisors instrumental support, R was significantly different from zero after step one, R =0.36, F (1, 176) = 25.92, p = 0.00. As demonstrated in Table 49, R increased after supervisors instrumental support was added, R = 0.42,

303

Finc (1, 175) = 10.73, p = 0.001, however it did not increase further when the interaction term was added, R = 0.42, F (1, 174) = 0.06, p = 0.80. Thus, inconsistent with hypothesis four, supervisor instrumental support did not moderate the effect of controllable work difficulties on job satisfaction.

304

Table 49- Hierarchical Regression Analyses Testing the Moderating Role of Instrumental Support Step 1 IV Controllable difficulties DV JS B -0.28 R =0.32** 2 Controllable difficulties Co-workers instrumental JS -0.26 0.31 R =0.49** 3 Controllable difficulties Co-workers instrumental Controllable difficulties x Co-workers instrumental JS -0.64 0.01 0.005 R =0.50* 1 Controllable difficulties JS -0.33 R =0.36** 2 Controllable difficulties Supervisors instrumental JS -0.28 0.16 R =0.42** 3 Controllable difficulties Supervisors instrumental Controllable difficulties x Supervisors instrumental JS -0.32 0.13 0.0005 R =0.42 *p<0.05, **p<0.01; JS- Job satisfaction 3.7.5 Hypothesis Five- Moderating Role of Emotional Support In order to test hypothesis five, proposing that emotional support moderates the effect of uncontrollable work difficulties on job satisfaction, two hierarchical multiple -0.32 R2=0.10 -0.29 0.37 R2=0.24 -0.72 0.02 0.55 R2=0.25 0.17 R2=0.13 -0.31 0.23 R2=0.18 -0.35 0.18 0.06 R2=0.18 AdjR2=0.17 sr2(unique) 10.30** AdjR2=0.10 8.49** 14.56** AdjR2=0.23 4.20 . 1.64* AdjR2=0.24 12.81** AdjR2=0.12 8.94** 5.02** AdjR2=0.17

305

regression analyses were conducted. The assumptions of normality, linearity, and homoscedasticity of residuals were met for both analyses, and there was no evidence of multicollinearity. For co-workers, R was significantly different from zero after uncontrollable difficulties had been entered, R = 0.33, F (1, 192) = 24.48, p = 0.00. R significantly increased after co-worker emotional support was entered, R = 0.55, Finc (1, 191) = 51.32, p = 0.00. However, as demonstrated in Table 50, the addition of the interaction term in step three was not significant, R = 0.56, Finc (1, 190) = 1.85, p = 0.18. Thus, inconsistent with hypothesis five, co-worker emotional support did not moderate the effect of uncontrollable work difficulties on job satisfaction. For supervisors, R was significantly different from zero after step one, R = 0.36, F (1, 173) = 24.95, p = 0.00. As demonstrated in Table 50, R did significantly increase after supervisors emotional support was entered, R = 0.45, Finc (1, 172) = 16.90, p = 0.00, but did not increase further when the interaction term was added, R =0.46, F (1, 171) = 0.76, p = 0.39. Inconsistent with hypothesis five, coworker and supervisor emotional support did not moderate the effect of uncontrollable work difficulties on job satisfaction.

306

Table 50- Hierarchical Regression Analyses Testing the Moderating Role of Emotional Support Step 1 IV Uncontrollable difficulties DV JS B -0.28 R =0.34** 2 Uncontrollable difficulties Co-workers emotional JS -0.23 0.38 R =0.55** 3 Uncontrollable difficulties Co-workers emotional Uncontrollable difficulties x Co-workers emotional JS -0.45 0.20 0.003 R =0.56 1 Uncontrollable difficulties JS -0.30 R =0.36** 2 Uncontrollable difficulties Supervisors emotional JS -0.24 0.20 R =0.45** 3 Uncontrollable difficulties Supervisors emotional Uncontrollable difficulties x Supervisors emotional JS -0.37 0.09 0.002 R =0.46 *p<0.05, **p<0.01; JS Job satisfaction -0.34 R2=0.11 -0.28 0.43 R2=0.30 -0.55 0.23 0.33 R2=0.31 -0.36 R2=0.13 -0.29 0.29 R2=0.20 -0.44 0.13 0.20 R2=0.21 sr2(unique) 11.29** AdjR2=0.11 7.67** 18.75** AdjR2=0.29 2.58**

AdjR2=0.30 11.22** AdjR2=0.12 7.95** 7.84** AdjR2=0.20 2.62*

AdjR2=0.19

307

3.7.5.1

Summary In summary, only co-worker instrumental support moderated the effect of

difficulties on job satisfaction. Supervisor instrumental support, supervisor emotional support and co-worker emotional support did not act as moderators. 3.7.6 Hypothesis Six- Major Predictors of Job Satisfaction In order to test hypothesis six, proposing that job autonomy, difficulties at work, control strategies, social support at work, and life satisfaction predict job satisfaction, a standard regression analysis was conducted. The assumptions of normality, linearity, and homoscedasticity of residuals were met for both analyses, and there was no evidence of multicollinearity. When all of the variables were entered, R was significantly different from zero, R = 0.74, F (12, 142) = 13.92, p = 0.00. As demonstrated in Table 51, controllable difficulties, job autonomy, life satisfaction, and co-workers emotional support uniquely predicted job satisfaction. Inconsistent with hypothesis six however, control strategies, uncontrollable difficulties, and supervisor social support did not uniquely predict job satisfaction.

308

Table 51- Standard Multiple Regression Predicting Job Satisfaction Predictor Controllable difficulties Uncontrollable difficulties Job autonomy Secondary control- Controllable situation Secondary control- Uncontrollable Situation Primary control- Controllable situation Primary control-Uncontrollable situation Life Satisfaction Co-workers emotional support Co-workers instrumental support Supervisors emotional support Supervisors instrumental support B -0.18 -0.11 0.24 1.30 -0.49 3.15 0.72 0.31 0.17 0.14 -0.04 -0.02 R =0.74** -0.18 -0.13 0.27 0.03 -0.01 0.09 0.03 0.19 0.19 0.16 -0.06 -0.02 R2=0.54 sr2(unique) 2.62** 4.08**

4.41** 1.49*

AdjR2=0.50

*p<0.05, **p<0.01; R is composed of 15.52% unique variance and 84.48% shared variance. 3.7.6.1 Summary When all of the variables in the hypothesised model of job satisfaction were entered into a regression equation, they accounted for 54% of the variance. However, only controllable difficulties, job autonomy, life satisfaction, and co-worker emotional support uniquely predicted job satisfaction. 3.7.7 Conclusion Employees reported using more primary than secondary control in both controllable and uncontrollable situations. Primary control was more adaptive than secondary control in both situations and was positively correlated with job satisfaction. However, primary and secondary control did not moderate the effect of work

309

difficulties on job satisfaction. Co-worker instrumental support did moderate the effect of controllable work difficulties on job satisfaction, however supervisor instrumental support did not. Furthermore, emotional support did not moderate the effect of uncontrollable difficulties on job satisfaction. The major predictors of job satisfaction were controllable difficulties, job autonomy, life satisfaction and co-worker instrumental support. These findings will now be discussed.

310

3.8

Discussion

The study proposed that the controllability of a work difficulty influences the use and adaptiveness of the control strategies used to handle that difficulty. The findings demonstrated, however, that for both controllable and uncontrollable difficulties, primary control strategies were used more than secondary control strategies, and primary control strategies were more adaptive than secondary control strategies. These findings, which are inconsistent with the discrimination model, suggest that trait control strategies may exist. The proposal that employees use similar control strategies in all situations questions the assumption that employees using primary control in uncontrollable situations will experience primary control failure. The results from this study also question the importance of the control strategies, as they, along with social support, did not moderate the effect of work difficulties on job satisfaction. These findings must be regarded with caution however as limitations have now been identified in the operationalisation of work difficulties. These hypotheses will now be examined. 3.9 Hypotheses Testing

The hypotheses can be grouped into three major proposals. The first proposal is that the controllability of a difficulty influences the use and adaptiveness of the control strategies used to handle that difficulty. The second proposal is that the control strategies and social support at work moderate the effects of work difficulties on job satisfaction. The third proposal is that general job autonomy, difficulties at work, control strategies, social support at work and life satisfaction predict job satisfaction.

311

Before these proposals are discussed, the conceptualisation of the control strategies requires explanation. 3.9.1 Primary Control, Self-Protective Secondary Control, and SelfAffirmative Secondary Control Factor analyses of the control strategies demonstrated that employees were using three types of control strategies, namely primary control, self-protective secondary control and self-affirmative secondary control. Although two types of secondary control were identified in chapter 2, it was not known whether the differences between them would be great enough to form separate factors. As this was the case however, the two types of secondary control require further exploration. All of the secondary control strategies involve people changing themselves to fit in with their situation, however there are two ways that this can be done. Selfprotective secondary control strategies reduce negative feelings about the situation. The strategies that loaded on the self-protective factor were attribution (Think that this difficulty is not my fault), predictive negative (Think that I knew this difficulty would happen), and wisdom (Think that I cant always get what I want). These strategies make the situation less concerning, and people conclude that a situation is not as bad as it seems. The second type of secondary control, self-affirmative, promotes positive feelings. The strategies that were identified as being self-affirmative were support (Tell someone about this difficulty to make me feel better), vicarious (Think of the success of my family or friends), present success (i.e., Think about my success in

312

other areas), active avoidance (Do something different, like going for a walk) and positive re-interpretation (Look for something else that is positive in the situation). These strategies make people feel good about themselves and their lives. It must be noted that this conceptualisation of self-protective and selfaffirmative secondary control was not completely supported. Four items did not load on the expected factors. Specifically, downward social comparison (Think that I am better off than many other people) did not load on the self-affirmative factor. Furthermore, goal disengagement (Think that the difficulty doesnt matter), illusory optimism (Think that this difficulty will work out okay in the end) and denial (Ignore this difficulty) did not load on the self-protective factor. There is no ready explanation as to why these items did not load on the expected factors. Clearly however, the majority of items were consistent with the conceptualisation of selfprotective and self-affirmative secondary control. Although this is a novel approach to secondary control strategies, it must be noted that the conceptualisation of these three strategies is still consistent with Heckhausen and Schulzs (1995) proposals. Specifically, primary control strategies involve attempts to change the environment to fit in with the self, and both types of secondary control strategies involve attempts to change the self to fit in with the environment. The new idea however is that some secondary control strategies reduce negative feelings, whilst others promote positive feelings. Although factor analyses have not been conducted on other primary and secondary control scales as they generally contain only one item (i.e., Thompson et al., 1996, 1994), they have been conducted on coping scales. The factors emerging from

313

these analyses can be compared to the three factors found in this study. As the most common coping scale is the Ways of Coping Questionnaire (Folkman & Lazarus, 1985; Folkman, Chesney, Cooke, Boccellari & Collette, 1994), factor analyses of this scale will be examined. Unlike the Situation Specific Primary and Secondary Control Scale (Maher et al., 2002), which uses one item for each strategy, the Ways of Coping Questionnaire (Folkman & Lazarus, 1985) uses multiple items for each strategy. As such, when factor analyses are conducted on the scale, the items cluster according to the type of strategy. For example, Folkman et al., (1986) demonstrated that a factor analysis, averaged across several samples yielded eight factors, including confrontative coping, distancing, self-controlling, seeking social support, accepting responsibility, escapeavoidance, planful problem solving and positive reappraisal. It must be noted however, that as discussed in chapter 1, factor analyses conducted on this scale are far from consistent (Edwards & ONeill, 1998). Factor analyses of the Ways of Coping Questionnaire are not comparable to those conducted on the Situation Specific Primary and Secondary Control Scale. Factor analyses of the Ways of Coping Questionnaire identify which items measure a particular strategy, whereas factor analyses of the Situation Specific Primary and Secondary Control Scale (Maher et al., 2002) identify which strategies cluster together. As such, the factor analyses in the current study are theoretically different from previous analyses. Rather than just examining whether items measure a strategy, they demonstrate how the strategies are related to each other. This means that the underlying purpose of the strategies can be examined.

314

The development of three types of control strategies is exploratory, and as such, further research is required. However, this conceptualisation may be useful in determining the best type of secondary control. It could be hypothesised that selfaffirmative secondary control would be more positively correlated with job and life satisfaction than self-protective secondary control, as rather than just decreasing negative feelings, they increase positive feelings. This proposal is not supported in the current study as both self-protective and self-affirmative secondary control strategies were not related to job satisfaction. Despite this however, further research may benefit from recognising there may be two types of secondary control. The three major proposals of this study will now be examined. 3.9.2 Proposal One: The Controllability of the Difficulty Influences the Amount and Adaptiveness of the Control Strategies Used to Manage that Difficulty

3.9.2.1

The Amount of Control Strategies Used for Controllable and Uncontrollable Difficulties It was hypothesised that employees would use more primary than secondary

control for controllable difficulties, and more secondary than primary control for uncontrollable difficulties. Support was found for the former part of the hypothesis, however no support was found for the latter as employees reported using more primary than secondary control for uncontrollable difficulties. These finding are partially consistent with the life span theory of control (Heckhausen & Schulz, 1995). This theory proposes that people prefer primary control

315

and that it has primacy over secondary control. Consistently, employees reported more primary than secondary control for controllable difficulties. However, the theory also proposes that when people are faced with uncontrollable situations, the probability of primary control failure increases, and control strategies focus on changing oneself rather than changing ones environment (Heckhausen & Schulz, 1995). This does not appear to be the case for the employees in this study however, as they report more primary than secondary control in uncontrollable situations. Only one other study has reported the amount of control strategies used in an uncontrollable situation. Thompson et al., (1996) demonstrated that HIV positive men in prison (i.e., low-control situation) reported slightly more primary control (M = 48.5%SM) than secondary control (M = 45%SM). This study, which is also inconsistent with the life span theory of control, was criticised in chapter 1 for measuring primary control using perceived control and secondary control using acceptance. However, it now appears that even when a new measure of primary and secondary control is used, employees report using more primary than secondary control for uncontrollable difficulties. Three explanations have been developed for this finding. 3.9.2.2 Why is Primary Control Used more than Secondary Control for Uncontrollable Difficulties? There are three possible explanations for the employees reporting more primary than secondary control for uncontrollable difficulties. First, it may be that

316

when completing the questionnaire, the respondents were unable to conceptualise uncontrollable difficulties. Second, primary control may be used first for all difficulties and secondary control may only be used if primary control fails. Third, the controllability of the situation may not influence the control strategies people use, and rather people may have trait control strategies. These explanations will be discussed. 3.9.2.3 a) Conceptualisation of Controllable and Uncontrollable Difficulties One reason why the employees may have reported higher primary control than secondary control in uncontrollable situations is that the employees were unable to conceptualise the distinction between controllable and uncontrollable difficulties. The terms controllable and uncontrollable were used because, although being abstract in nature, they did not bias the respondents as much as other constructs such as change, influence, do something about, or accept. Despite being abstract, it appears that the participants generally did understand these terms and the distinction between them. The results demonstrated that the majority of participants indicated that difficulties with time management, motivation and co-workers were controllable and difficulties with pay, amount of work, work-place rules and promotion were uncontrollable. As such, it appears that the participants understood what constituted a controllable and an uncontrollable difficulty, and hence this proposal does not explain why employees reported higher primary than secondary control in uncontrollable situations.

317

3.9.2.4

b) Primary control is Implemented First for Controllable and Uncontrollable Difficulties Another explanation for the finding that primary is used more than secondary

control for uncontrollable difficulties is that primary control is always implemented first. It was assumed that employees would rely on secondary control for uncontrollable difficulties in an attempt to avoid primary control failure. However, it must be noted that Heckhausen and Schulz (1995) proposed that primary control strategies are used first and it is possible that this applies in controllable and uncontrollable situations. Perhaps people attempt to change all situations using primary control, and if they fail, they then rely on secondary control strategies. If this is the case, it would be expected that people would use comparable amounts of primary control in controllable and uncontrollable situations, but that they would use more secondary control in uncontrollable situations. As demonstrated by the mean levels of primary and secondary control however, this does not appear to be case. The primary control levels were similar for controllable situations (M = 71.98) and uncontrollable situations (M = 64.56), however there was no difference in their levels of secondary control (controllable, M = 53.74, uncontrollable, M = 53.10). Hence, the proposal that employees report more primary control than secondary control in uncontrollable situations because they use primary control first in such situations and only use secondary control when primary control fails, does not appear to be supported.

318

3.9.2.5

c) Trait Control Strategies Another explanation for the finding that the employees reported more primary

control than secondary control for uncontrollable difficulties is that trait control strategies may exist. People may have a set of strategies that they consistently use to handle their difficulties, and they may not consider the usefulness of the strategy within that situation. The correlations between the control strategies used in controllable situations with the control strategies used in uncontrollable situations supports this proposal. Primary control for a controllable difficulty was strongly correlated with primary control for an uncontrollable difficulty (r = 0.69). Furthermore, secondary control for a controllable difficulty was strongly correlated with secondary control for an uncontrollable difficulty (r = 0.64). The correlations between primary and secondary control strategies were much weaker. Primary and secondary control strategies for controllable difficulties were not correlated. Primary and secondary control for uncontrollable difficulties were only weakly correlated (r = 0.20). The idea that peoples responses to difficulties are stable has been discussed in the coping literature. It is proposed that people have coping styles, dispositions, or traits that they bring to the situation (Carver et al., 1989). Accordingly, people do not approach each coping context anew, but rather bring to bear a preferred set of coping strategies that remains relatively fixed across time and circumstances (Carver et al., 1989, p. 270). Few researchers have examined trait coping, perhaps because Folkman and Lazarus (1986) disputed the idea, proposing that coping is contextual, and that it is

319

influenced by the persons appraisal of the situation. However, other studies besides the current research dispute this proposition. A study conducted by Schwartz, Neale, Marco, Shiffman and Stone (1999) assessed trait coping using the Daily Coping Questionnaire (Stone & Neale, 1984) and the Ways of Coping Questionnaire (Folkman & Lazarus, 1984). The question at the beginning of each scale was changed to how do you typically deal with stressful situations. They also measured coping using a momentary measure where participants were given a programmable palm-top computer. Participants would type in their stressful events and indicate how they coped with them immediately after the event. They examined how much of the variance in the momentary scales was due to differences among individuals. For example, for escape coping, they examined how much of the variance was due to the tendency for some individuals to report escape coping more than others. The results demonstrated that 42% of the variability in the momentary assessments was due to individual differences in escape coping. The other coping strategies accounted for less of the variance, ranging between 20-30% for the Ways of Coping Scale, and for 15-19% of the Daily Coping Scale. These findings suggest that a persons coping response could be partially predicted from a general coping scale, and thus supports trait coping. The proposal that coping is a trait or disposition can be used to explain the current findings. Employees may have reported using primary control for an uncontrollable difficulty because primary control strategies are within their disposition. Thus, rather than evaluating the situation, they evaluate the coping strategies they have in their repertoire.

320

3.9.2.6

Summary Consistent with the life span theory of control, employees reported using more

primary than secondary control for controllable difficulties. Inconsistently however, they also reported using more primary than secondary control for uncontrollable difficulties. Three explanations were developed to account for these findings. The first, proposing that employees did not understand uncontrollable difficulties, was dismissed, as employees seemed to classify their difficulties as expected. The second explanation proposed that people use more primary control for uncontrollable difficulties because they implement primary control first for all difficulties, and only use secondary control if primary control fails. This was not supported by the data, as the levels of secondary control were the same. The third explanation proposed that the controllability of the situation did not influence the control strategies the employees used. Rather, it is proposed that employees have trait control strategies. Employees may fail to evaluate the situation and rather simply use the strategies in their repertoire. The relationship between these control strategies and job satisfaction will now be examined. 3.9.2.7 Adaptiveness of Primary and Secondary Control for Controllable and Uncontrollable Difficulties It was hypothesised that primary control would be more adaptive than secondary control for controllable difficulties and that secondary control would be more adaptive than primary control for uncontrollable difficulties. Partial support was provided for this hypothesis, as primary control was more positively related to job

321

satisfaction than secondary control for controllable difficulties. Inconsistently however, primary control was also more positively related to job satisfaction than secondary control for uncontrollable difficulties. These findings are inconsistent with the discrimination model (Thompson et al., 1998), which proposes that primary control is the most adaptive strategy only for controllable situations. Rather, the findings support the primacy/back-up model (Thompson et al., 1998), which proposes that primary control is more adaptive than secondary control in both controllable and uncontrollable situations. As with the current study, past empirical studies have supported the primacy/back up model (Thompson et al., 1996; 1994; 1993; 1998). As limitations were identified in these studies however, it was thought that when these limitations were addressed, the discrimination model would be supported. These limitations, discussed in chapter 1, concern the measurement of perceived control and primary and secondary control strategies. A more notable flaw however is that these studies failed to adequately test the discrimination model and the primacy/back-up model. Rather than correlating the controllability of a situation with the control strategies used to handle that situation, these studies examined general levels of perceived control and control strategies. Some of the studies did examine the constructs at a more specific level (e.g., Thompson et al., 1996, 1994), however they then aggregated the items to obtain an overall measure of perceived control and an overall measure of primary and secondary control. It appears however that even when all of the limitations were addressed, the findings still supported the primacy/back up model.

322

The current findings, although referring to control strategies, can also be compared to the empirical studies on the goodness of fit hypothesis for coping strategies. These studies generally demonstrate that, consistent with the current findings, problem-focussed strategies are more adaptive than emotion-focussed strategies in controllable situations. They also demonstrate that emotion-focussed strategies are not more adaptive than problem-focussed strategies in uncontrollable situations (e.g., Bowman & Stern, 1995; Conway & Terry, 1992; Osowieki & Compas, 1998, 1999; Park, Folkman & Bostrom, 2001; Vitaliano et al., 1990). These studies were criticised for their research designs in the introduction. It appears however, that even when these problems are addressed, similar results are obtained. In summary, it appears that consistent with past studies, primary control is more adaptive than secondary control for both controllable and uncontrollable difficulties. As many flaws were identified with the past studies, it was expected that when these flaws were addressed, the results would be more consistent with the discrimination model. This is not the case however, and as such further exploration is needed to explain why primary control is more adaptive than secondary control in uncontrollable situations. 3.9.2.8 Why is Primary Control more Adaptive than Secondary Control in Uncontrollable Situations? The finding that primary control is adaptive in uncontrollable situations is contrary to intuition. As such, it is important that this finding can be explained theoretically. It was expected that if employees tried to change an uncontrollable

323

situation using primary control, they would fail and this failure would negatively influence perceived competence, self-efficacy, self-esteem (Heckhausen et al., 1997), and job satisfaction. The current findings, which demonstrate that primary control is positively related to job satisfaction for uncontrollable difficulties, challenge the assumptions regarding primary control failure. Primary control failure has not been measured in the past, or in the current study, as it is extremely difficult to assess. It requires the person to indicate how often they used each of the control strategies and indicate the successfulness of each strategy. This is cognitively taxing for the respondents, and if completed for primary and secondary control, would add another 17 items to each control scale (controllable and uncontrollable). More importantly however, it may not even be possible for people to recall this information. Whilst they may remember whether they solved a problem, it is unlikely that they can recall which strategy was more successful than others. Furthermore, it may actually be a combination of strategies that contributes to the problem being overcome. For these reasons, the successfulness of the strategies was not assessed. As primary control failure was not measured in the current study however, it is possible that it did not behave as expected. Firstly, it may be that the employees who are implementing primary control are not experiencing primary control failure. Secondly, employees may be experiencing primary control failure, yet experiencing few negative consequences. These explanations will be discussed.

324

3.9.2.9

Primary Control does not lead to Primary Control Failure In regard to the first explanation, employees who reported high primary control

for uncontrollable difficulties may have reported high job satisfaction because they successfully implemented the strategies. Perhaps people only use primary control when they know that they will be successful. Indeed, it seems maladaptive for people to use primary control if they know that it will lead to primary control failure. If it proposed that employees only use primary control when they know they will be successful, it must still be questioned how they could successfully change an uncontrollable situation using primary control. One possibility is that the difficulties reported by employees as being uncontrollable are only low-control difficulties. Most of the difficulties reported, such as pay, promotion and workplace-rules may not be completely uncontrollable. Other people determine them, and it is possible for the people to be influenced, and thus for primary control to be successful. Perhaps different results would be obtained if people were given difficulties that are clearly uncontrollable such as the death of a loved one or a natural disaster. 3.9.2.10 Primary Control Failure does not Negatively Influence Job Satisfaction

In regard to the second explanation, it may be that the employees are experiencing primary control failure, but that the primary control failure is not having negative effects. The life span theory of control proposes that primary control failure will threaten perceived competence, self-efficacy, and self-esteem (Heckhausen et al., 1997). It must be noted however that these effects have not been tested. Perhaps it is better to have tried to implement primary control and failed than to have not tried at

325

all. Employees can tell themselves that there was nothing more they could do, and thus they may feel better about their own control efforts. Both of these explanations are speculative, and indeed require empirical validation. To do this, future studies need to invest time in developing and measuring the successfulness of primary and secondary control strategies. 3.9.2.11 Summary

Although primary control was more adaptive than secondary control in controllable situations, it was also more adaptive in uncontrollable situations. These findings are inconsistent with the discrimination model and the goodness of fit hypothesis. It is difficult to explain as it was expected that employees who used primary control for uncontrollable difficulties would experience primary control failure. However, it may be that the employees only use primary control when they know that they will be successful. Alternatively, the employees may experience primary control failure, yet the consequences of primary control failure may be less damaging than not attempting at all. Further empirical research is required to examine these proposals. 3.9.3 Proposal Two: Moderators of Controllable and Uncontrollable Difficulties on Job Satisfaction Moderators of work difficulties were examined, as these variables may be more amenable to change than work difficulties. Employers may be reluctant to reduce work difficulties, where both the job and the organisation would need to undergo a

326

thorough assessment. Furthermore, it may be impossible to reduce some work difficulties if they are inherent in the nature of the work. It was hypothesised that the control strategies and social support at work would moderate the relationship between work difficulties and job satisfaction. Inconsistently however, primary control did not moderate the effect of controllable difficulties, and secondary control did not moderate the effect of uncontrollable difficulties. For the social support variables, only co-worker instrumental support moderated the effect of controllable work difficulties on job satisfaction. It must be noted that this finding was significant at 0.05, however it was not significant at the more stringent alpha level of 0.01. As few studies have examined the moderating role of social support in the workplace however, this finding was examined further. The finding that co-worker support played a greater role than supervisor support is inconsistent with other studies (e.g., Beehr, 1985; Fenlason & Beehr, 1994; Russell, Altmaier & Van Velzen, 1987). It was expected that as co-workers have less influence over difficulties at work, their support would not be as beneficial as supervisor support (Fenlason & Beehr, 1994). It must be noted however that the measure of supervisor support used in this scale was exploratory. Although the scale has face validity, there are no independent psychometric data for the scale. The findings will be compared to past studies. 3.9.3.1 Past Studies Examining the Moderators of Stress In regard to the control strategies, no other studies have examined whether the control strategies moderate the effect of work difficulties. However, a few studies

327

have demonstrated that coping strategies moderate the effect of stressors on stress (Aldwin & Revenson, 1987, Ashford, 1988; Parkes, 1990, 1994; Perrewe & Zellars, 1999; Osipow et al., 1985). These studies are inconsistent with the current findings, demonstrating that some coping strategies do moderate work stress. There is no conclusive evidence however, as to which control strategies moderate work stress. In regard to social support, a few studies, including study two, have demonstrated that social support at work has a moderating effect on job satisfaction (i.e., Karasek et al., 1982; Landsbergis et al., 1992). However, other studies have failed to find the moderating role of social support (Chay, 1993; de Jonge & Landeweerd, 1993; Melamed at al., 1991; Parkes & Von Rabenau, 1993). As discussed in chapter 3, one difference between the supportive and non-supportive studies is the measure of social support. Two of the supportive studies (i.e., current study and Landsbergis et al., 1992) relied on Karasek and Theorells (1990) scale. Although some of the items in Karasek and Theorells (1990) scale were criticised in this chapter, there is certainly no agreed way of measuring social support at work (Unden, 1996). The current study does not shed light on the problem however, as the social support scale did not factor as expected. The scale was only measuring two variables, supervisor support and co-worker support. Further research is needed on the operationalisation of social support to ensure that all four types of social support are assessed. In general, the findings on the moderating role of the control strategies and social support are somewhat inconsistent with other similar studies. One major difference between the current study and the other studies however is the independent

328

variable. Other studies have relied on job stress or work demands, whereas this study used work difficulties. This may have been problematic since work difficulties, controllable and uncontrollable, did not strongly predict job satisfaction. 3.9.3.2 Limitations in the Moderation Hypotheses The finding that work difficulties did not strongly predict job satisfaction is a concern for the robustness of this analysis. A moderation analysis tests whether the relationship between two variables (i.e., work difficulties and job satisfaction) varies depending on a moderator variable (i.e., control strategies or social support). A median split conducted on the moderator produces a low group and a high group. The relationship between work difficulties and job satisfaction for each group is then examined. If the relationship between the two variables is not strong for the average group however, it is unlikely that it will be strong when the moderator is low or high. It was expected that work difficulties would strongly predict job satisfaction, and as such, two explanations have been developed to account for the weak relationship. These concern the nature of work difficulties and the operationalisation of work difficulties. 3.9.3.3 Nature of Work Difficulties Researchers that have examined the moderating role of social support have examined job demands rather than work difficulties. Job demands are the psychological stressors in the work environment (i.e., high pressure of time, high working pace, difficult and mentally exacting work; Karasek & Theorell, 1990).

329

Work difficulties are much broader than job demands, and refer to any problems that employees face at work. 3.9.3.4 Operationalisation of Work Difficulties Work difficulties were measured by asking the employees to indicate how often they experienced their most commonly occurring difficulty. This is a difficult question to answer, as the employee needs to consider all of the difficulties that they face, think about how often they face each one and identify the one that they face the most. This item was useful in that it led people into thinking about how they handle that difficulty, however it may not have accurately assessed work difficulties. One person may experience one difficulty all the time yet rarely experience any other difficulties. Another person may experience ten difficulties all the time. Using the current scale however, these respondents would receive the same score. Thus, the difficulty at work scale requires revision. Perhaps the primary and secondary control scale could still include the item assessing the most common difficulty as this helps respondents to focus on a specific situation, however another measure of work difficulties is required. Developing a valid measure of work difficulties for a general sample of employees is problematic. The obvious solution is to ask respondents on average how often they face controllable and uncontrollable difficulties at work. These items may be prone to errors however as they are cognitively taxing, requiring the employee to mentally average their work difficulties.

330

Another solution is to ask respondents to indicate how often they experience each difficulty that they select from a list. Hence, as with the current scale, the respondents would be given a list of general work difficulties. They would tick which ones they experience and could control and then indicate how often they experience each difficulty. They would then do the same for uncontrollable difficulties. The problem however is that with the addition of the frequency item, the length of the scale doubles. Furthermore, there are an unlimited number of work difficulties and as such, some would be omitted. Another solution is to develop a list of occupational specific difficulties and ask employees how often they experience them. This solution, although it would enable the testing of the moderation hypotheses, is discouraged however, as different occupational groups cannot be compared. One final solution is offered. An open-ended format could be used, where respondents are asked to list their top five difficulties at work, and for each one, indicate how often they face it. This solution may be more time-consuming for the researcher to code, however it is not too cognitively taxing and it can be applied to a general sample of employees. 3.9.3.5 Summary There was little support for the moderating role of the control strategies and social support on the relationship between work difficulties and job satisfaction. These findings are limited however by the operationalisation of work difficulties. The scale only examined the most frequently occurring difficulty and as such, did not provide an

331

accurate assessment of work difficulties. Future researchers may need to use an openended format, where respondents are asked to list their difficulties at work and indicate how often they face each one. 3.9.4 Proposal Three: Predictors of Job Satisfaction It was hypothesised that job autonomy, difficulties at work, control strategies, social support at work, and life satisfaction would predict job satisfaction. Only controllable difficulties, job autonomy, life satisfaction, and co-worker emotional support uniquely predicted job satisfaction (15%). Thus, primary and secondary control strategies, uncontrollable difficulties, co-worker instrumental support, and supervisor emotional support and instrumental support did not uniquely predict job satisfaction. Possible explanations for these findings are discussed. The finding that the control strategies did not uniquely predict job satisfaction is particularly difficult to explain. It is intuitive that the control strategies that employees use to handle work difficulties influence their level of job satisfaction. One possibility is that it may not be primary and secondary control alone that predict job satisfaction, rather the effectiveness of the control strategies. Future studies may need to assess the control strategies and the effectiveness of them. The finding that uncontrollable difficulties did not uniquely predict job satisfaction may reflect the operationalisation of work difficulties. Employees were asked how often they face their most commonly occurring controllable and uncontrollable difficulty. As discussed previously, this measure may be flawed and as

332

such, more research is required to understand the importance of work difficulties in predicting job satisfaction. The finding that supervisor support did not uniquely predict job satisfaction may also be explained by its measurement. As discussed previously, the social support scale did not factor as expected, and it appeared as though the scale was only measuring two variables, supervisor support and co-worker support. The scale was a co-worker scale that was extended to supervisors. Perhaps separate scales are required for the different roles. As such, further research is needed on the operationalisation of social support to ensure that all four types of social support are assessed. 3.9.4.1 Summary Partial support was provided for the proposed predictors of job satisfaction, as controllable difficulties, job autonomy, life satisfaction and co-worker emotional support uniquely predicted job satisfaction. The finding that primary and secondary control strategies, uncontrollable difficulties, co-worker instrumental support and supervisor emotional and instrumental support did not uniquely predict job satisfaction may be due to operationalisation issues. 3.9.5 Conclusion The study tested three major proposals, which centered on job satisfaction, control strategies and the controllability of the situation. The first proposal that the controllability of the difficulty influences the use and adaptiveness of the control strategies used for that difficulty, was not supported. Employees reported using more primary control than secondary control for controllable and uncontrollable difficulties.

333

Three explanations were developed to account for these findings, however the most plausible was that people have trait control strategies. In addition to being used more than secondary control, primary control was also more adaptive than secondary control for controllable and uncontrollable difficulties. These findings, which are inconsistent with the discrimination model, challenge the assumptions about primary control failure. It is possible that primary control was adaptive for uncontrollable difficulties because it was being implemented successfully. If it was being implemented successfully, perhaps the work difficulties at work were low-control rather than being uncontrollable. Alternatively, the employees may have been experiencing primary control failure, however that failure may not have negatively affected job satisfaction. The second major proposal, that the control strategies and social support at work moderated the effects of work difficulties on job satisfaction, was generally not supported. The findings tended to be inconsistent with previous studies examining job stress, and the replacement of job stress with work difficulties was questioned. Specific problems with the operationalisation of work difficulties were identified that may have limited the findings. The third proposal, that general job autonomy, difficulties at work, control strategies, social support at work and life satisfaction predict job satisfaction was partially supported. On the basis of these findings, it was clear that measures of primary and secondary control, work difficulties, and social support require further exploration.

334

In summary, these findings suggest that a satisfied worker has high job autonomy, high social support, high life satisfaction, few work difficulties, and uses primary control to deal with controllable and uncontrollable difficulties. The implications of these findings will be discussed in chapter 5.

4 Chapter 5 - Final Discussion

335

336

4.1

Abstract

This thesis tested a model of job satisfaction that includes environmental and dispositional predictors. The major proposal of the model is that job autonomy influences the use and adaptiveness of primary and secondary control strategies. The model also examines other predictors of job satisfaction, including life satisfaction, work difficulties, and social support at work. Additionally, it proposes that the control strategies and social support at work moderate the relationship between work difficulties and job satisfaction. Empirical support offered for these proposals in chapters two, three and four are reviewed and a revised model of job satisfaction is presented. This model continues to include job autonomy, primary and secondary control, life satisfaction and work difficulties, however it also includes the successfulness of primary and secondary control and re-introduces personality variables.

337

4.2

The Development of A New Model of Job Satisfaction

This thesis developed a model of job satisfaction that includes environmental (i.e., job autonomy, social support at work, and work difficulties) and dispositional predictors (i.e., primary and secondary control, personality and life satisfaction). This model extended the job demand-control model (Karasek & Theorell, 1990), offering an alternative explanation for the positive relationship between job autonomy and job satisfaction. The job demand-control model was selected for further investigation because, unlike other dominant theories, it is highly applicable to the workplace and attractive to employers. The job demand-control model proposes that job demands and job decision latitude interact to predict job satisfaction, and that the most satisfied workers are those who have high job decision latitude and high job demands. The implication of this proposal is that employers can increase job satisfaction without reducing work demands. According to Karasek and Theorell (1990), employees with high job decision latitude can translate the physiological arousal produced from job demands into action through effective problem solving. They propose that workers with high job autonomy are given the freedom to decide what is the most effective course of action in response to a stressor (Karasek & Theorell, 1990, p. 36). However, this explanation has been criticised for being tautological. It proposes that job control, or the ability to choose at work, increases job satisfaction because it allows people to choose how they deal with their demands at work.

338

An alternative explanation for the positive link between job autonomy and job satisfaction is that employees with high job autonomy have higher job satisfaction because they respond differently to work difficulties. Employees can respond to work difficulties in two ways; they can either change the situation using primary control or they can change themselves using secondary control. It is expected that these primary and secondary control strategies mediate the relationship between job autonomy and job satisfaction. Job autonomy is expected to influence the amount of control strategies that employees use and the adaptiveness of those strategies. In regard to the amount of control strategies, employees with high job autonomy are expected to rely on more primary control and less secondary control than employees with low job autonomy. As primary control strategies are preferred over secondary control strategies, employees with higher job autonomy have higher job satisfaction than employees with lower job autonomy. In regard to the adaptiveness of the strategies, it is expected that primary control strategies are more adaptive than secondary control only when the situation is controllable. When the situation is uncontrollable, secondary control is expected to be the most adaptive strategy. The major proposal of the model of job satisfaction is thus that: 1) primary and secondary control mediate the relationship between job autonomy and job satisfaction. However, several other propositions are also examined, including that; 2) social support at work and life satisfaction are positively related to job satisfaction and; 3) the

339

control strategies and social support at work moderate the relationship between work difficulties and job satisfaction. Empirical tests of these proposals will be examined.

1) Primary and Secondary Control Strategies Mediate the Relationship Between Job Autonomy and Job Satisfaction It is expected that primary and secondary control strategies explain the relationship between job autonomy and job satisfaction. Job autonomy is expected to influence the amount of control strategies than employees report, and the adaptiveness of the control strategies. 4.2.1.1 Job Autonomy Influences the Use of Primary and Secondary Control Strategies It is expected that all employees, with either low or high job autonomy, implement primary and secondary control strategies. According to the life span theory of control (Heckhausen & Schulz, 1995), primary control has primacy over secondary control as it is preferred and is implemented first. If primary control is implemented successfully, the problem is resolved. If primary control fails however, the person is expected to implement secondary control strategies to compensate for, and avoid, future primary control failure. When these propositions are applied to the workplace, it is expected that job autonomy influences the likelihood of primary control failure. It is proposed that job autonomy is inversely related to the probability of primary control failure, which in turn, influences the use of secondary control strategies. Thus, employees with high job

340

autonomy are expected to experience less primary control failure and as such, use less secondary control than employees with low job autonomy. This proposal was tested by comparing the control strategies reported by low job autonomy workers with those reported by high job autonomy workers. Study one compared supermarket workers and academics, whilst study two compared teachers and academics. Both of these studies provided minimal support. In study one, the supermarket workers reported similar levels of primary control and more secondary control, than the academics. Although these results suggest that job autonomy influences the use of secondary control, but not primary control, it must be noted that these results are based on the levels of job autonomy inferred from type of occupation. Thus, it is assumed that supermarket workers are low in job autonomy and academics are high in job autonomy. When the same analysis was conducted with the reported levels of job autonomy, the results changed, in that only primary control was related to job autonomy. Thus the findings from study one suggest that job autonomy influences the use of primary control, but not secondary control. These findings were limited however, as the primary and secondary control scale used in this study was flawed. The rating scale did not assess how much control strategies the person was using, rather how much they agreed with the strategies presented to them in the scale. The primary and secondary control scale was revised for study two and administered to teachers and academics. This study was not supportive of the proposals however, as the groups reported similar levels of primary and secondary control. When studies one and two are considered together, it appears as though there

341

is little support for the proposal that job autonomy influences the use of the control strategies. These findings were attributed to, in part, the specificity of the hypotheses. Studies one and two examined the proposal that job autonomy influences the control strategies at a general level, measuring how much control employees have over their work environment and how they generally handle work difficulties. It was expected that this relationship may increase in strength however if the hypotheses were more specific. In this case, the controllability of one situation would be correlated with the control strategies used to handle that situation. As such, study three examined the amount of control strategies that employees used for controllable and uncontrollable difficulties. It was hypothesised that employees would use more primary than secondary control for controllable difficulties and more secondary than primary control for uncontrollable difficulties. Inconsistently however, employees reported more primary than secondary control strategies for controllable and uncontrollable difficulties. One explanation for this finding is that employees have trait control strategies. The use of primary and secondary control for controllable difficulties was highly correlated with the use of primary and secondary control for uncontrollable difficulties. Thus, people may have a set of strategies that they consistently use to handle their difficulties. Rather than evaluating the controllability of each work difficulty, employees may simply use the strategies that they know. If the amount of primary and secondary control used by employees remains stable across situations, then a dispositional factor, such as personality may predict the

342

control strategies. A few researchers have previously examined how personality variables relate to coping strategies (Brebner, 2001; Carver et al., 1989; Gunthert, Armeli & Cohen, 1999; Saklofske & Kelly, 1995; Scheier, Weintraub & Carver, 1986). It is proposed that people with high extroversion use active coping strategies where they talk out their problems and people with high neuroticism use passive coping strategies where they tend to blame themselves, and also other people (Costa, Somerfield & McCrae, 1996). These proposals are extended to the control strategies, where it is expected that extroversion is positively related to primary control and neuroticism is positively related to secondary control. The correlations between extroversion and neuroticism and the control strategies were examined in studies one and two. In regard to extroversion, study one demonstrated that primary control was positively correlated with extroversion for the academics, r = 0.25, and the supermarket workers, r = 0.42. Furthermore, in study two, primary control was positively related to extroversion for the academics, r = 0.20. Thus, these findings suggest that people high on extroversion use more primary control. In regard to neuroticism, study one demonstrated that there was no relationship between secondary control and neuroticism. Study two provided some support, as teachers levels of neuroticism were positively related to secondary control (r = 0.19). Although these findings suggest that neuroticism is at best, only weakly correlated with secondary control, studies using coping strategies have demonstrated that neuroticism is strongly correlated with emotion-focussed coping (i.e., Brebner, 2001; Saklofske & Kelly, 1995). These higher correlations may reflect the difference

343

between emotion-focussed coping strategies and secondary control strategies. In general emotion-focussed strategies tend to be more negative than secondary control strategies and thus may be more positively correlated with neuroticism. As the secondary control strategies developed for this study included more positive strategies, further research may be required to examine which personality variables predict secondary control. It might be useful to examine how the remaining personality variables (i.e., conscientiousness, agreeableness and openness) relate to secondary control. In regard to the proposed model of job satisfaction, the finding that employees reported similar levels of control strategies in controllable and uncontrollable situations suggests that changes need to be made to the model. As such, rather than job autonomy, it is proposed that personality predicts the use of the control strategies. 4.2.1.2 Summary There was marginal support for the proposal that job autonomy predicts the use of the control strategies. When this proposal was changed to be more specific, the controllability of the difficulty did not influence the use of primary and secondary control strategies. The finding that employees reported similar levels of primary and secondary control for controllable and uncontrollable difficulties suggests that trait control strategies may exist. Employees may have a set of control strategies that they regularly use, irrespective of the controllability of the problem. As such, the model of job satisfaction is changed so that personality predicts the control strategies rather than job autonomy or the controllability of the situation.

344

4.2.1.3

Job Autonomy Influences the Adaptiveness of Primary and Secondary Control The relationship between the control strategies and job satisfaction is expected

to change depending on the level of job autonomy. This hypothesis is based on the discrimination model, which proposes that primary control is the more adaptive strategy in controllable situations and that secondary control is the more adaptive strategy in uncontrollable situations. This model underlies the philosophy of the serenity prayer; Grant me the strength to change what I can, the patience to accept what I cannot, and the wisdom to know the difference (Thompson et al., 1998, p. 587). An alternative model has also been developed, namely the primacy/back-up model. This model proposes that primary control is more adaptive than secondary control in controllable and relatively uncontrollable situations. Previous empirical studies have supported the primacy/back-up model (Thompson et al., 1996; 1994; 1993; 1998), however these studies were criticised for their measurement of the controllability of the situation and the control strategies. In studies one and two, the correlations between the control strategies and job satisfaction for the low job autonomy group were compared to the correlations for the high job autonomy group. Study one supported the primacy/back up model, demonstrating that primary control was the most adaptive strategy for both the academics and the supermarket workers. As the scale of primary control was subsequently criticised, the proposal was re-tested with a revised scale in study two. Study two did not support the primacy/back-up model or the discrimination model, demonstrating that primary and secondary control strategies were not related to job satisfaction.

345

As mentioned previously, in both of these studies, the hypotheses were not consistent with the definition of the discrimination model or the primacy/back-up model. The hypotheses were too general and as such were made more specific in study three. In this study, the relationships between the control strategies for controllable and uncontrollable difficulties and job satisfaction were examined. The findings from study three refuted the discrimination model and supported the primacy/back-up model. Primary control was more adaptive than secondary control for controllable and uncontrollable difficulties. These findings suggest that employees should use primary control whenever they face a difficulty at work, even if it is uncontrollable. The proposal that primary control is adaptive in uncontrollable situations is difficult to explain as is it is assumed that they are likely to experience primary control failure. It must be noted however that primary control failure was not measured, and as such, the assumption that primary control in uncontrollable situations results in primary control failure may be inaccurate. It is possible that employees using primary control for uncontrollable difficulties report higher job satisfaction because they implemented it successfully. As such, the successfulness of the control strategies must be measured in future studies. This proposal is incorporated in the revised model of job satisfaction. It is now proposed that primary and secondary control strategies are not directly related to job satisfaction, rather that they indirectly influence job satisfaction through the successfulness of the control strategies. For example, suppose two employees report having primary control strategies in their repertoire, however only one of them

346

implements primary control successfully. It would be expected that the employee who is successfully implementing primary control would report higher job satisfaction than the employee experiencing primary control failure. As such, the successfulness of primary control may be a better predictor of job satisfaction than primary control directly. It is expected that if employees successfully implement the strategies, they will report higher job satisfaction. 4.2.1.4 Summary The controllability of the difficulty did not influence the relationship between the control strategies and job satisfaction. Even when the situation was uncontrollable, primary control was the most adaptive strategy. These findings, along with previous research, refute the discrimination model and support the primacy/back-up model. The primacy/back-up model is difficult to explain as people who use primary control in uncontrollable situations are expected to experience primary control failure. Primary control failure was not measured however, and as such, the successfulness of the control strategies must also be measured in future studies. The model of job satisfaction is revised where it is proposed that the control strategies are indirectly related to job satisfaction through the successfulness of the control strategies. 4.2.2 Conclusion: Do the Control Strategies Mediate the Relationship Between Job Autonomy and Job Satisfaction? The above findings demonstrate that the control strategies do not explain the relationship between job autonomy and job satisfaction. As such, the question of why job autonomy is related to job satisfaction remains unanswered. One possibility is

347

self-determination. According to DeCharms (1968) and Deci and Ryan (1986), humans have an innate need for competence and self-determination. Individuals attempt to seek out situations that challenge them. They find these activities rewarding and experience positive emotions such as enjoyment and excitement (Fay & Frese, 2001; Ryan & Deci, 2001). Another possibility is job status. Employees with high job autonomy have jobs that generally involve more responsibility and job status than employees with low job autonomy. Although few studies have examined the relationship between job status and job satisfaction, one study has demonstrated that female employees with higher job status tend to report higher job satisfaction than females employees with lower job status (Secret & Green, 1998). Another possibility is self-esteem. Employees with high job autonomy may feel that their employer trusts them and thus may value themselves more than employees with low job autonomy. Self-esteem has been shown to be positively related to job satisfaction, where the average correlation is r = 0.26 (Judge & Bono, 2001). Thus, job status or self-esteem may mediate the relationship between job autonomy and job satisfaction. It is important that researchers continue to examine why job autonomy is related to job satisfaction as the explanation offered by Karasek and Theorell (1990) in the job demand-control model (Karasek & Theorell, 1990) is tautological and vague. It is necessary that researchers understand the mechanism underlying the proposal that job autonomy can reduce the influence of job demands.

348

4.2.3

2) Social Support at Work and Life Satisfaction Directly Predict Job Satisfaction The next major proposal of the model of job satisfaction is that social support

at work and life satisfaction predict job satisfaction. 4.2.3.1 Social Support at Work Social support is expected to be directly related to job satisfaction. In study two, supervisor support (r = 0.64, r = 0.46), and co-worker support (r = 0.39, r = 0.42) were positively correlated with job satisfaction for the teachers and academics, respectively. Furthermore, study three demonstrated that supervisor support (r = 0.37) and co-worker support (r = 0.48) were moderately correlated with job satisfaction. In regard to the proposed model of job satisfaction, social support at work appears to be an important predictor. 4.2.3.2 Life Satisfaction Life satisfaction is expected to be positively related to job satisfaction. In study one, life satisfaction was not related to job satisfaction for the supermarket workers, however it was weakly related for the academics, r = 0.20. The results were stronger in study two, where life satisfaction was moderately correlated with job satisfaction for the academics, r = 0.38 and the teachers, r = 0.46. Study three also demonstrated, using a general sample of employees that r = 0.46. On the basis of these findings, it is concluded that life satisfaction is a direct predictor of job satisfaction.

349

The positive correlations between life satisfaction and job satisfaction support the spillover model, which proposes that satisfaction in one domain of an individuals life extends into other areas. Life satisfaction may spillover into job satisfaction or job satisfaction may spillover into life satisfaction. Thus, employers need to ensure that their employees are satisfied with all major areas of their lives, not just the workplace. Employees also need to be satisfied with their standard of living, their health, their personal relationships, their safety, and feeling part of their community. The levels of life satisfaction reported by the employees are particularly interesting. According to Cummins (2000b), life satisfaction is held under homeostatic control. Using two standard deviations to define the normative range, it is predicted that the mean subjective life satisfaction of Western population samples lay within the range 70-80%SM (Cummins, 1995). Consistently, all mean levels lay within the 7080%SM range, M = 78.31, M = 73.09, M = 74.20, M = 75.61, M = 73.68. The finding that life satisfaction can be predicted within such a small range is remarkable. Even the employees with low job autonomy (i.e., supermarket workers and teachers) reported levels of life satisfaction that were within the normative range. The mechanisms that underlie this prediction involve personality, perceived control, optimism, and self-esteem (Cummins, 2000b). More empirical studies are needed to examine how these predictors are related to life satisfaction. These results are not only important in developing a theory of life satisfaction, but these predictors are important for employers attempting to increase job satisfaction.

350

4.2.3.3

Summary Social support at work and life satisfaction both directly predicted job

satisfaction and are included in the revised model of job satisfaction. It is proposed that social support influences job satisfaction and that life satisfaction and job satisfaction influence each other. 4.2.4 3) The Control Strategies and Social Support at Work Moderate the Relationship Between Work Difficulties and Job Satisfaction

4.2.4.1

Moderating Role of Control Strategies Previous researchers have suggested that it is not the stressor that predicts job

satisfaction, but rather how the person deals with the stressor (Aldwin & Revenson, 1987, Ashford, 1988; Parkes, 1990, 1994; Perrewe & Zellars, 1999; Osipow, Doty & Spokane, 1985). Thus, it is expected that if employees match their control strategies to the situation, the negative influence of work difficulties on job satisfaction is lessened. Study three did not support this proposal however, demonstrating that primary and secondary control did not act as moderators. These findings suggest that even if employees match their control strategies to the situation, the negative influence of work difficulties on job satisfaction is not lessened. As such, this part of the model of job satisfaction requires revision. An alternative proposal is offered. Rather than the control strategies moderating the effect of work difficulties on job satisfaction, the successfulness of the strategies may be important. Thus, it is expected that if employees successfully

351

implement the matching control strategies, the influence of work difficulties on job satisfaction decreases. The model of job satisfaction is thus altered, where the successfulness of primary control moderates the effect of controllable difficulties, and the successfulness of secondary control moderates the effect of uncontrollable difficulties. 4.2.4.2 Moderating Role of Social Support In regard to social support at work, it is expected that co-worker and supervisor support moderate the effect of work difficulties on job satisfaction. Study two demonstrated that supervisor support, but not co-worker support moderated work difficulties. As supervisor support increased, the relationship between work difficulties and job satisfaction decreased. Study three examined the types of social support, proposing that instrumental support buffers the effects of controllable difficulties and emotional support buffers the effects of uncontrollable difficulties. Marginal support was found for this proposal, as co-worker instrumental support moderated the relationship between controllable work difficulties and job satisfaction. These findings in study two and three are somewhat inconsistent. This inconsistency may be attributed to, in part, the measurement of work difficulties. In study two, general work difficulties were measured by the item how often do you face difficulties at work? This item is prone to errors as it is cognitively taxing, requiring the employee to mentally average their work difficulties.

352

In study three, controllable and uncontrollable work difficulties were measured by asking employees how often they face their most commonly occurring controllable and uncontrollable difficulty. This scale was also criticised as it only focused on one difficulty. One person may experience one difficulty all the time yet rarely experience any other difficulties. Another person may experience ten difficulties all the time. Using the current scale however, these respondents would receive the same score. It is thus concluded that further research is needed to examine the variables that moderate the relationship between work difficulties and job satisfaction. This research needs to measure work difficulties using an open-ended format, where respondents list their top five difficulties at work and indicate how often they face each one. This scale is not expected to be excessively taxing and can be administered to a general sample of employees. Using this scale, it is expected that instrumental support will moderate the effect of controllable difficulties and emotional support will moderate the effect of uncontrollable difficulties. 4.2.4.3 Summary There was no support for the moderating role of the control strategies on the relationship between work difficulties and job satisfaction, however there was some support for social support at work. As primary and secondary control did not act as moderators, the model of job satisfaction was revised to examine the successfulness of the control strategies. Although there was only minimal support for the moderating role of social support, the indirect relationship is retained in the model of job satisfaction as the operationalisation of work difficulties was criticised.

353

4.3

Revised Model of Job Satisfaction

This discussion has combined the results from three studies to develop a revised model of job satisfaction (refer to Figure 11). The revisions are based on the current results, findings from other research, or are purely speculative. The bolded arrows and variables represent changes made to the model. The first proposal that personality influences the use of the control strategies is based on past studies of coping and personality. Although further research needs to be conducted to determine which personality variables predict secondary control, it is proposed that extroversion is positively related to primary control. Past research has also demonstrated that extroversion is also positively related to job satisfaction and life satisfaction, and that neuroticism is negatively related to job satisfaction and life satisfaction. Thus, personality influences the control strategies, job satisfaction, and life satisfaction. Primary and secondary control strategies are no longer directly related to job satisfaction; rather it is speculated that they are indirectly related to job satisfaction through the successfulness of the control strategies. It is expected that if employees successfully implement the strategies, they will report higher job satisfaction. In addition to the successfulness of the control strategies, job autonomy and social support at work are expected to be positively related to job satisfaction. Job satisfaction is also expected to be reciprocally related to life satisfaction. These relationships have all been demonstrated in the current findings. Based on the findings of study three, controllable and uncontrollable difficulties are expected to be negatively related to job satisfaction. The relationship

354

between work difficulties and job satisfaction is hypothesised to be moderated by the successfulness of the control strategies and social support at work. Specifically, it is expected that the successfulness of primary control moderates the effect of controllable difficulties, and the successfulness of secondary control moderates the effect of uncontrollable difficulties. The effect of controllable work difficulties on job satisfaction is expected to be less when primary control is successful, and the effect of uncontrollable difficulties on job satisfaction is expected to be less when secondary control is successful. Social support at work is also expected to moderate the effect of work difficulties on job satisfaction. As demonstrated in study three, instrumental support is expected to moderate the effect of controllable difficulties on job satisfaction. It is also hypothesised that emotional support will moderate the effect of uncontrollable difficulties on job satisfaction. Although this proposal was not supported in study three, it is expected that when a new measure of work difficulties is used, it will be supported.

355

Figure 11- Revised Model of Job Satisfaction

Secondary Control Instrumental Support

Primary Control

Success of

Success of PC Controllable Difficulties Controllable Diff x Instrumental Support

Controllable Diff x Success PC

Personalit

Life Satisfaction

Job Satisfaction

Job Autonomy

Uncontrol Diff x Success of SC

Uncontrollable Difficulties

Uncontrol Diff x Emotional Support

Success of PC

Success of SC

Secondary Control Primary Control

Emotional Support

356

4.4

Conclusion

This thesis extended the job demand-control model (Karasek & Theorell, 1990), offering an alternative explanation for the relationship between job autonomy and job satisfaction. A model of job satisfaction was developed which included job autonomy, primary and secondary control, life satisfaction, work difficulties and social support at work. The major proposal of this model was that job autonomy influences the use and adaptiveness of primary and secondary control strategies. Empirical testing of the model demonstrated that primary and secondary control did not mediate the relationship between job autonomy and job satisfaction. Employees reported using more primary control than secondary control for controllable and uncontrollable difficulties. Furthermore, primary control was more adaptive than secondary control for both types of difficulties. Using these findings, a revised model of job satisfaction was developed. This model proposes that rather than job autonomy, personality influences the use of the control strategies. Furthermore, it is proposed that the control strategies do not directly relate to job satisfaction, rather they are indirectly related through the successfulness of the control strategies. In addition to these variables, job autonomy, social support at work, life satisfaction and work difficulties continue to be included as predictors of job satisfaction.

357

4.5

Final Word

This study developed a model of job satisfaction that offered an alternative explanation to Karasek and Theorell (1990) for the relationship between job autonomy and job satisfaction. Based on the life span theory of control (Heckhausen & Schulz, 1995) and the discrimination model (Thompson et al., 1998), it was proposed that employees with high job autonomy reported high job satisfaction because they relied on more primary control and less secondary control strategies than employees with low job autonomy. These proposals were not supported, as primary control was the most commonly used and most adaptive strategy for controllable and uncontrollable difficulties. These findings suggest that the serenity prayer might best be changed to Grant me the strength to change the things I can. and the things I cannot.

358

4.6

References

Abouserie, R. (1996). Stress, coping strategies and job satisfaction in university academic staff. Educational Psychology, 16, 49-57. Agho, A.O., Price, J.L., & Mueller, C.W. (1992). Discriminant validity of measures of job satisfaction, positive affectivity and negative affectivity. Journal of Occupational and Organizational Psychology, 65, 185-196. Alderfer, C.P. (1969). An empirical test of a new theory of human needs. Organizational Behavior and Human Performance, 4, 142-175. Aldwin, C.M., & Revenson, T.A. (1987). Does coping help? A reexamination of the relation between coping and mental health. Journal of Personality and Social Psychology, 53, 337-348. Algera, J.A. (1980). Kenmerken van werk. Unpublished Ph.D. Thesis, University of Leiden. Ambrose, M.L., & Kulik, C.T. (1999). Old friends, new faces: Motivation research in the 1990s. Journal of Management, 25, 231-237. Armstrong, T.B. (1971). Job content and context factors related to satisfaction for different occupational levels. Journal of Applied Psychology, 55, 57-65. Arnold, H.J., & House, R.J. (1980). Methodological and substantive extensions to the Job Characteristics Model of motivation. Organizational Behavior and Human Decision Processes, 25, 161-183. Arvey, R.D., Bouchard, T.J., Segal, N.L., & Abraham, L.M. (1989). Job satisfaction: Environmental and genetic components. Journal of Applied Psychology, 74, 187-192.

359

Arvey, R.D., McCall, B.P., Bouchard, T.J., Taubman, P., & Cavanaugh, M.A. (1994). Genetic influences on job satisfaction and work values. Personality and Individual Differences, 17, 21-33. Ashford, S.J. (1988). Individual strategies for coping with stress during organisational transitions. Journal of Applied Behavioral Science, 24, 19-36. Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182. Beehr, T.A. (1985). The role of social support in coping with organizational stress. In T.A. Beehr., & R.S. Bhagat (Eds.), Human Stress and Cognition in Organizations: An Integrated Perspective. (pp. 375-398). New York: Wiley Beutell, N.L., & Wittig-Berman, U. (1999). Predictors of work-family conflict and satisfaction with family, job, career, and life. Psychological Reports, 85, 893-903. Bobko, P. (2001). Correlation and Regression: Applications for IndustrialOrganizational Psychology and Management. London: Sage. Boey, K.W. (1998). Coping and family relationships in stress resistance: A study of job satisfaction of nurses in Singapore. International Journal of Nursing Studies, 35, 353-361. Bogg, J., & Cooper, C. (1995). Job satisfaction, mental health, and occupational stress among senior civil servants. Human Relations, 48, 327-341. Boomsma, A. (1983). On the robustness of LISREL (maximum likelihood estimation) against small sample sizes of nonnormality. Unpublished Ph.D. thesis, University of Groningen, The Netherlands.

360

Boonzaier, B., Ficker, B., & Rust, B. (2001). A review of research on the Job Characteristics Model and the attendant job diagnostic survey. South African Journal of Business Management, 32, 11-35. Bowman, G.D., & Stern, M. (1995). Adjustment to occupational stress: The relationship of perceived control to effectiveness of coping strategies. Journal of Counseling Psychology, 42, 294-303. Brayfield, A.H., & Rothe, H.F. (1951). An index of job satisfaction. Journal of Applied Psychology, 35, 307-311. Breaugh, J.A. (1989). The Work Autonomy Scales: Additional validity evidence. Human Relations, 42, 1033-1056. Breaugh, J.A. (1998). The development of a new measure of global work autonomy. Educational and Psychological Measurement, 58, 119-129. Brebner, J. (2001). Personality and stress coping. Personality and Individual Differences, 31, 317-327. Brenner, V.C., Carmack, C.W., & Weinstein, M.G. (1971). An empirical test of the motivation-hygiene theory. Journal of Accounting Research, 9, 359-366. Burger, J.M., & Cooper, H.M. (1979). The desirability of control. Motivation and Emotion, 3, 381-393. Burke, R.J., & Greenglass, E.R. (2000). Hospital restructuring and nursing staff well-being: The role of coping. International Journal of Stress Management, 7, 49-59. Cahill, J. (1998). Stories of success and survival: The role of primary and secondary control mechanisms in the subjective well-being of people on methadone

361

maintenance treatment programs. Unpublished Honours thesis, Deakin University, Melbourne, Australia. Cammann, C., Fichman, M., Jenkins, D., & Klesh, J. (1979). The Michigan Organizational Assessment Questionnaire. Unpublished manuscript, University of Michigan: Ann Arbor. Campbell, J.P., & Pritchard, R.D. (1976). Motivation theory in Industrial and Organizational Psychology. In M. Dunnette (Ed.), Handbook of Industrial and Organizational Psychology. (pp 63-130). Chicago: Rand McNally. Caplan, R.D., Cobb, S., French, J.R.P. Jr., Harrison, R.U., & Pinneau, S.R. Jr. (1975). Job Demands and Worker Health. U.S. Department of Health, Education, and Welfare Publication No. 75-160. Washington D.C.: U.S. Government Printing Office, The Institute for Social Research. Carson, K.D., Lanier, P.A., & Carson, P.P. (2001). A glimpse inside the ivory tower: A cross-sectional comparison of work orientations in academia. International Journal of Public Administration, 24, 479-492. Carver, C.S., Scheier, M.F., & Weintraub, J.K. (1989). Assessing coping strategies: A theoretically based approach. Journal of Personality and Social Psychology, 56, 267-283. Champoux, J.E. (1980). A three sample test of some extensions to the Job Characteristics Model of work motivation. Academy of Management Journal, 23, 466478.

362

Chay, Y.W. (1993). Social support, individual differences and well-being: a study of small business entrepreneurs and employees. Journal of Occupational and Organizational Psychology, 66, 285-302. Chipperfield, J.G., Perry, R.P., & Menec, V.H. (1999). Primary and secondary control-enhancing strategies. Journal of Aging and Health, 11, 517-539. Clark, A.E. (1996). Job satisfaction in Britain. British Journal of Industrial Relations, 34, 189-217. Clark, A.E., & Oswald, A.J. (1996). Satisfaction and comparison income. Journal of Public Economics, 69, 57-81. Coakes, S.J., & Steed, L.G. (1999). SPSS: Analysis without Anguish. Brisbane: John Wiley. Cohen, J., & Cohen, P. (1983). Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. Hillsdale, N.J: Lawrence Erlbaum. Cohen, S., & Kamarack, T., Mermelstein, R., & Hoberman, H.M. (1985). Measuring the functional components of social support. In I.G. Sarason., & B.R. Sarason (Eds.), Social Support: Theory, Research and Applications. The Hague, Netherlands: Nijhoff. Cole, M. (1989). The politics of stress in teaching. In M.Cole., & S. Walker (Eds.), Teaching and Stress. (pp.161-170). Milton Keynes: Open University Press. Constantinople, A. (1967). Perceived instrumentality of the college as a measure of attitudes towards college. Journal of Personality and Social Psychology, 5, 196-201.

363

Conway, V.J., & Terry, D.J. (1992). Appraised controllability as a moderator of the effectiveness of different coping strategies: A test of the goodness of fit hypothesis. Australian Journal of Psychology, 44, 1-7. Cooper, C.L., Sloan, S., & Williams, S. (1988). Occupational Stress Indicator. Windsor: NFER-Nelson. Costa, P.T. Jr., & McCrae, R.R. (1985). The NEO Personality Inventory Manual. Odessa, FL: Psychological Assessment Resources. Costa, P.T. Jr., & McCrae, R.R. (1989). Personality as a life long determinant of wellbeing. In L.Z. Malatesta., & C.E. Izzard (Eds.), Emotion in Adult Development. (p.141-157). Beverly Hills, CA: Sage. Costa, P.T. Jr., & McCrae, R.R. (1992). NEO PI-R professional manual. Odessa, FL: Psychological Assessment Resources. Costa, P.T. Jr., Somerfield, M.R., & McCrae, R.R. (1996). Personality and coping: A reconceptualisation. In M. Zeidner., & N.S. Endler (Eds.), Handbook of Coping: Theory, Research, Applications. New York: Wiley Cousins, R. (2001). Predicting subjective quality of life: The contributions of personality and perceived control. Unpublished Doctorate thesis, Deakin University, Melbourne, Australia. Cramer, D. (1995). Life and job satisfaction: A two-wave panel study. The Journal of Psychology, 129, 261-268. Cummins, R.A. (1995). On the trail of the gold standard for subjective wellbeing. Social Indicators Research, 35, 179-200.

364

Cummins, R.A. (1996). The domain of life satisfaction: An attempt to order chaos. Social Indicators Research, 38, 303-328. Cummins, R.A. (1997). Comprehensive Quality of Life Scale-Adult Manual (ComQol-A5). Melbourne, Australia: Deakin University. Cummins, R.A. (2000a). Objective and subjective quality of life: An interactive model. Social Indicators Research, 52, 55-72. Cummins, R.A. (2000b). Normative life satisfaction: Measurement issues and a homeostatic model. In B. Zumbo (Ed.), Social Indicators and Quality of Life Research Methods: Methodological Development and Issues. Amsterdam: Klumer. Cummins, R.A. (2002). Caveats to the Comprehensive Quality of Life Scale. http://acqol.deakin.edu.au. Cummins, R.A., Eckersley, R., Pallant, J., Van Vugt, J., Shelley, J., Pusey, M., & Misajon, R. (2001). The Australian Unity Index of Well-Being: Survey 1: Report 1. (on-line). Available: http://acqol.deakin.edu.au. Cummins, R.A., & Nistico, H. (in press). Maintaining life satisfaction: The role of positive cognitive bias. Journal of Happiness Studies DeCharms, R. (1968). Personal Causation. New York: Academic Press. Deci, E.L., & Ryan, R.M. (1985). Intrinsic Motivation and Self-Determination in Human Behaviour. New York: Plenum Press. de Jonge, J., Breukelen, G.J.P., Landeweerd, J.A., & Nijhuis, F.J.N. (1999a). Comparing group and individual level assessments of job characteristics in testing the Job Demand-Control Model: A multilevel approach. Human Relations, 52, 95-97.

365

de Jonge, J., Mulder, M.J.G.P., & Nijhuis, F.J.N. (1999b). The incorporation of different demand concepts in the job demand-control model: Effects of health care professionals. Social Science and Medicine, 48, 1149-1160. DeNeve, K.M. (1999). Happy as an extroverted clam? The role of personality for subjective well-being. Psychological Science, 8, 141-144. De Rijk, A.E., Le Blanc, P., Schaufeli, W.B., & de Jonge, J. (1998). Active coping and need for control as moderators of the job demand-control model: Effects on burnout. Journal of Occupational and Organizational Psychology, 71, 1-18. Diener, E., Suh, E.M., Lucas, R.E., & Smith, H.L. (1999). Subjective wellbeing: Three decades of progress. Psychological Bulletin, 125, 276-302. Dollard, M.F., Winefield, H.R., Winefield, A.H., & Jonge, J.D. (2000). Psychosocial service workers: A test of the demand control-support model. Journal of Occupational and Organizational Psychology, 73. 501-510. Doran, L.I., Stone, V.K., Brief, A.P., & George, J.M. (1991). Behavioral intentions as predictors of job attitudes: The role of economic choice. Journal of Applied Psychology, 76, 40-45. Ducharme, L.J., & Martin, J. (2000). Unrewarding work, co-worker support, and job satisfaction. Work and Occupations, 27, 223-244. Dwyer, D.H., & Ganster, D.C. (1991). The effects of job demands and control in employee attendance and satisfaction. Journal of Organizational Behavior, 12, 595608. Edwards, A.L. (1959). Edwards Personal Preference Schedule. New York: Psychological Corporation.

366

Edwards, J.R., & ONeill, R. (1999). The construct validity of scores on the Ways of Coping Questionnaire: Confirmatory analysis of alternative factor structures. Educational and Psychological Measurement, 58, 955-983. Evans, B.K., & Fischer, D.G. (1992). A hierarchical model of participatory decision-making, job autonomy and perceived control. Human Relations, 45, 11691190. Evans, M.G. (1991). The problem of analyzing multiplicative composites: interactions revisited. American Psychologist, 46, 6-15. Ewen, R.B. (1964). Some determinants of job satisfaction: A study of the generalisability of Herzbergs theory. Journal of Applied Psychology, 48, 161-163. Fay, D., & Frese, M. (2000). Self-starting behavior at work: Toward a theory of personal initiative. In J. Heckhausen (Ed.), Motivational Psychology of Human Development. Amsterdam: Elservier Science. Felce, D., & Perry, J. (1995). Quality of life: Its definition and measurement. Research in Developmental Disabilities, 16, 51-74. Fenlason, K.J., & Beehr, T.A. (1994). Social support and occupational stress: Effects of talking to others. Journal of Organizational Behavior, 15, 157-175. Ferris, K.R. (1977). A test of the expectancy theory of motivation in an accounting environment. The Accounting Review, 3, 605-615. Finlay, W., Martin, J.K., Roman, P.M., & Blum, T.C. (1995). Organizational structure and job satisfaction: Do bureaucratic organizations produce more satisfied employees? Administration and Society, 27, 427-451.

367

Fisher, C.D. (2000). Mood and emotions while working: missing pieces of job satisfaction? Journal of Organizational Behavior, 21, 185-202. Fisher, S. (1994). Stress in Academic Life: The Mental Assembly Line. Buckingham: Open University Press. Fletcher, B.C., & Jones, F. (1993). A refutation of Karaseks DemandDiscretion Model of Occupational Stress with a range of dependent measures. Journal of Organizational Behavior, 14, 319-330. Folkman, S., Chesney, M.A., Cooke, M., & Bocccellari, A. (1994). Caregiver burden in the HIV-positive and HIV-negative partners of men with AIDS. Journal of Consulting and Clinical Psychology, 62, 746-756. Folkman, S., & Lazarus, R.S. (1980). An analysis of coping in a middle-aged community sample. Journal of Health and Social Behavior, 21, 219-239. Folkman, S., & Lazarus, R.S. (1985). If it changes, it must be a process: Study of emotion and coping during three stages of a college examination. Journal of Personality and Social Psychology, 48, 150-170. Folkman, S., Lazarus, R.S., Dunkel-Schetter, C., DeLongis, A., & Gruen, R.J. (1986). Dynamics of a stressful encounter: Cognitive appraisal, coping and encounter outcomes. Journal of Personality and Social Psychology, 50, 992-1003. Forsythe, C.J., & Compas, B.E. (1987). Interaction of cognitive appraisals of stressful events and coping: Testing the goodness of fit hypothesis. Cognitive Therapy and Research, 11, 473-485.

368

Fox, M.L., Dwyer, D.J., & Ganster, D.C. (1993). Effects of stressful job demands and control on physiological and attitudinal outcomes in a hospital setting. Academy of Management Journal, 36, 289-319. Fox, S., & Feldman, G. (1988). Attention state and critical psychological states as mediators between job dimensions and job outcomes. Human Relations, 41, 229-245. Fried, Y. (1991). Meta-analytic comparisons of the Job Diagnostic Survey and Job Characteristics Inventory as correlates of work satisfaction and performance. Journal of Applied Psychology, 76, 690-697. Fried, Y., & Ferris, G.R. (1987). The validity of the Job Characteristics Model: A review and meta-analysis. Personnel Psychology, 40, 287-322. Fried, Y., & Ferris, G.R. (1991). The dimensionality of job characteristics: Some neglected issues. Journal of Applied Psychology, 71, 419-426. Friedman, L.C., Nelson, D.V., Baer, P.E., Lane, M., Smith, F.E., & Dworkin, R.J. (1992). The relationship of dispositional optimism, daily life stress, and domestic environment to coping methods used by cancer patients. Journal of Behavioral Medicine, 15, 127-141. Friesen, D., Holdaway, E.A., & Rice, A.W. (1983). Satisfaction of school principals with their work. Educational Administration Quarterly, 19, 35-58. Frone, M.R., Russell, M., & Cooper, M.L. (1994). Relationship between job and family satisfaction: Causal or noncausal covariation? Journal of Management, 20, 565-579.

369

Fung-kam, L. (1998). Job satisfaction and autonomy of Hong Kong registered nurses. Journal of Advanced Nursing, 27, 355-364. Ganster, D.C. (1989). Worker control and well-being: A review of research in the workplace. In S. Sauter., J. Hurrell., & C. Cooper (Eds.), Job Control and Worker Health. (pp. 3-24). Chichester, U.K: Wiley. Ganster, D.C., Dwyer, D.J., & Fox, M.L. (2001). Explaining employees health care costs: A prospective examination of stressful job demands, personal control, and physiological reactivity. Journal of Applied Psychology, 86, 954-964. Ganster, D.C., & Fusilier, M.R. (1989). Control in the workplace. In C.L. Cooper., & I. Robertson (Eds.), International Review of Industrial and Organizational Psychology. (pp. 235-280). London: John Wiley Gardner, G. (1977). Is there a valid test of Herzbergs two-factor theory? Journal of Occupational Psychology, 50, 197-204. Gaziel, H.H. (1989). Determinants of perceived deficiency of autonomy among elementary school administrators. Social Behavior and Personality, 17, 57-66. Geyer, P.D., & Daly, J.P. (1998). Predicting job satisfaction for relocated workers: Interaction of relocating consequences and employee age. The Journal of Psychology, 132, 417-427. Gillet, B., & Schwab, D.P. (1975). Convergent and discriminant validities of corresponding Job Descriptive Index and Minnesota Satisfaction Questionnaire Scales. Journal of Applied Psychology, 60, 313-317. Graham, W.K., & Balloun, J. (1973). An empirical test of Maslows need hierarchy theory. Journal of Humanistic Psychology, 13, 97-108.

370

Grant, G.P., & Murray, C.E. (1999). Teaching in America: The Slow Revolution. Cambridge, Mass: Harvard University Press. Gunthert, K.C., Cohen, L.H., & Armeli, S. (1999). The role of neuroticism in daily stress and coping. Journal of Personality and Social Psychology, 5, 1087-1100. Hackman, J.R., & Oldham, G.R. (1975). Development of the Job Diagnostic Survey. Journal of Applied Psychology, 60, 159-170. Hackman, J.R., & Oldham, G.R. (1976). Motivation through the design of work: Test of a theory. Organizational Behavior and Human Performance, 16, 250279. Hackman, J.R., & Oldham, G.R. (1980). Work Redesign. Reading, Mass: Addison-Wesley. Hall, D.T., & Nougaim, K.E. (1968). An examination of Maslows need hierarchy theory in an organizational setting. Organizational Behavior and Human Performance, 3, 12-35. Halliday, C.A., & Graham, S. (2000). If I get locked up, I get locked up: Secondary control and adjustment among juvenile offenders. Personality and Social Psychology Bulletin, 26, 548-559. Hallqvist, J., Diderichsen, F., Theorell, T., Reuterwall, C., & Ahlbom, A. (1998). Is the effect of job strain on myocardial infarction risk due to interaction between high psychological demands and low decision latitude? Results from Stockholm Heart Epidemiology Program (SHEEP). Social Science and Medicine, 46, 1405-1415.

371

Hart, P.M. (1999). Predicting employee life satisfaction: A coherent model of personality, work and non-work experiences, and domain satisfaction. Journal of Applied Psychology, 84, 564-584. Harwood, M.K., & Rice, R.W. (1992). An examination of the referent selection processes underlying job satisfaction. Social Indicators Research, 27, 1-39. Hatfield, J.D., Robinson, R.B., & Huseman, R.C. (1985). An empirical evaluation of a test for assessing job satisfaction. Psychological Reports, 56, 39-45. Hawkins, M.J., Hawkins, W.E., & Ryan, E.R. (1989). Self-actualisation as related to age of faculty members at a large Midwestern University. Psychological Reports, 65, 1120-1122. Heckhausen, J., & Schulz, R. (1995). A life span theory of control. Psychological Review, 102, 284-304. Heckhausen, J., Schulz, R., & Wrosch, C. (1997). Technical Report: Optimization in primary and secondary control: OPS Scales, a measurement instrument for general and domain specific strategies of control and development regulation. Berlin: Max Planck Institute for Human Development. Heeps, L., Croft, C., & Cummins, R.A. (2000). The Primary and Secondary Control Scale (2nd ed). Melbourne: Deakin University. Herzberg, F. (1966). Work and the Nature of Man. Cleveland: World Publ Co. Herzberg, F., Mausner, B., & Snyderman, B.B. (1959). The Motivation to Work. New York: Wiley.

372

Herzberg, F., Mausner, B., & Snyderman, B.B. (1993). The Motivation to Work. New Brunswick, N.J: Transaction Publishers. Highhouse, S., & Becker, A.S. (1993). Facet measures and global job satisfaction. Journal of Business and Psychology, 8, 117-127. Hill, M.D. (1986). A theoretical analysis of faculty job satisfaction/dissatisfaction. Educational Research Quarterly, 10, 36-44. Himle, D.P., & Jayaratne, S. (1991). Buffering effects of four social support types on burnout among social workers. Social Work Research and Abstracts, 27, 2228. Hirschfeld, R.R. (2000). Does revising the intrinsic and extrinsic subscales of the Minnesota Satisfaction Questionnaire short form make a difference? Educational and Psychological Measurement, 60, 255-271. Hoppock, R. (1935). Job Satisfaction. New York: Harper. Howard, J.L., & Frink, D.D. (1996). The effects of organizational restructure on employee satisfaction. Group and Organization Management, 21, 278-303. Iaffaldano, M.T., & Muchinsky, P.M. (1985). Job satisfaction and job performance: A meta-analysis. Human Relations, 44, 287-307. Ironson, G.H., Smith, P.C., Brannick, M.T., Gibson, W.M., & Paul, K.B. (1989). Construction of a job in general scale: A comparison of global, composite, and specific measures. Journal of Applied Psychology, 74, 193-200. Iverson, R.D., & Maguire, C. (2000). The relationship between job and life satisfaction: Evidence from a remote mining community. Human Relations, 53, 807813.

373

Jansen, P.G.M., Kerkstra, A., Abud-Saad, H.H., & Van der Zee, J. (1996). The effects of job characteristics and individual characteristics of job satisfaction and burnout in community nursing. International Journal of Nursing Studies, 33, 407-421. Johnson, J.V., & Hall, E.M. (1988). Job strain, work place social support, and cardiovascular disease: a cross-sectional study of a random sample of the Swedish working population. American Journal of Public Health, 78, 1336-1342. Johnson, J.V., & Hall, E.M. (1994). Social support in the work environment and cardiovascular disease. In S.A. Schumaker., & S.M. Czajkowski (Eds.), Social Support and Cardiovascular Disease. Plenum Series in Behavioral Psychophysiology and Medicine. (pp. 145-166.) New York: Plenum Press. Johnson, J.V., Hall, E.M., & Theorell, T. (1989). Combined effects of job strain and social isolation on cardiovascular disease morbidity and mortality in a random sample of the Swedish male working population. Scandinavian Journal of Work, Environment and Health, 15, 271-279. Johnson, S.M., Smith, P.C., & Tucker, S.M. (1982). Response format of the Job Descriptive Index: Assessment of reliability and validity by the multitraitmultimethod matrix. Journal of Applied Psychology, 67, 500-505. Judge, T.A., & Bono, J.E. (2001). Relationship of core self-evaluations traitsSelf-esteem, generalised self-efficacy, locus of control and emotional stability with job satisfaction and job performance: A meta-analysis. Journal of Applied Psychology, 86, 80-93.

374

Judge, T.A., Bono, J.E., & Locke, E.A. (2000). Personality and job satisfaction: The mediating role of job characteristics. Journal of Applied Psychology, 85, 237-249. Judge, T.A., & Locke, E.A. (1993). Effect of dysfunctional thought processes on subjective well-being and job satisfaction. Journal of Applied Psychology, 78, 475490. Judge, T.A., Locke, E.A., Durham, C.C., & Kluger, A.N. (1998). Dispositional effects on job and life satisfaction: The role of cognitive evaluations. Journal of Applied Psychology, 83, 17-34. Judge, T.A., & Watanabe, S. (1994). Individual differences in the nature of the relationship between job and life satisfaction. Journal of Occupational and Organizational Psychology, 67, 101-107. Kandel, D.B., Davies, M., & Raveis, V.H. (1985). The stressfulness of daily social roles for women: Marital, occupational and household roles. Journal of Health and Social Behavior, 26, 64-78. Karasek, R.A. Jr. (1979). Job demands, job decision latitude, and mental strain: Implications for job redesign. Administrative Science Quarterly, 24, 335-357. Karasek, R.A., Brisson, C., Kawakami, N., Houtman, I., Bongers, P., & Amick, B. (1998). The Job Content Questionnaire (JCQ): An instrument for internationally comparative assessments of psychosocial job characteristics. Journal of Occupational Health Psychology, 3, 322-355. Karasek, R.A., & Theorell, T. (1990). Healthy work: Stress, productivity and the Reconstruction of Working Life. New York: Basic Books.

375

Karasek, R.A., Triantis, K.P., & Chaudry, S.S. (1982). Co-worker and supervisor support as moderators of associations between task characteristics and mental strain. Journal of Occupational Behaviour, 3, 181-200. Kelly, J. (1992). Does job re-design theory explain job re-design outcomes? Human Relations, 45, 753-775. Keppel, G. (1991). Design and analysis: A Researchers Handbook (3rd ed.). New Jersey: Prentice Hall. King, N. (1970). Clarification and evaluation of the two factor theory of job satisfaction. Psychological Bulletin, 74, 18-31. Klecker, B.M., & Loadman, W.E. (1999). Male elementary school teachers ratings of job satisfaction by years of teaching experience. Education, 119, 504-604 Koeske, G.F., Kirk, S.A., & Koeske, R.D. (1993). Coping with job stress: Which strategies work best? Journal of Occupational and Organizational Psychology, 66, 319-335. Kohn, P.M., Hay, B.D., & Legere, J.J. (1994). Hassles, coping styles, and negative well-being. Personality and Individual Differences, 17, 169-179. Kunin, T. (1955). The construction of a new type of attitude measure. Personnel Psychology, 8, 65-77. Lahey, K.E., & Vihtelic, J.L. (2000). Finance faculty demographics, career history, diversity, and job satisfaction. Financial Practice and Education, 10, 111-123. Landeweerd, J.A., & Boumans, N.P.G. (1994). The effect of work dimensions and need for autonomy on nurses work satisfaction and health. Journal of Occupational and Organizational Psychology, 67, 207-218.

376

Landry, M.B. (2000). The effects of life satisfaction and job satisfaction on reference librarians and their work. Reference and user Services Quarterly, 40, 166178. Landsbergis, P.A., Schnall, P.L., Deitz, D., Friedman, R., & Pickering, T. (1992). The patterning of psychological attributes and distress by job strain and social support in a sample of working men. Journal of Behavioral Medicine, 15, 379405. LaRocco, J.M., House, J.S., & French, J.R.P. (1980). Social support, occupational stress, and helath. Journal of Health and Social Behavior, 21, 202. Laschinger, H.K.S., Finegan, J., & Shamian, J. (2001). Promoting nurses health: Effect of empowerment on job strain and work satisfaction. Nursing Economics, 19, 42-52. Latack, J.C. (1986). Coping with job stress: Measures and future direction for scale development. Journal of Applied Psychology, 71, 377-385. Lawler, E.E., & Suttle, J.L. (1972). A causal correlational test of the need hierarchy concept. Organizational Behavior and Human Performance, 7, 265-287. Lazarus, R.S., & Folkman, S. (1984). Stress, Appraisal and Coping. New York: Springer. Leong, F.T.L., & Dollinger, S.J. (1991). NEO Personality Inventory. In D.J. Keyser., & R.C. Sweetland (Eds.), Test Critiques, Vol VIII. (p.527-539). Texas: ProEd

377

Leung, K. (1997). Relationships among satisfaction, commitment and performance: A group level analysis. Applied Psychology: An International Review, 46, 199-205. Leung, T., Siu, O., & Spector, P.E. (2000). Faculty stressors, job satisfaction and psychological distress among university teachers in Hong Kong: The role of locus of control. International Journal of Stress Management, 7, 121-138 Locke, E.A. (1969). What is job satisfaction? Organizational Behavior and Human Performance, 4, 309-336. Locke, E.A. (1976). Nature and causes of job satisfaction. In M. Dunnette (Ed.), Handbook of Industrial and Organizational Psychology. (pp. 1297-1350). Chicago, IL: Rand McNally. Loher, B.T., Noe, R.A., Moeller, N.L., & Fitzgerald, M.P. (1985). A metaanalysis of the relation of job characteristics to job satisfaction. Journal of Applied Psychology, 70, 280-289. Lu, L. (1999). Work motivation, job stress and employees well-being. Journal of Applied Management Studies, 8, 61-69. Lykken, D., & Tellegen, A. (1996). Happiness is a stochastic phenomenon. Psychological Science, 7, 186-189. Ma, X., & MacMillan, R.B. (1999). Influences of workplace conditions on teachers job satisfaction. Journal of Educational Research, 93, 39-47. Maher, E., & Cummins, R.A. (2001). Subjective quality of life, perceived control, and dispositional optimism among older people. Australasian Journal on Aging, 20, 139-146.

378

Maher, E., & Cummins, R.A. (2002). Situation Specific Primary and Secondary Control Scale. Melbourne: Deakin University. Maher, E., Misajon, R., Heeps, L., & Cummins, R.A. (2001). Primary Control and Secondary Control Scale (4th ed.). Melbourne: Deakin University. Mannheim, B., Baruch, Y., & Tal, J. (1997). Alternative models for antecedents and outcomes of work centrality and job satisfaction of high-tech personnel. Human Relations, 50, 1537-1562. Marriott, A., & Sexton, L. (1994). Components of job satisfaction in psychiatric social workers. Health and Social Work, 19, 199-206. Maslow, A.H. (1954). Motivation and Personality. New York: Harper. Maslow, A.H. (1970). Motivation and Personality. New York: Harper & Row. McConatha, J.T., & Huba, H.M. (1999). Primary, secondary and emotional control across adulthood. Current Psychology, 18, 164-170. McCrae, R.R., & Costa, P.T. (1991). Adding Liebe und Arbeit: The full FiveFactor Model and well-being. Personality and Social Psychology Bulletin, 17, 227232. McCrae, R.R., & Costa, P.T. Jr. (1992). Discriminant validity of NEO-PIR Facet Scale. Educational Psychological Measurement, 52, 229-237. Melamed, S., Kushnir, T., & Meir, E.I. (1991). Attentuating the impact of job demands: Additive and interactive effects of perceived control and social support. Journal of Vocational Behavior, 39, 40-53.

379

Michalos, A.C. (1985). Multiple Discrepancies Theory (MDT). Social Indicators Research, 16, 347-413. Miles, E.W., Patrick, S.L., & King, W.C. (1996). Job level as a systemic variable in predicting the relationship between supervisory communication and job satisfaction. Journal of Occupational and Organizational Psychology, 69, 277-285 Misajon, R. (2002). The Homeostatic Mechanism: Subjective Quality of Life and Chronic Pain. Unpublished Ph.D. Thesis, Deakin University, Melbourne. Misajon, R., & Cummins, R.A. (in press). Subjective quality of life, control and the impact of chronic illness. Arthritis Care and Research. Mitchell, T.R. (1974). Expectancy models of job satisfaction, occupational preference and effort. Psychological Bulletin, 81, 1053-1077. Moorman, R.H. (1993). The influence of cognitive and affective based job satisfaction measures on the relationship between satisfaction and organizational citizenship behavior. Human Relations, 46, 759-776. Munro, L., Rodwell, J., & Harding, L. (1998). Assessing occupational stress in psychiatric nurses using the full Job Strain Model: The value of social support to nurses. International Journal of Nursing Studies, 35, 339-345. Narayanan, L., Menon, S., & Spector, P.E. (1999). Stress in the workplace: a comparison of gender and occupation. Journal of Organizational Behavior, 20, 63-73. Neher, A. (1991). Maslows theory of motivation: A critique. Journal of Humanistic Psychology, 31, 89-112.

380

Nicolle, P.G. (1994). The effect of work dimensions and need for autonomy on nurses work satisfaction and health. Journal of Occupational and Organizational Psychology, 64, 207-218. Niemann, Y.F., & Dovidio, J.F. (1998). Relationship of solo status, academic rank, and perceived distinctiveness to job satisfaction of racial/ethnic minorities. Journal of Applied Psychology, 83, 55-71. Norman, P., Collins, S., Conner, M., Martin, R., & Rance, J. (1995). Attributions, cognitions, and coping styles: Teleworkers reactions to work-related problems. Journal of Applied Social Psychology, 25, 117-128. ODriscoll, M.P., & Beehr, T.A. (2000). Moderating effects of perceived control and need for clarity on the relationship between role stressors and employee affective reactions. The Journal of Social Psychology, 140, 151-157. Olsen, D. (1993). Work satisfaction and stress in the first and third year of academics appointment. Journal of Higher Education, 64, 453-462. Osipow, S.H., Doty, R.E., & Spokane, A.R. (1985). Occupational stress, strain and coping across the life span. Journal of Vocational Behavior, 27, 98-108. Osowiecki, D., & Compas, B.E. (1998). Psychological adjustment to cancer: Control beliefs and coping in adult cancer patients. Cognitive Therapy and Research, 22, 483-499. Osowiecki, D.M., & Compas, B.E. (1999). A prospective study of coping, perceived control and psychological adjustment to breast cancer. Cognitive Therapy and Research, 23, 169-180.

381

Parahoo, K., & Barr, O. (1994). Job satisfaction of community nurses working with people with a mental handicap. Journal of Advanced Nursing, 20, 1046-1055. Park, C.L., Folkman, S., & Bostrom, A. (2001). Appraisals of controllability and coping in caregivers and HIV+ men: Testing the goodness-of-fit hypothesis. Journal of Consulting and Clinical Psychology, 69, 481-488. Parker, S.K., & Sprigg, C.A. (1999). Minimising strain and maximising learning: The role of job demands, job control and proactive personality. Journal of Applied Psychology, 84, 925-939. Parkes, K.R. (1984). Locus of control, cognitive appraisal and coping in stressful episodes. Journal of Personality and Social Psychology, 46, 655-668. Parkes, K.R. (1989). Personal control in an occupational context. In A. Steptoe., & A. Appels (Eds.), Stress, Personal Control, and Health (pp. 21-47). Chichester: Wiley. Parkes, K.R. (1990). Coping, negative affectivity and the work environment: Additive and interactive predictors of mental health. Journal of Applied Psychology, 75, 399-409. Parkes, K.R. (1994). Personality and coping as moderators of work stress process: Models, methods and measures. Work and Stress, 8, 110-129. Parkes, K.R., & von Rabenau, C. (1993). Work characteristics and wellbeing among psychiatric health care staff. Journal of Community and Applied Social Psychology, 3, 243-260. Parsons, M.B. (1998). A review of procedural acceptability in organizational behavior management. Journal of Organizational Behavior Management, 18, 173-190.

382

Payne, R., & Fletcher, B.C. (1983). Job demands, supports, and constraints as predictors of psychological strain among schoolteachers. Journal of Vocational Behavior, 22, 136-147. Pearson, C.A.L., & Chong, J. (1997). Contribution of job content and social information in organizational commitment and job satisfaction: an exploration in a Malaysian nursing context. Journal of Occupational and Organizational Psychology, 70, 357Pelfrene, E., Vlerick, P., Mak, R.P., De Smets, P., Kornitzers, M., & De Backer, G. (2001). Scale reliability and validity of the Karasek Job Demand Control Support model in the Belstress study. Work and Stress, 15, 297-313. Perrewe, P.L., & Zellars, K.I. (1999). An examination of attributions and emotions in the transactional approach to the organizational stress process. Journal of Organizational Behavior, 20, 739-752. Porter, L.W. (1961). A study of perceived need satisfaction in bottom and middle management jobs. Journal of Applied Psychology, 45, 1-10. Porter, L.W. (1963). Job attitudes in management: II. Perceived importance of needs as a function of job level. Journal of Applied Psychology, 47, 141-148. Pritchard, R.D., & Sanders, M.S. (1973). The influence of valence, instrumentality, and expectancy on effort and performance. Journal of Applied Psychology, 57, 55-60. Pulakos, E.D., & Schmitt, N. (1983). A longitudinal study of a valence model approach for the prediction of job satisfaction of new employees. Journal of Applied Psychology, 68, 307-312.

383

Quinn, R.O., & Staines, G.L. (1979). The 1977 Quality of Employment Survey: Descriptive statistics with comparison data from the 1960-1970 and the 197273 surveys. University of Michigan: Ann Arbor. Rain, J.S., Lane, I.M., & Steiner, D.D. (1991). A current look at the job satisfaction/life satisfaction relationship: Review and future considerations. Human Relations, 44, 287-307. Reinharth, L., & Wahba, R.A. (1976). A test of alternative models of expectancy theory. Human Relations, 29, 257-272. Renn, R.W., & Vandenberg, R.J. (1995). The critical psychological states: An underrepresented component on Job Characteristics Model research. Journal of Management, 21, 279-303. Rice, R.W., McFarlin, D.B., & Bennett, D.,E. (1989). Standards of comparison and job satisfaction. Journal of Applied Psychology, 74, 591-598. Roberts, K.H., & Glick, W. (1981). The job characteristics approach to task design: A critical review. Journal of Applied Psychology, 66, 193-217. Roberts, K.H., Walter, G.A., & Miles, R.E. (1971), A factor analytic study of job satisfaction items designed to measure Maslow need categories. Personnel Psychology, 24, 205-220. Roberts, S.M. (1995). Applicability of the goodness of fit hypothesis to coping with daily hassles. Psychological Reports, 77, 943-954. Roberts, T.B. (1982). Comments on Mathess article. Journal of Humanistic Psychology, 22, 97-98.

384

Rodriguez, I., Bravo, M.J., Peiro, J.M., & Schaufeli, W. (2001). The demandscontrol-support model, locus of control and job dissatisfaction: a longitudinal study. Work and Stress, 15, 97-114. Rothbaum, F., Weisz, J.R., & Synder, S.S. (1982). Changing the world and changing the self: A two process model of perceived control. Journal of Personality and Social Psychology Bulletin, 42, 5-37. Roznowski, M. (1989). Examination of the measurement properties of the Job Descriptive Index with experimental items. Journal of Applied Psychology, 74, 805814. Russell, D.W., Altmaier, E., & Van Velzen, D. (1987). Job-related stress, social support, and burnout among classroom teachers. Journal of Applied Psychology, 72, 269-274. Ryan, E.M., & Deci, E.L. (2001). On happiness and human potentials: A review of research on hedonic and eudaimonic well-being. Annual Review of Psychology, 141-155. Saklofske, D.H., & Kelly, I.W. (1995). Coping and personality. Psychological Reports, 77, 481-482. Salancik, G.R., & Pfeffer, J. (1977). An examination of need satisfaction models of job attitudes. Administrative Science Quarterly, 22, 427-456. Schappe, S.P. (1998). The influence of job satisfaction, organizational commitment, and fairness perceptions on organizational citizenship behavior. The Journal of Psychology, 132, 277-291.

385

Schaubroeck, J., & Merritt, D.E. (1997). Divergent effects of job control on coping with work stressors: The key role of self-efficacy. Academy of Management Journal, 40, 738-754. Scheier, M.F., Weintraub, J.K., & Carver, C.S. (1986). Coping with stress: Divergent strategies of optimists and pessimists. Journal of Personality and Social Psychology, 51, 1257-1264. Schmidt, G.L. (1976). Job satisfaction among secondary school administrators. Educational Administration Quarterly, 12, 68-86. Schonfeld, I.S. (2000). An updated look at depressive symptoms and job satisfaction in first year women teachers. Journal of Occupational and Organizational Psychology, 73, 363-371. Schwab, D.P., Olian-Gottlieb, J.D., & Heneman, H.G. (1979). Betweensubjects expectancy theory research: A statistical review of studies predicting effort and performance. Psychological Bulletin, 86, 139-147. Schulz, R., & Heckhausen, J. (1996). A life-span model of successful aging. American Psychologist, 51, 702-714. Schulz, R., & Heckhausen, J. (1999). Aging, culture and control. Journal of Gerontology: Psychological Sciences, 54, 139-145. Schwartz, J.E., Neale, J., Marco, C., Shiffman, S.S., & Stone, A.A. (1999). Does trait coping help? A momentary assessment approach to the evaluation of traits. Journal of Personality and Social Psychology, 77, 360-369. Secret, M., & Green, R.G. (1998). Occupational status differences among three groups of married mothers. Journal of Women and Social Work, 13, 47-70.

386

Senate Employment, Education and Training Referees Committee. (1998). A Class Act: Inquiry into the Status of the Teaching Profession. Canberra. Sergiovanni, T. (1967). Factors which affect satisfaction and dissatisfaction of teachers. Journal of Educational Administration, 5, 66-82 Shoura, M.M., & Singh, A. (1999). Motivation parameters for engineering managers using Maslows theory. Journal of Management in Engineering, 15, 44-55. Silver, P.F. (1987). Job satisfaction and dissatisfaction revisited. Educational and Psychological Research, 7, 1-20. Smith, C.A., Organ, D.W., & Near, J.P. (1983). Organizational citizenship behavior: Its nature and antecedents. Journal of Applied Psychology, 68, 653-663. Smith, C.S., Tisak, J., Hahn, S.E., & Schmeider, R.A. (1997). The measurement of job control. Journal of Organizational Behavior, 18, 225-237. Smith, P.C., Kendall, L.M., & Hulin, C.L. (1969). The Job Descriptive Index. Bowling Green, OH: Department of Psychology, Bowling Green State University. Sobel, R.S. (1971). Tests of preperformance and postperformance models of satisfaction with outcomes. Journal of Personality and Social Psychology, 19, 213321. Spector, P.E. (1986). Perceived control by employees: A meta-analysis of studies concerning autonomy and participation at work. Human Relations, 39, 10051017. Spector, P.E. (1987). Interactive effects of perceived control and job stressors on affective reaction and health outcomes for clerical workers. Work and Stress, 1, 155-162.

387

Spector, P.E. (1997). Job Satisfaction: Applications, Assessment, Causes and Consequences. London: Sage Publications. Spector, P.E., Dwyer, D.J., & Jex, S.M. (1988). Relation of job stressors to affective, health and performance outcomes: A comparison of multiple data sources. Journal of Applied Psychology, 73, 11-19. Spector, P.E., & O'Connell, B.J. (1994). The contribution of personality traits, negative affectivity, locus of control and Type A to the subsequent reports of job stressors and job strains. Journal of Occupational and Organizational Psychology, 67, 1-11. Spokes, A. (1998). Subjective quality of life of people with schizophrenia: Examination of perceived control and positive illusions. Unpublished Honours Thesis, Deakin University, Melbourne, Australia. Staw, B.M., Bell, N.E., & Clausen, J.A. (1986). The dispositional approach to job attitudes: A lifetime longitudinal test. Administrative Science Quarterly, 31, 5677. Staw, B.M., & Ross, J. (1985). Stability in the midst of change: A dispositional approach to job attitudes. Journal of Applied Psychology, 70, 469-480. Stone, A.A., & Neale, J.M. (1984). New measure of daily coping: Development and preliminary results. Journal of Personality and Social Psychology, 46, 892-906. Tabachnick, B.G., & Fidell, L.S. (1996). Using Multivariate Statistics (3rd ed.). Northridge: Harper Collins.

388

Taber, T.D., & Taylor, E. (1990). A review and evaluation of the psychometric properties of Job Diagnostic Survey. Personnel Psychology, 43, 467-512. Tait, M., Padgett, MY., & Baldwin, T.T. (1989). Job and life satisfaction: A reevaluation of the strength of the relationship and gender effects as a function of the date of the study. Journal of Applied Psychology, 74, 502-507. Teacher Stress in Victoria: A survey of teachers views: The Report (1990). Victoria: Ministry of Education. Teas, R.K. (1981). A within-subject analysis of valence models of job preference and anticipated satisfaction. Journal of Occupational Psychology, 54, 109124. Terry, D.J., & Hynes, G.J. (1998). Adjustment to a low-control situation: Reexamining the role of coping responses. Journal of Personality and Social Psychology, 74, 1078-1092. Terry, D.J., Nielsen, M., & Perchard, L. (1999). Effects of work stress on psychological well-being and job satisfaction: The stress-buffering role of social support. Australian Journal of Psychology, 45, 168-175. Thurstone, L.L., & Jones, L.V. (1957). The rational origin for measuring subjective values. Journal of the American Statistical Association, 52, 458-471. Tiegs, R.B., Tetrick L.E., & Fried, Y. (1992). Growth need strength and context satisfactions as moderators of the relations of the Job Characteristics Model. Journal of Management, 18, 575-593. Tinsley, H.E.A. (1994). NEO Personality Inventory-Revised. In D.J. Keyser., & R.C. Sweetland (Eds.), Test Critiques, Vol X. (p. 443-456). Texas: ProEd.

389

Thompson, S.C., Collins, M.A., Newcomb, M.D., & Hunt, W. (1996). On fighting versus accepting stressful circumstances: Primary and secondary control among HIV-positive men in prison. Journal of Personality and Social Psychology, 70, 1307-1317. Thompson, S.C., Nanni, C., & Levine, A. (1994). Primary versus secondary and central versus consequence-related control in HIV-Positive men. Journal of Personality and Social Psychology, 67, 540-547. Thompson, S.C., Sobolew-Shubin, A., Galbraith, M.E., Schwankovsky, L., & Cruzen, D. (1993). Maintaining perceptions of control: Finding perceived control in low-control circumstances. Journal of Personality and Social Psychology Bulletin, 64, 293-304. Thompson, S.C., Thomas, C., Rickabaugh, C.A., Tantamjarik, P., Otsuki, T., Pan, D., Garcia, B.F., & Sinar, E. (1998). Primary and secondary control over agerelated changes in physical appearance. Journal of Personality, 66, 583-605. Thurber, C.A., & Weisz, J.R. (1997). You can try or you can just give up: The impact of perceived control and coping style on childhood homesickness. Developmental Psychology, 33, 508-517. Thurstone, L.L., & Jones, L.V. (1957). The rational origin for measuring subjective values. Journal of the American Statistical Association, 52, 458-471. Tokar, D.M., & Subich, L.M. (1997). Relative contributions of congruence and personality dimensions to job satisfaction. Journal of Vocational Behavior, 50, 482-491.

390

Unden, A. (1996). Social support at work and its relationship to absenteeism. Work and Stress, 10, 46-61. Valentiner, D.P., Holahan, C.J., & Moos, R.H. (1994). Social support, appraisals of event controllability and coping: An integrative model. Journal of Personality and Social Psychology, 66, 1094-1102. Van Der Doef, M., & Maes, S. (1999). The job demand-control (-support) model and psychological well-being: a review of 20 years of empirical research. Work and Stress, 13, 87-114. Van Eerde, W., & Thierry, H. (1996). Vrooms expectancy models and workrelated criteria: A meta-analysis. Journal of Applied Psychology, 81, 575-586. Vitaliano, P.P., DeWolfe, D., Maiuro, R.D., Russo, J., & Katon, W. (1990). Appraised changeability of a stressor as a modifier of the relationship between coping and depression: A test of the hypothesis of fit. Journal of Personality and Social Psychology, 59, 582-592. Vitaliano, P.P., Russo, J., Carr, J.E., Maiuro, R.D., & Becker, J. (1985). The Ways of Coping Checklist: Revision and psychometric properties. Multivariate Behavioral Research, 20, 3-26. Vroom, V.H. (1964). Work and motivation. New York: Wiley. Wahba, M.A., & House, R.J. (1976). Expectancy theory in work and motivation: Some logical and methodological issues. Human Relations, 27, 121-147. Wahba, M.A., & Bridwell, L.G. (1976). Maslow reconsidered: A review of the research on the need hierarchy theory. Organizational Behavior and Human Performance, 15, 212-240.

391

Wall, T.D., Clegg, C.W., & Jackson, P.R. (1978). An evaluation of the Job Characteristics Model. Journal of Occupational Psychology, 51, 183-196. Wall, T.D., Jackson, P.R., Mullarkey, S., & Parker, S.K. (1996). The demands-control model of job strain: A more specific test. Journal of Occupational and Organizational Psychology, 69, 153-166. Wanous, J.P., Reichers, A.E., & Hudy, M.J. (1997). Overall job satisfaction: How good are single-item measures? Journal of Applied Psychology, 82, 247-252. Warr, P., Cook, J., & Wall, T. (1979). Scales for the measurement of some work attitudes and aspects of psychological well-being. Journal of Occupational Psychology, 52, 129-148. Warr, P.B. (1990). Decision latitude, job demands, and employee well-being. Work and Stress, 4, 285-294. Waters, L.K., & Waters, C.W. (1969). Correlates of job satisfaction and job dissatisfaction among female clerical workers. Journal of Applied Psychology, 53, 388-391. Weiss, D.J., Dawis, R.V., England,G.W., & Lofquist, L.H. (1967). Manual for the Minnesota Satisfaction Questionnaire. Minneapolis: Industrial Relations Centre, University of Minnesota. Weiss, H.M., Nicholas, J.P., & Daus, C.S. (1999). An examination of the joint effect of affective experienced and job beliefs on job satisfaction and variations in affective experiences over time. Organizational Behavior and Human Decision Processes, 78, 1-24.

392

Wernimont, P.F. (1966). Intrinsic and extrinsic factors in job satisfaction. Journal of Applied Psychology, 50, 41-50. White, R.W. (1959). Motivation reconsidered: The concept of competence. Psychological Review, 66, 297-333. Whitsett, D.A., & Winslow, E.K. (1967). An analysis of studies critical of the motivation-hygiene theory. Personnel Psychology, 391-414. Wicker, F.W., Brown, G., Wiehe, J.A., Hagen, A.S., & Reed, J.L. (1993). On reconsidering Maslow: An examination of the Deprivation/Domination proposition. Journal of Research in Personality, 27, 118-133. Wicker, F.W., & Wiehe, J.A. (1999). An experimental study of Maslows deprivation-domination proposition. Perceptual and Motor Skills, 88, 1356-1358. Wilensky, H.L. (1960). Work, careers and social integration. International Social Science Journal, 12, 543-560. Williams, S., & Cooper, C.L. (1996). Occupational Stress Indicator, Version 2. Harrogate, North Yorkshire: RAD. Winefield, A.H. (2000). Stress in academe: Some recent research findings. In D.T. Kenny., & J.G. Carlson., F.J. McGuigan., J.L. Sheppard (Eds.), Stress and Health: Research and Clinical Applications. (pp. 437-446). Amsterdam: Harwood Academic. Witt, L.A., Andrews, M.C., & Kacmar, K.M. (2000). The role of participation in decision-making in the organisational politics-job satisfaction relationship. Human Relations, 53, 341- 350.

393

Wong, K.S., & Cheuk, W.H., & Rosen, S. (2000). The influence of job stress and supervisor support on negative affects and job satisfaction in kindergarten principals. Journal of Social Behavior and Personality, 15, 85-99. Wright, T.A., Bennett, K.K., & Dun, T. (1999). Life and job satisfaction. Psychological Reports, 84, 1025-1028. Wyrwich, K.W., Nienaber, N.A., Tierney, W.M., & Wolinsky, F.D. (1999). Linking clinical relevance and statistical significance in evaluating individual changes in health-related quality of life. Medical Care, 37, 469-478.

394

4.7

Appendices

395

Appendix A- Plain Language Statement for Study One Dear Sir/Madam, My name is Elise Maher, and I am completing my Ph.D. in Psychology at Deakin University. As part of my studies, I am undertaking a research project under the supervision of Professor Robert Cummins, a researcher in the School of Psychology. This study is investigating job satisfaction and control. The study aims to provide useful information about how the amount of choice that an employee has influences job satisfaction. The results will provide information that will enhance programs that increase job satisfaction. You are invited to participate in this research. If you agree, you will be asked to complete the enclosed questionnaire. Any information you provide will be anonymous and confidential. Only group results will be reported and no individuals will be identified. Upon completion of the study, data will be secured in a locked cabinet in the School of Psychology, Deakin University, for a minimum period of six years from the date of publication. The questionnaire should take around 30 minutes to complete and your participation would be greatly appreciated. Examples of questions are: "My work is boring", "In my job, I can choose the amount I earn", "I am not a worrier" and "How satisfied are you with your close relationships with family or friends". You are free to withdraw at any time during the study in which event your participation in the research study will immediately cease and any information obtained will not be used. You are free to refuse to answer any questions. Following the completion of the study, I will provide your employer with a summary of the results. If you would like a copy of the summary sent directly to you, please contact Elise Maher. If you have any further questions regarding the study, please contact: Elise Maher on 9251 7153 or Email: ecmaher@deakin.edu.au Or you can contact Professor Robert Cummins on 9244-6845 or Email: robert.cummins@deakin.edu.au. If you are happy to be involved in this study, please complete the enclosed questionnaire and return it in the reply-paid envelope supplied (i.e., NO STAMP NEEDED).
________________________________________________________________________ Should you have any concerns about the conduct of this research project, please contact the Secretary, Deakin University Ethics Committee, Research Services, Deakin University, 221 Burwood Highway, BURWOOD, VIC, 3125, Tel (03) 9251 7123

396

Appendix B- Job Autonomy Scale used in Study One (Revision of Ganster, 1989, cited in Dwyer & Ganster, 1991) Indicate your agreement with the following 13 statements by ticking () a number ranging from 1 to 10, where 1= Do not agree at all, and 10= Agree completely. All of the statements begin with In my job, I can choose. In my job, I can choose: 1) In my job, I can choose among a variety of tasks or projects to do.
0
Do not agree at all

10
Agree Completely

2) In my job, I can choose the order in which I do my work.

0
Do not agree at all

10
Agree Completely

3) In my job, I can choose how quickly I work.

0
Do not agree at all

10
Agree Completely

4) In my job, I can choose how I schedule my rest breaks.

0
Do not agree at all

10
Agree Completely

5) In my job, I can choose the physical conditions of my workstation.

0
Do not agree at all

10
Agree Completely

6) In my job, I can choose when I interact with others.

0
Do not agree at all

10
Agree Completely

7) In my job, I can choose the amount I earn.

0
Do not agree at all

10
Agree Completely

8) In my job, I can choose the number of times I am interrupted at work.

0
Do not agree at all

10
Agree Completely

397

9) In my job, I can choose how my work is evaluated.

0
Do not agree at all

10
Agree Completely

10) In my job, I can choose the quality of my work.

0
Do not agree at all

10
Agree Completely

11) In my job, I can choose the policies and procedures in my work unit.
0
Do not agree at all

10
Agree Completely

12) In my job, I can choose among a variety of methods to complete my work.

0
Do not agree at all

10
Agree Completely

13) In my job, I can choose how much work I get done.

0
Do not agree at all

10
Agree Completely

14) In general, how much are you able to influence work and work-related matters?
0
Very Little

10
Very Much

398

Appendix C- Primary and Secondary Control Scale used in Study One (Revision of Heeps et al., 2000) Indicate your agreement with the following statements by selecting a number ranging from 1 to 10, where 1=Do not agree at all, and 10= Agree Completely
1) When a goal that I have at work is difficult to reach, I think about different ways to achieve it. 0
Do not agree at all

10
Agree Completely

2) When I want something at work to change, I think I can make it happen. 0

Do not agree at all

10
Agree Completely

3) When a work task really matter to me, I think about it a lot. 0

Do not agree at all

10
Agree Completely

4) When I really want to reach a goal at work, I believe I can achieve it. 0
Do not agree at all

10
Agree Completely

5) When faced with a difficult work situation, I believe I can overcome it. 0
Do not agree at all

10
Agree Completely

399

Indicate your agreement with the following statements by selecting a number ranging from 1 to 10 where 1=Do not agree at all, and 10=Agree completely. All the statements begin with When something bad happens that I cannot change When something bad happens at work that I cannot change: When something bad happens at work that I cannot change
1) I can see that something good will come of it. 0
Do not agree at all

10
Agree Completely

When something bad happens at work that I cannot change

2) I remember you cant always get what you want. 0
Do not agree at all

10
Agree Completely

When something bad happens at work that I cannot change

3) I know things will work out OK in the end. 0
Do not agree at all

10
Agree Completely

When something bad happens at work that I cannot change

4) I remember I am better off than many other people. 0
Do not agree at all

10
Agree Completely

When something bad happens at work that I cannot change

5) I remember I have already accomplished a lot in life. 0
Do not agree at all

10
Agree Completely

When something bad happens at work that I cannot change

6) I remember the success of my family or friends. 0
Do not agree at all

10
Agree Completely

When something bad happens at work that I cannot change

7) I think nice thoughts to take my mind off it. 0
Do not agree at all

10
Agree Completely

When something bad happens at work that I cannot change

400

8) I remind myself the situation will change if I am just patient. 0

Do not agree at all

10
Agree Completely

When something bad happens at work that I cannot change

9) I tell myself it doesnt matter. 0
Do not agree at all

10
Agree Completely

When something bad happens at work that I cannot change

10) I think about my success in other areas. 0
Do not agree at all

10
Agree Completely

When something bad happens at work that I cannot change

11) I dont feel disappointed because I knew it might happen. 0
Do not agree at all

10
Agree Completely

When something bad happens at work that I cannot change

12) I can see it was not my fault. 0
Do not agree at all

10
Agree Completely

When something bad happens at work that I cannot change

13) I ignore it by thinking about other things. 0
Do not agree at all

10
Agree Completely

When something bad happens at work that I cannot change

14) I realise I didnt need to control it anyway. 0
Do not agree at all

10
Agree Completely

401

Appendix D- Job Satisfaction Scale used in Study One (Revision of Roznowski, 1989) Indicate your agreement with the following 15 statements by ticking () a number ranging from 1 to 10, where 1= Do not agree at all, and 10= Agree completely. 1) My work is boring.
0
Do not agree at all

10
Agree Completely

2) My co-workers are stupid.

0
Do not agree at all

10
Agree Completely

3) My pay is bad.
0
Do not agree at all

10
Agree Completely

4) My supervisors know how to supervise.

0
Do not agree at all

10
Agree Completely

5) There is a good chance for promotion in my job.

0
Do not agree at all

10
Agree Completely

6) My co-workers are responsible.

0
Do not agree at all

10
Agree Completely

7) I am well-paid.
0
Do not agree at all

10
Agree Completely

8) My work is dull.
0
Do not agree at all

10
Agree Completely

402

9) There is a fairly good chance for promotion in my job.

0
Do not agree at all

10
Agree Completely

10) My supervisors are bad.

0
Do not agree at all

10
Agree Completely

11) My work is interesting.

0
Do not agree at all

10
Agree Completely

12) My pay is unfair.

0
Do not agree at all

10
Agree Completely

13) My supervisors are annoying.

0
Do not agree at all

10
Agree Completely

14) My co-workers are a waste of time.

0
Do not agree at all

10
Agree Completely

15) There are good opportunities for advancement in my job.

0
Do not agree at all

10
Agree Completely

403

Appendix E- Life Satisfaction Scale used in Study One (Cummins, 1997) Please tick () the box that best describes how SATISFIED you are with each area. Do not spend too much time on any one question. There are no right or wrong answers. 1) How Satisfied are you with the THINGS YOU OWN ?
0
Completely dissatisfied

10
Completely satisfied

2) How Satisfied are you with your HEALTH?

0
Completely dissatisfied

10
Completely satisfied

3) How Satisfied are you with what you ACHIEVE IN LIFE ?

0
Completely dissatisfied

10
Completely satisfied

4) How Satisfied are you with your CLOSE RELATIONSHIPS with FAMILY or FRIENDS ?
0
Completely dissatisfied

10
Completely satisfied

5) How Satisfied are you with HOW SAFE YOU FEEL ?

0
Completely dissatisfied

10
Completely satisfied

6) How Satisfied are you with feeling part of your COMMUNITY?

0
Completely dissatisfied

10
Completely satisfied

7) How Satisfied are you with YOUR OWN HAPPINESS ?

0
Completely dissatisfied

10
Completely satisfied

404

Appendix F- Personality Scale used in Study One (Costa & McCrae, 1992) This questionnaire contains 24 statements. Read each statement carefully. For each statement tick () the box with the response that best represents your opinion. 1) I am not a worrier.
0
Do not agree at all

10
Agree Completely

2) I like to have a lot of people around me.

0
Do not agree at all

10
Agree Completely

3) I often feel inferior to others.

0
Do not agree at all

10
Agree Completely

4) I laugh easily.
0
Do not agree at all

10
Agree Completely

5) When I'm under a great deal of stress, sometimes I feel like I'm going to pieces.
0
Do not agree at all

10
Agree Completely

6) I don't consider myself especially light hearted.

0
Do not agree at all

10
Agree Completely

7) I rarely feel lonely or blue.

0
Do not agree at all

10
Agree Completely

8) I really enjoy talking to people.

0
Do not agree at all

10
Agree Completely

9) I often feel tense and jittery.

405

0
Do not agree at all

10
Agree Completely

10) I like to be where the action is.

0
Do not agree at all

10
Agree Completely

11) Sometimes I feel completely worthless.

0
Do not agree at all

10
Agree Completely

12) I usually prefer to do things alone.

0
Do not agree at all

10
Agree Completely

13) I rarely feel fearful or anxious.

0
Do not agree at all

10
Agree Completely

14) I often feel as if I am bursting with energy.

0
Do not agree at all

10
Agree Completely

16) I often get angry at the way people treat me.

0
Do not agree at all

10
Agree Completely

16) I am a cheerful, high-spirited person.

0
Do not agree at all

10
Agree Completely

17) Too often, when things go wrong, I get discouraged and feel like giving up.
0
Do not agree at all

10
Agree Completely

18) I am not a cheerful optimist.

406

0
Do not agree at all

10
Agree Completely

19) I am seldom sad or depressed.

0
Do not agree at all

10
Agree Completely

20) My life is fast-paced.

0
Do not agree at all

10
Agree Completely

21) I often feel helpless and want someone else to solve my problems.
0
Do not agree at all

10
Agree Completely

22) I am a very active person.

0
Do not agree at all

10
Agree Completely

23) At times I have been so ashamed I just want to hide.

0
Do not agree at all

10
Agree Completely

24) I would rather go my own way than be a leader of others.

0
Do not agree at all

10
Agree Completely

407

Appendix G-Levels of Job Satisfaction reported by various occupational groups No 1 Author Leung et al., (2000) Judge et al., (1998) Judge et al., (2000) Renn & Vandenberg (1995) Hackman & Oldham (1975) Dollard et al., (2000) Wall, Jackson, Mullarkey & Parker (1996) ODriscoll & Beehr (2000) Fletcher & Jones (1993) Fox et al., 1993 Mannheim, Baruch & Tal (1997) Beutell & WittigBerman (1999) Agho, Price & Mueller (1992) Fisher (2000) Occupation Academics Scale Job Satisfaction Scale (OSI-2; Williams & Cooper, 1996) Brayfield & Rothe (1951) Brayfield & Rothe (1951) Job Diagnostic Survey (Hackman & Oldham, 1975) Job Diagnostic Survey (Hackman & Oldham, 1975) Global measure Warr, Cook & Wall (1979) intrinsic satisfaction Facet measure (12 facets) Global measure N 106 Job Satisfaction (%SM) 57.6

2 3 4 5 6 7

Physicians General Management, counseling, administration General Public sector welfare workers Manufacturing employees Accounting firms Manual and non-manual workers Nurses Managerial personnel MBA students

164 384 107 188 658 786 1451

72.67 69.6, 76.53 64.8 60.3 64 51.33

8 9

236 501 (manual) 788 (nonmanual) 136 39 177

68 60 58 76.33 70 52.5

10 11 12

Faces scale (Kunin, 1955) Job Diagnostic Survey (Hackman & Oldham, 1975) Global measure (4 items) Brayfield & Rothe (1951) Faces scale (Kunin,

13 14

Employees of medical centre General

550 124

62.04 87.0

408

1955) Howard & Frink (1996) Managers, administrators police, firefighters, labourers Nurses Job Diagnostic Survey (Hackman & Oldham, 1975) 248 65.83

15 16

17 18 19

Jansen, Kerkstra, Abud-Saad & Van der Zee (1996) Laschinger et al., (2001) Schonfeld (2000) Finlay, Martin, Romas & Blum (1995) Ma & Macmillan (1999) Geyer & Daly (1998) Schappe (1998) Parsons (1998) Pearson & Chong (1997) Miles, Patrick & King (1996) Witt, Andrews & Kacmar (2000) Bogg & Copper (1995) Weiss,

Facet measure (Algera, 1980)

355 nurses, 92 nurse auxiliaries 600 184 169

63.75 (nurses) & 68.0 (nurse auxiliaries) 44.75 80 78.5

Nurses Graduate teachers Administrators

Job Diagnostic Survey (Hackman & Oldham, 1975) Quinnes & Staines (1979)-Global Hoppock (1935)

20 21 22

Teachers Private sector organisation Insurance company workers Nurses Nurses Manufacturing employees Public sector organisation Civil servants, executives Managers

Global items (4) Global items (3) Minnesota Satisfaction Questionnaire (Weiss et al., 1967) Brayfield & Rothe (1951) Job Diagnostic Survey (Hackman & Oldham, 1975) Job perception scale (Hatfield, Robinson & Huseman, 1985) Hoppock (1935)

2,202 174 150

80.25 57 71.50

23 24 25 26

47 286 713 1251

71.36 68.88 57.5 63.75

27 28

OSI (Cooper, Sloan & Williams, 1988) Faces Scale (Kunin,

1051 1056 24

55.6 62.0 82

409

29

Nicholas & Daus (1999) Moorman (1993)

Manufacturers

1955), and global scale Brayfield & Rothe (1951)

62.25 71

30 31

Parahoo & Barr (1994) Spector, Dwyer & Jex (1988)

Nurses Secretaries

Global measure (1 item) Michigan Orgnizational Assessment (Cammann, Fichman, Jenkins & Klesh, 1979) Michigan Orgnizational Assessment (Cammann et al., 1979) Global item National follow-up survey of teachers education graduates- 7 facets Global measure- 1 item Global Scale (Kandel, Davies & Raveis, 1985)

35 155

75 78

32

Spector & OConnell (1994) Wong et al., (2000) Klecker & Loadman (1999)

University Graduates

66.66

33 34

Kindergarten principles Teachers

108 1874

54 68.16

35 36

Marriott & Social workers Sexton (1994) Frone, Random Russell & Cooper (1994)

188 631

66.9 73.33

MEAN (N= 41)

66.75

410

Appendix H- Primary and Secondary Control Scale for Study Two (Maher et al., 2001)

The following items assess the difficulties that you have at work. Please tick () the areas in which you experience difficulties in your work.

Time management (making time to do everything) Motivation Interpersonal relationships (colleagues, or supervisors) Nature of the Work Promotions Pay Other

b) How often do you have difficulty doing something at work? (ie., think of the examples given above, or other difficulties you may have had at work)
0
Never

10
All the time

411

Here are ways people deal with difficult situations at work. How often have you had these thoughts when facing a difficulty at work OVER THE PAST WEEK?

I thought 1) It will work out okay in the end.

0
Never

10
Every time

2) I knew it would happen.

0
Never

10
Every time

3) I can't always get what I want.

0
Never

10
Every time

4) It doesnt matter.
0
Never

10
Every time

5) I am better off than many other people.

0
Never

10
Every time

6) It was not my fault.

0
Never

10
Every time

412

Here are other ways people deal with difficult situations at work. How often have you done these things when facing a difficulty at work

OVER THE PAST WEEK?

7) I looked for different ways to overcome it.
0
Never

10
Every time

8) I kept trying.
0
Never

10
Every time

9) I told someone about it.

0
Never

10
Every time

10) I worked to overcome it.

0
Never

10
Every time

11) I thought of the success of my family or friends.

0
Never

10
Every time

12) I thought about my success in other areas.

0
Never

10
Every time

13) I did something different, like going for a walk.

0
Never

10
Every time

14) I ignored it.

0
Never

10
Every time

15) I worked out how to remove obstacles.

413

0
Never

10
Every time

16) I looked for something else that was positive in the situation.
0
Never

10
Every time

414

Appendix I- Plain Language Statement for Study Two

Dear Sir/Madam, My name is Elise Maher, and I am completing my Ph.D. in Psychology at Deakin University. As part of my studies, I am undertaking a research project under the supervision of Professor Robert Cummins, a researcher in the School of Psychology. This study is investigating job satisfaction and control. The study aims to provide useful information about how the amount of choice that an employee has influences job satisfaction. The results will provide information that will enhance programs that increase job satisfaction. You are invited to participate in this research. If you agree, you will be asked to complete the enclosed questionnaire. Any information you provide will be anonymous and confidential. Only group results will be reported and no individuals will be identified. Upon completion of the study, data will be secured in a locked cabinet in the School of Psychology, Deakin University, for a minimum period of six years from the date of publication. The questionnaire should take around 30 minutes to complete and your participation would be greatly appreciated. Examples of questions are: "I am satisfied with the praise I get for doing a good job", "How much can you choose the amount that you earn", "I am not a worrier", "How satisfied are you with your close relationships with family or friends", and Which management style do you prefer. You are free to withdraw at any time during the study in which event your participation in the research study will immediately cease and any information obtained will not be used. You are free to refuse to answer any questions. Following the completion of the study, I will provide your employer with a summary of the best coping strategies. If you would like a copy of the summary sent directly to you, please contact Elise Maher. If you have any further questions regarding the study, please contact: Elise Maher on 9251 7153 or Email: elisem@deakin.edu.au, or you can contact Professor Robert Cummins on 9244-6845 or Email: cummins@deakin.edu.au If you are happy to be involved in this study, please complete the enclosed questionnaire and return it in the reply-paid envelope supplied (i.e., NO STAMP NEEDED). Thank you very much for your time. _________________________________________________________________
Should you have any concerns about the conduct of this research project, please contact the Secretary, Deakin University Ethics Committee, Research Services, Deakin University, 221 Burwood Highway, BURWOOD, VIC, 3125, Tel (03) 9251 7123

Appendix J- Job Autonomy Scale for Study Two (Hackman & Oldham, 1975)

415

The following 3 items assess how much freedom you have at your work. For each item, please tick () a number ranging from 1 to 10, where 1= Do not agree at all, and 10= Agree completely. 1) In my job, I can decide on my own how to go about doing my work.
0
Do not agree at all

10
Agree completely

2) In my job, I have the chance to use my personal initiative and judgement in carrying out the work.
0
Do not agree at all

10
Agree completely

3) In my job, I have considerable opportunity for independence and freedom.

0
Do not agree at all

10
Agree completely

416

Appendix K- Need for Autonomy Scale for Study Two (de Rijk et al., 1998) The following 4 items assess how important it is for you to do certain things at work. Please tick a box ranging from 1=Not important at all to 10=Could not be more important. 1) How important is it for you to set the pace of your tasks at work?
0
Not important at all

10
Could not be more important

2) How important is it for you to have control over what you do at work and the way that you do it?
0
Not important at all

10
Could not be more important

3) How important is it for you to do your own planning at work?

0
Not important at all

10
Could not be more important

4) How important is it for you to give orders at work instead of receiving them?
0
Not important at all

10
Could not be more important

417

Appendix L- Job Satisfaction Scale for Study Two (Weiss et al., 1967) Please indicate how satisfied you are with the following aspects of your work. Please tick a box ranging from (1) Very dissatisfied to (10) Very satisfied. On my present job, this is how I feel about.. 1) Being able to keep busy all the time.
0
Very dissatisfied

10
Very satisfied

2) The chance to work alone on the job.

0
Very dissatisfied

10
Very satisfied

3) The chance to do different things from time to time.

0
Very dissatisfied

10
Very satisfied

4) The chance to be somebody in the community.

0
Very dissatisfied

10
Very satisfied

5) The way my boss handles his/her work.

0
Very dissatisfied

10
Very satisfied

6) The competence of my supervisor in making decisions.

0
Very dissatisfied

10
Very satisfied

7) Being able to do things that dont go against my conscience.

0
Very dissatisfied

10
Very satisfied

418

8) The way my job provides for steady employment.

0
Very dissatisfied

10
Very satisfied

9) The chance to do things for other people.

0
Very dissatisfied

10
Very satisfied

10) The chance to tell people what to do.

0
Very dissatisfied

10
Very satisfied

11) The chance to do something that makes use of my abilities.

0
Very dissatisfied

10
Very satisfied

12) The way company politics are put into practice.

0
Very dissatisfied

10
Very satisfied

13) My pay and the amount of work I do.

0
Very dissatisfied

10
Very satisfied

14) The chances for advancement on my job.

0
Very dissatisfied

10
Very satisfied

15) The freedom to use my own judgement.

0
Very dissatisfied

10
Very satisfied

16) The chance to try my own methods of doing the job.

0
Very dissatisfied

10
Very satisfied

419

17) The working conditions.

0
Very dissatisfied

10
Very satisfied

18) The way my co-workers get along with each other.

0
Very dissatisfied

10
Very satisfied

19) The praise I get for doing a good job.

0
Very dissatisfied

10
Very satisfied

20) The feeling of accomplishment I get from the job.

0
Very dissatisfied

10
Very satisfied

420

Appendix M- Social Support Scale for Study Two (Revision of Karasek & Theorell, 1990) The following 8 questions ask about your supervisor and your co-workers. Please circle a number 1= Not true at all to 10= Could not be more true. 1) My supervisor shows concern for me.
0
Not true at all

10
Could not be more true

2) My supervisor pays attention to me.

0
Not true at all

10
Could not be more true

3) My supervisor is helpful getting work done.

0
Not true at all

10
Could not be more true

4) My supervisor creates a good teamwork environment for me.

0
Not true at all

10
Could not be more true

5) My co-workers are friendly to me.

0
Not true at all

10
Could not be more true

6) My co-workers are helpful to me.

0
Not true at all

10
Could not be more true

7) My co-workers are personally interested in me.

0
Not true at all

10
Could not be more true

8) My co-workers are competent.

0
Not true at all

10
Could not be more true

421

Appendix N-Plain Language Statement used in Study Three Dear Sir/Madam, My name is Elise Maher, and I am completing my Ph.D. in Psychology at Deakin University. As part of my studies, I am undertaking a research project under the supervision of Professor Robert Cummins, a researcher in the School of Psychology. This study is investigating job satisfaction and coping. The study aims to provide useful information about the best type of coping strategies that workers should use. The results will provide information that will enhance programs that increase job satisfaction. You are invited to participate in this research. If you agree, you will be asked to complete the enclosed questionnaire. Any information you provide will be anonymous and confidential. Only group results will be reported and no individuals will be identified. Upon completion of the study, data will be secured in a locked cabinet in the School of Psychology, Deakin University, for a minimum period of six years from the date of publication. The questionnaire should take around 20 minutes to complete and your participation would be greatly appreciated. Examples of questions are: "I can decide on my own about how to go about doing my work", "Your co-workers really care about you", "What type of difficulties do you face at work?" and "How satisfied are you with your close relationships with family or friends." You are free to withdraw up until you have returned the survey, in which event your participation in the research study will immediately cease and any information obtained will not be used. You are free to refuse to answer any questions. Following the completion of the study, I am happy to provide you a summary of the best coping strategies. If you would like a copy of the summary or if you have any further questions regarding the study, please contact: Elise Maher on (03) 9251 7153 or Email: ecmaher@deakin.edu.au, or Professor Robert Cummins on (03) 9244-6845 or Email: cummins@deakin.edu.au. If you are happy to be involved in this study, please complete the enclosed questionnaire and return it in the reply-paid envelope supplied (i.e., NO STAMP NEEDED). Thank you very much for your time.
__________________________________________________________________
Should you have any concerns about the conduct of this research project, please contact the Secretary, Deakin University Ethics Committee, Research Services, Deakin University, 221 Burwood Highway, BURWOOD, VIC, 3125, Tel (03) 9251 7123

422

Appendix O- Primary and Secondary Control Scale for Study Three (Maher & Cummins, 2002) People may experience several kinds of difficulties in their work. They can control some of them, but not others. For example, Worker A, a teacher, can control difficulties involving students, parents and time management. They cannot however control difficulties involving school policies and work times. Another example, Worker B a supermarket operator, can control difficulties involving customers and co-workers. They cannot control difficulties involving pay, promotion and holiday leave. 1) Tick the difficulties you experience at work that you CAN CONTROL.

Difficulties with Supervisor(s) Difficulties with Promotion Difficulties with Co-worker(s) Difficulties with Time Management Difficulties with Kind of work you do Difficulties with Motivation Difficulties with Pay Difficulties with Work Times

Difficulties with Work-place rules Difficulties with Amount of Work Other

2) Consider the difficulty that you experience MOST OFTEN, and which you CAN CONTROL. How often do you experience this difficulty?
1
Rarely

2
Sometimes

3
Often

4
Always

423

3) When you face this difficulty that you CAN CONTROL, how often do you do the following? a) Discuss solutions with the people involved
0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

b) Think that the difficulty doesnt matter

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

c) Think that this difficulty will work out okay in the end
0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

d) Choose a solution and act on it

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

e) Think that I knew this difficulty would happen

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

f) Think that I cant always get what I want

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

g) Work harder
0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

h) Think that I am better off than many other people

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

i) Think that this difficulty is not my fault

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

j) Keep trying

424

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

k) Tell someone about this difficulty to make me feel better

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

l) Think of the success of my family/friends

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

m) Think about my success in other areas

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

n) Do something different, like going for a walk

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

o) Ignore this difficulty

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

p) Look for something else that is positive in the situation

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

q) Other ..(please specify)

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

425

The following questions examine difficulties that you CANNOT CONTROL 4) Tick the difficulties you experience at work that you CANNOT CONTROL.

Difficulties with Supervisor(s) Difficulties with Promotion Difficulties with Co-worker(s) Difficulties with Time Management Difficulties with Kind of work you do Difficulties with Motivation Difficulties with Pay Difficulties with Work Times

Difficulties with Work-place rules Difficulties with Amount of Work Other

5) Consider the difficulty that you experience MOST OFTEN, and which you CANNOT CONTROL. How often do you experience this difficulty?
1
Rarely

2
Sometimes

3
Often

4
Always

6) When you face this difficulty that you CANNOT CONTROL, how often do you do the following? a) Discuss solutions with the people involved
0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

b) Think that the difficulty doesnt matter

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

c) Think that this difficulty will work out okay in the end
0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

d) Choose a solution and act on it

426

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

e) Think that I knew this difficulty would happen

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

f) Think that I cant always get what I want

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

g) Work harder
0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

h) Think that I am better off than many other people

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

i) Think that this difficulty is not my fault

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

j) Keep trying
0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

k) Tell someone about this difficulty to make me feel better

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

l) Think of the success of my family/friends

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

m) Think about my success in other areas

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

n) Do something different, like going for a walk

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

427

o) Ignore this difficulty

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

p) Look for something else that is positive in the situation

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

q) Other (please specify)

0
Never

1
Rarely

2
Sometimes

3
Often

4
Always

428

Appendix P- Life Satisfaction Scale for Study 3 (Cummins et al., 2001)

1) How satisfied are you with your standard of living?

0
Completely dissatisfied

10
Completely satisfied

2) How satisfied are you with your health?

0
Completely dissatisfied

10
Completely satisfied

3) How satisfied are you with what you achieve in life?

0
Completely dissatisfied

10
Completely satisfied

4) How satisfied are you with your personal relationships?

0
Completely dissatisfied

10
Completely satisfied

5) How satisfied are you with how safe you feel?

0
Completely dissatisfied

10
Completely satisfied

6) How satisfied are you with feeling part of your community?

0
Completely dissatisfied

10
Completely satisfied

7) How satisfied are you with your future security?

0
Completely dissatisfied

10
Completely satisfied

429

Appendix Q- Social Support Scale for Study 3 (Ducharme & Martin, 2000) The following 6 questions ask about your co-workers. Please circle a number ranging from 0 to 10, where 0= Do not agree at all and 10= Agree completely. 1) My co-workers really care about me.
0
Do not agree at all

10
Agree completely

2) I feel close to my co-workers.

0
Do not agree at all

10
Agree completely

3) My co-workers take a personal interest in me.

0
Do not agree at all

10
Agree completely

4) My co-workers assist with unusual work problems.

0
Do not agree at all

10
Agree completely

5) My co-workers are helpful in getting the job done.

0
Do not agree at all

10
Agree completely

6) My co-workers give useful advice on job problems.

0
Do not agree at all

10
Agree completely

430

The following 6 questions ask about your supervisor(s). For each item, please circle a number ranging from 0 to 10, where 0= Do not agree at all, and 10= Agree completely 1) My supervisor really cares about me.
0
Do not agree at all

10
Agree completely

2) I feel close to my supervisor.

0
Do not agree at all

10
Agree completely

3) My supervisor takes a personal interest in me.

0
Do not agree at all

10
Agree completely

4) My supervisor assists with unusual work problems.

0
Do not agree at all

10
Agree completely

5) My supervisor is helpful in getting the job done.

0
Do not agree at all

10
Agree completely

6) My supervisor gives useful advice on job problems.

0
Do not agree at all

10
Agree completely