You are on page 1of 19

International Journal of Stress Management, Vol. 8, No.

2, 2001

The Assessment of Workaholism as Behavioral Tendencies: Scale Development and Preliminary Empirical Testing
Peter E. Mudrack1,3 and Thomas J. Naughton2

This article reports on the development and preliminary empirical testing of two new scales designed to assess workaholic behavioral patterns. One scale assessed employee tendencies to perform nonrequired work (typically, spending time thinking of ways to perform work better), and the other measured tendencies to intrude on and attempt to control the work of others. These scales seem consistent with both conceptual definitions and anecdotal accounts of workaholism. Moreover, the scales were designed to assess behavioral tendencies rather than work attitudes, to circumvent the possibility of rationalization and denial among workaholic respondents, and to be applicable across multiple work contexts. Evidence from a sample of 278 employed respondents indicated that the scales were internally consistent, and represented distinct constructs. The scales also demonstrated evidence of convergent and discriminant validity. For example, scores on the nonrequired work scale correlated positively with hours worked and with the needs for achievement, autonomy, and affiliation. Scores on the control of others scale correlated positively with job involvement, stress, work-nonwork conflict, role conflict, dominance needs, and self-monitoring. In summary, the workaholism scales developed for this investigation appear both reliable and valid and thus potentially able to shed light on the phenomenon of workaholism.
KEY WORDS: workaholism; scale development; construct validity.

1 2

Department of Management, Kansas State University. Department of Management and Organization Sciences, Wayne State University. 3 Correspondence should be directed to Peter E. Mudrack, Department of Management, 101 Calvin Hall, Kansas State University, Manhattan, KS, 66506-0507; e-mail pmudrack@ksu.edu. 93
1072-5245/01/0400-0093/$19.50/0 2001 Human Sciences Press, Inc.

94

Mudrack and Naughton

INTRODUCTION Almost 30 years ago, Oates (1971) affixed the label workaholism to a potentially destructive set of behaviors. Workaholism was intended to evoke the word alcoholism and to describe a pattern of inappropriate dependence common to both domains. The notion of workaholism has been discussed regularly in the business and popular press (e.g., Baechler, 1996; Costello, 2000; Gini, 1998) and thus appears to be meaningful to business practitioners. In spite of its obvious relevance, little empirical research has been devoted to workaholism. One possible reason for the relative absence of research is that existing survey instruments may not adequately capture the phenomenon of workaholism. The existence of reliable and valid measures is crucial in order to increase our understanding of workaholism (or, for that matter, of any phenomenon). The purpose of this article is to report on the development and preliminary empirical testing of two new scales designed to assess behavioral manifestations of workaholism. Before discussing our new scales, we must first define workaholism and articulate why existing measures of workaholism may not be appropriate. Scott, Moore, and Miceli (1997) defined workaholism as consisting of three behavioral patterns: (1) spending discretionary time in work activities; (2) thinking about work when not at work; (3) working beyond organizational or economic requirements. Spence and Robbins (1992) developed three self-report workaholism scales, but these do not seem totally consistent with such a definition. Their scales assess the degree to which respondents are highly involved in their work, feel driven to work, and have a sense of enjoyment of work. Undoubtedly, such notions are germane to the phenomenon of workaholism. However, an obvious shortcoming with the Spence and Robbins scales is that they more closely assess attitudes or affect rather than behavioral tendencies. As such, these scales may be conceptually indistinguishable from other well-established measures in the organizational literature such as satisfaction, involvement, or commitment. Measures of workaholism, in order to be meaningful and valid, must not overlap closely with existing measures of distinct constructs. Survey items that are consistent with the preceding definition of workaholism must clearly be relevant to work activities that are not obligatory and that involve an element of free choice. For example, employees who work an extraordinarily large number of hours per week would not necessarily be workaholics if such work habits were expected by others or required in order to complete assigned tasks. Employees who work hard and focus exclusively on tasks at hand during scheduled work hours would not necessarily be workaholics if they also forget about work at other times. Accordingly, survey items pertaining to how hard respondents work, how intensely they focus on work activities, or how many hours they work would seem to be imperfect indicators of worka-

Workaholism: Scale Development

95

holism. Such items do not directly assess the notions of working beyond the expectations of others, thinking about work when not at work, or choosing to spend free time in work activities. A possible approach in designing workaholism survey items might be to ask respondents if they work more than absolutely necessary to do their jobs or if they spend discretionary time on optional work activities. Such an approach may not be appropriate, however, in light of the element of denial and rationalization that seems to characterize workaholism. For example, Scott et al. (1997) observed that workaholic individuals may attempt to construct plausible but perhaps misleading explanations for their work habits (see also Schaef & Fassel, 1988). Such persons may claim (irrespective of their true motives) that they work as they do because they have no real choice in the matter. Such claims may serve to justify (whether to themselves or to others) their activities in a socially acceptable manner by conveniently removing the element of discretion. Therefore, it may be imprudent to ask respondents directly if the work that they perform is somehow not technically required or has an element of free choice. Given our suspicion that some persons might be unable or unwilling to admit that the work they do is actually discretionary, we concluded that asking about time and energy spent on specific activities would likely provide more candid responses than asking directly about time spent on optional work. Designing survey items relevant to specific activities, however, may require researchers to have in-depth knowledge of respondents work situations in order to identify those behaviors that are truly discretionary. One difficulty in such an approach is that items relevant to a specific context may be neither applicable nor meaningful to respondents in a different work setting (or even to respondents in the same work setting who occupy other positions). Consequently, different workaholism survey items may be needed for different groups of respondents. However, comparing workaholism scores among respondents who did not complete the same survey items seems ill advised. The ultimate value of developing context-specific workaholism survey items also seems questionable if other researchers cannot simply use the items to help answer research questions, but must first undertake to develop their own unique items. In order to be useful in research, workaholism survey items must refer to work activities that are neither required nor obligatory across multiple work contexts. For most (if not all) employees, we assumed that performing their jobs and executing assigned work would typically be required, but that initiating work and thinking of ways to do work better would typically not be required. We also assumed that most (if not all) employees would not necessarily be obligated to monitor the work activities of other employees in an active and intrusive manner. Even supervisors would seemingly not be compelled to be intrusive when performing their assigned tasks. These two patterns of work activities (thinking about how to do work better and intrusive interference in the

96

Mudrack and Naughton

work of others) seem both relevant and applicable in many work domains, as well as consistent with the conceptual definition of workaholism offered earlier. Such activities are, we argue, not required of most employees, are discretionary, and are demanding enough so that employees are able to think about work in their spare time. Workaholism has traditionally been operationalized as an intrapersonal phenomenon. Spence and Robbins (1992), for example, developed items that asked respondents to report on their own levels of work involvement, sense of feeling driven to work, and enjoyment of work. The preceding discussion suggests, however, that workaholism may also have an important interpersonal component. Indeed, workaholic tendencies often affect others profoundly. Relations between workaholic persons and nonworkaholic coworkers may tend to be adversarial or even hostile (Oates, 1971; Scott et al., 1997) because these coworkers seemingly cannot match the workaholics own intense dedication to work. Machlowitz (1980), for example, described how workaholics tend to create a pressure cooker atmosphere for the people around them and seem inclined to meddle in the work of others and look over the shoulders of fellow employees (see also Scott et al., 1997; Seybold & Salomone, 1994). In short, workaholic individuals seem to affect coworkers negatively and may be unable or unwilling to allow others to work without interference. Thus, an interpersonal and controlling element to workaholism seems consistent with earlier theorizing on the subject.

SURVEY ITEMS Next, we describe the specific survey items that emerged from the process discussed above. One set of survey items pertained to self-reports of the amount of time and energy typically spent on solitary activities (thinking of ways to perform jobs better), and the other set of items dealt with time and energy spent on attempting to control the work of others. Each of these two workaholism subscales contained four items. On the surveys administered to respondents (addressed below), the following sentences preceded the eight items themselves. Please indicate how much time and energy you spend on each of the job related activities listed below. Use the following scale: 1 = None or not applicable; 2 = Little time and energy spent; 3 = Moderate time and energy spent; 4 = A considerable amount of time and energy spent; 5 = A very large amount of time and energy spent. For your job, how much time and energy do you now spend on the following activities? Three items in the 4-item Non-Required Work scale asked respondents how much time and energy they spend on thinking about ways to improve their work,

Workaholism: Scale Development

97

and the fourth item asked about time and energy spent on initiating projects. Specific items that followed the sentences presented above were: (1) Thinking of ways to improve the quality of work provided to customers and/or coworkers; (2) Thinking of ways to improve the quantity of work provided to customers and/or coworkers; (3) Thinking of ways to be more productive; and (4) Taking responsibility for initiating assignments and projects. As discussed earlier, we assumed that most (if not all) employees would typically be required to perform assigned tasks but would not generally be required to spend time and energy in such thought patterns or in being responsible for initiating work. The four items designed to assess Control of Others reflect the interpersonal and intrusive nature of workaholism. Three of these items asked about the time and energy respondents typically spend on: (1) Fixing problems created by other people; (2) Checking on the accuracy of other peoples work; and (3) Taking responsibility for the work of other people. Under most circumstances, even supervisors seem unlikely to be obligated to spend a considerable amount of time and energy taking responsibility for others work. High scores on this item thus seem to reflect time and energy spent on discretionary activitiesthus workaholism. The fourth item in this scale asked about time and energy spent on dealing with crisis situations. We included this item because a crisis seems unlikely to be experienced by an individual in isolation and thus has an important interpersonal component. Moreover, some observers have suggested that workaholics may be inclined to create crises in an attempt to spend more time at work and to make work more interesting. Machlowitz (1980, pp. 5657), for example, noted that individuals who enjoy fighting fires might sometimes want to start fires or to fan the flames of existing fires. Conceptually, the control of others facet of workaholism resembles organizational citizenship behaviors (Scott et al., 1997). Such behaviors include almost anything beneficial to an organization or its employees (Organ, 1994). Workaholic attempts to control others, however, could be beneficial (if welcomed by others) or harmful (if resented by those being helped; Scott et al., 1997). Operationally, we did not think it prudent to assess control of others based on whether or not attempts at control were interpreted as helpful (and were thus citizenship) or harmful (and were thus simply intrusive). Such an operationalization might imply that a given pattern of behavior on the part of one employee might be interpreted differently from the same pattern of behavior on the part of another employee. Certainly, our operationalization of control of others could include helpful citizenship behaviors, but the focus on behavior suggests that actual tendencies to get involved in the work of others is the critical factor in determining workaholism and not necessarily how that involvement is regarded by others.

98

Mudrack and Naughton

RESEARCH QUESTIONS We now report the results of an empirical study that used both of these newly developed workaholism measures (Non-Required Work and Control of Others). This study was designed to answer some fundamental questions about the construct validity of these scales. For example, are scores on the scales internally consistent? Does a 2-factor model in which the two workaholism scales are positioned as distinct constructs provide a better fit to the data than a 1-factor model in which all eight items are regarded as indicators of a single workaholism construct? Do scores on the workaholism scales relate as expected to other relevant variables and also in different ways to these variables? Briggs and Cheek (1986) suggested that subscales of a potentially unitary construct might be considered important to the extent that they were conceptually meaningful and empirically useful; that is, differentially related to other measures. If our eight workaholism items indeed assess two distinct constructs, then additional evidence of the construct validity of these two dimensions would be provided by differential patterns of relationships with other measures. For example, if scores on the Non-Required Work and Control of Others scales related similarly to other variables, then the practical necessity of separating these scales in research might not seem obvious. However, if scores on the workaholism scales related in different ways to scores on some other variables, then these scales might seem important and worth examining separately in research applications. Specifically, we examined relationships between the two workaholism indicators developed in this study and a number of work attitudes and personality measures. The specific variables included in the study are described below and were included because similar variables had been featured in other empirical studies of workaholism, because such variables seemed likely (on an a priori basis) to shed light on possible causes or consequences of workaholic behavioral tendencies, or because we expected the variables to provide evidence of empirical utility (e.g., a significant relationship with one workaholism scale, but a nonsignificant relationship with the other; Briggs & Cheek, 1986). Although certain results were anticipated, we advanced no formal hypotheses concerning relationships between workaholism scales and other variables. Our purpose was simply to provide an informative picture of the relations between the workaholism scales and other widely used measures.

Workaholic Behaviors and Work Attitudes Job involvement refers to the degree to which individuals psychologically identify with their present job (Kanungo, 1982). Spence and Robbins (1992) reported positive correlations between job involvement (an 8-item scale devel-

Workaholism: Scale Development

99

oped by those authors) and scores on their three workaholism scales (see also Burke, 1999a, 1999b). We expect that job involvement will correlate positively with Non-Required Work in light of the apparent disinterest of highly job involved persons in nonwork activities (e.g., Cohen, 1995; Ettington, 1998; Porter, 1996; Rosin & Korabik, 1995). Persons for whom pursuits outside of work hold little appeal might seem inclined to think about their jobs when away from work. Recent evidence also suggests positive associations between job involvement and citizenship behaviors (Cohen, 1999; Hoffi-Hofstetter & Mannheim, 1999). These associations suggest that highly involved employees may be inclined to assist others on the job. In light of the possibility that assistance represents attempts to exert control over others, we expected a positive relationship between job involvement and Control of Others. Conceptual writings on the subject of workaholism differ in the extent that workaholic employees are thought to enjoy their work. Some such persons claim to work because they love to work (although they may be suffering from acute denial and rationalization; e.g., Scott et al., 1997), while others seem grimly determined to work despite an absence of satisfaction or fun. In response to these apparent differences among workaholics, Spence and Robbins (1992) categorized persons who scored highly on both work involvement and feeling driven to work as enthusiastic workaholics if they also scored highly on work enjoyment and as workaholics if their work enjoyment scores were low. On balance, however, research findings suggest moderate positive relationships between work involvement and work enjoyment and weak relationships at best between feeling driven to work and work enjoyment (Burke, 1999a, 1999b; Spence & Robbins, 1992). The positive relationships between work involvement and work enjoyment are not unexpected in light of the positive relationships observed regularly in the literature between job involvement and job satisfaction (e.g., Furnham, Brewin, & OKelly, 1994; Morrow & Crum, 1998) and the possibility that work involvement, like job involvement, is a work attitude (e.g., Scott et al., 1997). In summary, we did not anticipate a strong directional relationship between either of our workaholic behaviors scales (Non-Required Work, Control of Others) and either job satisfaction or self-reports of fun at work. Spence and Robbins (1992) reported positive relationships between two of their workaholism measures (work involvement and feeling driven to work) and scores on their own 9-item measure of job stress. Burke (2000) found that stress correlated positively and significantly only with feeling driven scores. With regard to the workaholic behaviors scales at issue in this study, we anticipated a positive relationship between job stress and Control of Others but no clear directional relationship with Non-Required Work. High scorers on the Control of Others scale are inclined to get involved in various problems and crises at work. Regardless of whether or not such problems are largely self-created

100

Mudrack and Naughton

(e.g., Machlowitz, 1980), their mere existence seems likely to be associated with stressful reactions. In contrast, there seems nothing particularly stressful about thinking about ways to perform ones job better. We also anticipated that the presence of problems and crises at work would tend to be associated with conflict between the work and nonwork spheres of life. High scorers on the Control of Others scale may, after spending a full day dealing with problems and crises, be irritable at home or be too tired for other activities. High scorers on the Non-Required Work scale seem unlikely to suffer from irritability and fatigue merely by thinking about their jobs. Accordingly, we expected a positive relationship between Control of Others scores and work-nonwork conflict scores. Employees who intrude on the work of others and who attempt to control others inappropriately may tend to score highly on a measure of role conflict. In a sense, such employees are overstepping their authority and assuming a quasi-supervisory role that may not be accepted by some others. Role conflict also involves a sense that things at work should be done differently, and that existing policies sometimes have to be defied in order for work to be done properly. For highly workaholic employees, this desire to do things differently and properly may take the form of active intrusion in the work of others and beliefs that they themselves should be in charge. Employees who score highly on Non-Required Work, however, seem unlikely to score highly on role conflict simply by thinking about their jobs at their own discretion. Although they may tend to experience high role conflict, it is less clear that high scorers on the Control of Others scale will necessarily suffer from role ambiguity. On the one hand, employees who intrude on the work of others are behaving in ways that do not conform with the requirements of their positions. Such employees, as a result, may truly be unaware of what their responsibilities are or how much authority they actually have at work. On the other hand, employees who spend their time checking up on others and fixing the problems of others would seem to require a degree of certainty and conviction in the essential rightness of their intrusive work habits (cf. Baechler, 1996). Role ambiguity survey items do not ask, for example, whether one behaves correctly with regard to ones actual level of responsibility or authority, but rather the degree to which respondents are certain about these things. Respondents may be certain without necessarily being correct. As Control of Others scores could seemingly be both positively and negatively related to role ambiguity, we expected no directional relationship. As discussed earlier, we argued that number of hours worked per week would be an imperfect indicator of workaholism. Employees, after all, could work large numbers of hours because these are required or essential in order to accomplish assigned tasks. Nonetheless, workaholic employees seem likely (all other things being equal) to work more hours per week than nonworkaholic

Workaholism: Scale Development

101

employees. Therefore, a possible indicator of construct validity in our workaholism measures would be positive associations with the typical number of hours spent on work activities. In particular, employees who spend time thinking about how to do their work better (i.e., who receive high scores on the Non-Required Work scale) would likely tend to work more hours than employees who spend little time on such pursuits. Of course, such thinking does not necessarily have to occur in ones place of employment. Workaholism, by definition, incorporates the notion of working in ones discretionary time and working even when not physically at work. Accordingly, Non-Required Work scores may not necessarily correlate positively and significantly with the number of hours actually spent at ones place of employment. Therefore, in order to test fairly the predicted positive association between Non-Required Work and number of hours worked, we asked respondents to estimate their time spent at work broadly. Specifically, we asked On the average, how many hours would you say you spend per week engaged in job/career related activities, including time spent at the office, time spent traveling to/from work, time spent working at home on job activities, and time spent in graduate school activities? Parenthetically, all respondents both worked full time and were enrolled as part-time students seeking to earn a Master of Business Administration (MBA) degree (as described below). This broadly defined estimate of the number of hours worked per week should have correlated positively and significantly with Non-Required Work scores. Employees who attempt to control the work of others may not necessarily require large numbers of hours in order to be successful. Therefore, we did not necessarily anticipate a positive and significant correlation between Control of Others scores and hours worked.

Workaholic Behaviors and Personality Traits High scorers on the Control of Others scale want to influence others at work and to solve their coworkers problems. The interpersonal aspects of controlling others suggest that such persons may also tend to score highly on personality traits that also have implications for interpersonal relationships. We examined two such traits in this investigation: dominance and self-monitoring. Dominant individuals are inclined to assume leadership and influence roles and to be controlling (Jackson, 1984; Steers & Braunstein, 1976). We expected a positive relationship between dominance and Control of Others. Another personality trait with interpersonal implications is self-monitoring. High self-monitors are keenly aware of the moods and reactions of other people and are inclined to alter their own behavior in order to elicit the approval of others. High selfmonitoring seems useful in a number of social contexts such as leadership (e.g.,

102

Mudrack and Naughton

Zaccaro, Foti, & Kenny, 1991), romantic popularity (Speed & Gangestad, 1997), and tips received by restaurant staff (Lynn & Simons, 2000). In the absence of skillful persuasion and care, attempts to control the work of others might be interpreted negatively by the targets of control. Therefore, high self-monitoring may also be useful in order for attempts at control to be successful (or at least not to be resisted actively). We thus anticipated a positive relationship between Control of Others and self-monitoring. Other traits may be associated with tendencies to do more work than absolutely necessary on the job. For example, the hardworking qualities of highneed achievers might lead them to go above and beyond the call of duty at work. A desire for autonomy and independence might produce tendencies to engage in unnecessary work because one is able to be autonomous when spending time and energy thinking about ways to perform work better. Individuals also may engage in unnecessary work out of a desire to be liked and accepted by others who may approve of and appreciate such tendencies. Baechler (1996), for example, discussed the workaholics need to be needed by others. In short, we anticipated that Non-Required Work scores would correlate positively with achievement, autonomy, and affiliation needs. We also explored relationships between our workaholism scales and self-esteem in case workaholic individuals behave as they do to compensate for a poor self-image and feelings of inferiority. We did not necessarily anticipate significant associations, but wanted to investigate the possibility that workaholism has seemingly pathological origins.

METHOD Participants We administered self-report surveys containing items assessing the variables described below to 278 full-time employed individuals in a major metropolitan area in the United States. These employees worked in a variety of industries (e.g., automotive, banking, health care) and attended classes as part of an evening MBA program (mean age = 29.9 years, SD = 5.6 years; 163 men, 115 women; 116 not married, 162 married; 127 managers, 151 non-managers). On average, these respondents had been working for their current organization for 5.2 years. Respondents remained anonymous and earned no course credit for survey completion. An additional 44 surveys were completed by students who did not work full time at paying jobs. We did not use the survey data provided by these respondents in this investigation, however. All 278 respondents completed the two workaholism scales and the measures of work attitudes described below. A subset of these respondents (n = 183)

Workaholism: Scale Development

103

from four separate class sections completed the personality measures described below, as well as the workaholism and work attitudes scales. Sample items in each scale are provided in Table 1.

Measures of Work Attitudes Job involvement was assessed with the 9-item version of the Kanungo (1982) scale recommended by Paterson and ODriscoll (1990; = .83). We assessed job satisfaction with the 3-item scale of Hackman and Lawler (1971; = .79), and fun at work with a 4-item scale created for this investigation (see Table 1 for items). Coefficient alpha in these data was .76. We assessed job stress with the 4-item scale of Motowidlo, Packard, and Manning (1986; = .86) and work-nonwork conflict with the 8-item scale developed by Kopelman, Greenhaus, and Connolly (1983; = .88). Role conflict ( = .80; eight items) and role ambiguity ( = .78; six items) were assessed with the scales developed by Rizzo, House, and Lirtzman (1970). All of the above measures appeared in a 7-point response format ranging from strongly disagree (1 point) to strongly agree (7 points). Higher scores on all measures suggest higher levels of the work attitude at issue (e.g., higher job involvement or role ambiguity). Unless otherwise indicated, the personality measures described below also used the preceding 7-point response format. For example, we assessed dominance in this format with five items from the Manifest Needs Questionnaire (MNQ; Steers & Braunstein, 1976; = .80). The MNQ may also be used to assess achievement, autonomy, and affiliation. In light of the generally poor reliability estimates associated with these MNQ scales, however (e.g., Mannheim, Baruch, & Tal, 1997; Moorman & Podsakoff, 1992; Schneer & Chanin, 1987; Young & Parker, 1999), we elected to use the lengthier scales (16 items each) from the Personality Research Form (PRF Form E; Jackson, 1984) to assess the needs for achievement ( = .74), autonomy ( = .69), and affiliation ( = .72). These PRF scales featured a true-false response format, and scores could range from 0 to 16. Self-monitoring was assessed with the 18-item scale of Snyder and Gangestad (1986; = .70). This scale appeared in a true-false response format. The surveys also included the 10-item Rosenberg (1965) selfesteem scale ( = .82).

RESULTS Coefficient alpha estimates for the two workaholism scales in these data were .74 for Non-Required Work and .82 for Control of Others. The mean item correlation was .41 for Non-Required Work, and .37 for Control of Others.

104 Table 1. Sample Survey Items

Mudrack and Naughton

Job Involvement I am very much involved personally in my job. I live, eat, and breathe my job. Job Satisfaction Generally speaking, I am satisfied with my job. I frequently think of quitting my job (scoring reversed). Fun at Work I have fun at work. We are always laughing or joking about something at work. I hear the funniest jokes and stories at work. I find very little to laugh about at work (scoring reversed). Job Stress I feel a great deal of stress because of my job. Very few stressful things happen to me at work (scoring reversed). Work-Nonwork Conflict Because my work is demanding, at times I am irritable at home. After work, I come home too tired to do some of the things Id like to do. Role Conflict I have to do things at work that should be done differently. I often have to buck a rule or policy in order to carry out an assignment at work. Role Ambiguity I feel certain about how much authority I have at work (scoring reversed). I know exactly what is expected of me on the job (scoring reversed). Achievement I enjoy difficult work. I try to work just hard enough to get by (scoring reversed). Affiliation People consider me to be quite friendly. Sometimes I have to make a real effort to be sociable (scoring reversed). Autonomy I would like to have a job in which I didnt have to answer to anyone. My greatest desire is to be independent and free. Dominance I strive to be in command when I am working in a group. I avoid trying to influence those around me to see things my way (scoring reversed). Self-Monitoring I can only argue for ideas which I already believe (scoring reversed). Im not always the person I appear to be. Self-Esteem On the whole, I am satisfied with myself. I feel that I have a number of good qualities.

Briggs and Cheek (1986) suggested that an optimal level of homogeneity within survey items occurs when the mean interitem correlation lies between .20 and .40. In this range, a single total score represents the complexity of the items adequately without being overly redundant. Scores on these scales correlated positively but moderately (r = .25, p < .05). This evidence suggests that the for-

Workaholism: Scale Development

105

mat of these scales is appropriate and that these scales do not assess the same underlying construct. In order to provide more definitive evidence for the preceding assertion, we tested the factor structure of the eight workaholism items using LISREL 8 reskog & So rbom, 1996). Specifically, we compared the fit of three models. (Jo The first model (Model A) was a null model in which the existence of no factor structure is assumed (all relationships among the eight items were set equal to zero). Model B consisted of one general factor in which the eight items were forced to load on a unitary workaholism factor. If Model B provided a good fit to the data, this would suggest that the items in our two workaholism scales do not assess separate and distinct constructs and that these items should perhaps be combined in our analyses. Model C (the third of the three models tested) had two factors of four items each; that is, the factor structure that we implicitly assumed when developing the items. Specifically, four items assessed Non-Required Work and four items assessed Control of Others. Five fit indices relevant to these models are provided in Table 2. These were the Akaike Information Criterion (AIC; Akaike, 1987), the Goodness-of reskog & So rbom, 1986), the Non-Normed Fit Index (NNFI; Fit Index (GFI; Jo Bentler & Bonett, 1980), the Incremental Fit Index (IFI; Bollen, 1989), and the Root Mean Square Error of Approximation (RMSEA; Steiger, 1990). Lower values for both the AIC and the RMSEA suggest a better fit of the model to the data than do higher values. Higher values for the GFI, NNFI, and IFI suggest a better fit than do lower values. Model C (2-factor solution) performed considerably better than Models A (null) and B (1-factor) in terms of fit. For example, the AIC value for Model C was substantially lower than the values for the other two models. The GFI, IFI, and NNFI for Model C approached or exceeded the threshold level of .90, and the RMSEA equaled the threshold value of .08 that suggests a reasonably good fit (Browne & Cudeck, 1993). In sum, this confirmatory factor analysis suggests

Table 2. Significance Test and Fit Indices for Confirmatory Factor Analysis of Scales Designed to Measure Two Workaholic Behavior Dimensions Model Null Model One Factor Model Two Factor Model A B C 2 490.55* 202.98* 63.02* DF 28 20 19 AIC 506.55 273.93 93.77 GFI 0.82 0.95 NNFI 0.45 0.86 IFI 0.61 0.91 RMSEA 0.20 0.08

Note. DF = degrees of freedom; AIC = Akaikes Information Criterion; GFI = Goodness of Fit Index; NNFI = Non-Normed Fit Index; IFI = Incremental Fit Index; RMSEA = Root Mean Square Error of Approximation. *p < .05.

106

Mudrack and Naughton Table 3. Descriptive Statistics and Correlation Matrix

Variables

Non-Required Work (1) Control of Others (2) .25* Job Involvement (3) .18* .31* Job Satisfaction (4) .13* .10* .28* Job Stress (5) .15* .49* .33* .18* Work-Nonwork Conflict (6) .19* .38* .51* .09 .62* Role Conflict (7) .10* .41* .16* .28* .51* .50* Role Ambiguity (8) .01 .18* .02 .45* .28* .11* .36* Fun at Work (9) .02 .13* .04 .28* .13* .12* .14* .24* Hours Worked (10) .20* .14* .17* .06 .27* .36* .26* .09 Age (11) .05 .01 .00 .16* .02 .00 .04 .09 .14* .06 .05 .09 .06 .05 .12* .05 Gender (12)a Marital Status (13)b .12* .06 .05 .20* .04 .09 .04 .01 .12* .15* .13* .09 .11* .13* .05 .01 Managerial Status (14)c Achievement Needs (15) .29* .17* .60* .38* .12 .24* .08 .01 Autonomy Needs (16) .23* .03 .04 .09 .01 .01 .17* .04 Affiliation Needs (17) .26* .10 .17* .10 .02 .10 .16* .05 Dominance Needs (18) .20* .35* .07 .09 .10 .17* .19* .00 Self-Monitoring (19) .06 .32* .08 .16* .15* .18* .49* .13* Self-Esteem (20) .11 .05 .29* .17* .19* .35* .10 .03 Mean 11.83 11.49 31.06 14.27 18.90 30.43 33.09 19.66 SD 3.13 3.09 9.60 4.24 5.17 10.00 8.51 5.94 Note. Sample size for correlations involving personality variables (Variables 15 to 20, inclusive) was 183. Otherwise, sample size was 278. *p < .05. a Gender was scored as 1 for men and 2 for women. b Marital Status was scored as 1 for not married (or equivalent) and 2 for married (or equivalent). c Managerial Status was scored as 1 for does not supervise other employees and 2 for supervises other employees.

that the eight workaholism items assess two distinct constructs, and should be separated in empirical applications. A correlation matrix is provided in Table 3. Correlations between most of the work attitudes assessed here and the Control of Others workaholism scale were generally larger than correlations with the Non-Required Work scale. On the whole, high scorers on the Control of Others scale were more likely than low scorers to be involved in their jobs, to report job stress and work-nonwork conflict, and to experience role conflict. Although positive and significant, correlations between Non-Required Work and job involvement, job stress, worknonwork conflict, and role conflict were relatively weak (lower than .20). These findings suggest that the two workaholism scales developed here have empirical utility (Briggs & Cheek, 1986). As expected, neither workaholism scale shared much variance with role ambiguity, job satisfaction, or fun at work.

Workaholism: Scale Development

107

10

11

12

13

14

15

16

17

18

19

20

.11* .12* .16* .06 .17* .13* .00 .19* .31* .02 .11* .07 .05 .18* .09 .14* .23* .05 .12 .14* .25* .04 .19* .10 .07 .27* .20* .05 .12 .14* 18.99 63.44 29.89 4.32 10.87 5.68

.12* .11* .06 .07 .01 .01 .14* .07

.03 .04 .11 .01 .05 .25* .08 .03 .10 .01 .02 .07 .05

.11 .01 .29* .17* .03 10.96 3.55

.13 .23* .27* .17* .16* .26* .09 .27* .21* .02 6.40 9.77 26.35 10.80 58.96 3.37 3.44 4.73 3.37 6.98

Employees with high dominance needs and high self-monitoring tendencies tended to attain high Control of Others scores. In contrast, persons who reported engaging in Non-Required Work seemed inclined to score highly on the needs for achievement, affiliation, and autonomy. Neither workaholism scale correlated significantly with self-esteem. As discussed earlier, we anticipated these differential patterns of relationships. With regard to demographic measures, neither workaholism scale shared much variance with either respondent age or gender. Not unexpectedly, NonRequired Work scores correlated positively and significantly ( p < .05) with selfreports of the number of hours spent per week on work activities as broadly defined. This positive association suggests construct validity in our workaholism measures.

DISCUSSION This article reported on the development and preliminary testing of two new scales designed to assess behaviorally based manifestations of workahol-

108

Mudrack and Naughton

ism. In developing these scales, we were acutely aware of the need for them to be consistent with conceptual definitions of workaholism and to assess behavioral tendencies rather than work attitudes. Our measures also had to circumvent the reported inclination of workaholic individuals to rationalize their work habits and to be in a state of denial about these. Finally, our measures needed to be applicable in multiple work settings and not be context-specific. Two distinct scales emerged from this developmental process. One of the two workaholism scales developed here assessed tendencies to spend considerable time and energy engaged in work activities that are unlikely to be technically required of most (if not all) employees. These nonrequired activities dealt primarily with thinking of ways to perform work better. Persons who received high scores on this Non-Required Work scale tended to work large numbers of hours each week and also tended to score highly on three personality measures (achievement needs, affiliation needs, and autonomy needs). The second of the two workaholism scales created for this investigation incorporated largely interpersonal aspects of workaholism because the behavior of workaholic employees often affects others profoundly. In particular, workaholic employees seem inclined to seek to influence and control the work of others (perhaps in order to be sure that this work is performed properly). The Control of Others scale assesses tendencies to take responsibility for others work, to fix problems created by others, to check on the accuracy of others work, and to respond to crises. Among our sample of respondents, high scorers on the Control of Others scale generally scored highly on two personality measures (dominance needs and self-monitoring) and four work attitudes (job involvement, job stress, work-nonwork conflict, role conflict). What is evident from the preceding discussion is that the pattern of relationships involving Non-Required Work scores was markedly different from the pattern involving Control of Others. Such evidence speaks directly to the conceptual and practical differences between the two workaholism scales (cf. Briggs & Cheek, 1986). The necessity of separating these scales in research applications was also suggested by the relatively weak correlation between the two scales and the fact that a model in which the two workaholism measures were positioned as distinct constructs provided a good fit to the data. The observed patterns of relationships also suggests that the two workaholism scales possess both convergent and discriminant validity. If scales possess convergent validity, then scores on these scales should relate as expected with scores on other scales. Evidence suggesting convergent validity was described above. If scales possess discriminant validity, then scores on these scales should share little variance with scores on scales that seem conceptually distinct. As expected, neither workaholism scale related strongly with self-esteem, role ambiguity, job satisfaction, or reported fun experienced at work. None of the corre-

Workaholism: Scale Development

109

lations between Non-Required Work and such work attitudes as job involvement, job stress, work-nonwork conflict, and role conflict were particularly strong. Control of Others scores were also not strongly linked with the needs for achievement, autonomy, or affiliation. Much more research is needed before we can conclude definitively that the workaholism scales developed here possess construct validity. Preliminary evidence, however, is indeed encouraging and seems to justify further research using these scales. It is our sincere hope that these scales may assist in achieving a deeper understanding of the important and relevant phenomenon of workaholism.

ACKNOWLEDGMENTS Thanks are owed to Brian Niehoff and Bill Turnley for helpful suggestions on an earlier version of this article.

REFERENCES
Akaike, H. (1987). Factor analysis and AIC. Psychometrika, 52, 317332. Baechler, M. (1996). Im Mary, and Im a workaholic. Inc., 18(5), 2930. Bentler, P. M., & Bonett, D. G. (1980). Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin, 88, 588606. Bollen, K. A. (1989). A new incremental fit index for general structural equation models. Sociological Methods and Research, 17, 303316. Briggs, S. R., & Cheek, J. M. (1986). The role of factor analysis in the development and evaluation of personality scales. Journal of Personality, 54, 106148. Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. A. Bollen & J. S. Long (Eds.), Testing structural equation models (pp. 136162). Newbury Park, CA: Sage. Burke, R. J. (1999a). Its not how hard you work but how you work hard: Evaluating workaholism components. International Journal of Stress Management, 6, 225239. Burke, R. J. (1999b). Workaholism in organizations: Measurement validation and replication. International Journal of Stress Management, 6, 4555. Burke, R. J. (2000). Workaholism in organizations: Psychological and physical well-being consequences. Stress Medicine, 16, 1116. Cohen, A. (1995). An examination of the relationships between work commitment and nonwork domains. Human Relations, 48, 239263. Cohen, A. (1999). The relation between commitment forms and work outcomes in Jewish and Arab culture. Human Relations, 54, 371391. Costello, J. (2000). Full-time dilemma: Economy bedevils workaholics group. Wall Street Journal, February 8, p. A1. Ettington, D. R. (1998). Successful career plateauing. Journal of Vocational Behavior, 52, 7288. Furnham, A., Brewin, C. R., & OKelly, H. (1994). Cognitive style and attitudes to work. Human Relations, 47, 15091521. Gini, A. (1998). Working ourselves to death: Workaholism, stress, and fatigue. Business and Society Review, 100101, 4546. Hackman, J. R., & Lawler, E. E. III. (1971). Employee reactions to job characteristics. Journal of Applied Psychology, 55, 259286.

110

Mudrack and Naughton

Hoffi-Hofstetter, H., & Mannheim, B. (1999). Managers coping resources, perceived organizational patterns, and responses during organizational recovery from decline. Journal of Organizational Behavior, 20, 665685. Jackson, D. N. (1984). Personality Research Form manual (3rd ed.). Port Huron, MI: Research Psychologists Press. Jo reskog, K. G., & So rbom, D. (1986). LISREL VI: Analysis of linear structural relationships by maximum likelihood and least squares methods. Mooresville, IN: Scientific Software, Inc. Jo reskog, K. G., & So rbom, D. (1996). LISREL 8: Users Reference Guide (2nd ed.). Chicago: Scientific Software. Kanungo, R. N. (1982). Measurement of job and work involvement. Journal of Applied Psychology, 67, 341349. Kopelman, R. E., Greenhaus, J. H., & Connolly, T. F. (1983). A model of work, family and interrole conflict: A construct validation study. Organizational Behavior and Human Performance, 32, 198215. Lynn, M., & Simons, T. (2000). Predictors of male and female servers average tip earnings. Journal of Applied Social Psychology, 30, 241252. Machlowitz, M. (1980). Workaholics: Living with them, working with them. Reading, MA: AddisonWesley. 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, 15371562. Moorman, R. H., & Podsakoff, P. M. (1992). A meta-analytic review and empirical test of the potential confounding effects of social desirability response sets in organizational behaviour research. Journal of Occupational and Organizational Psychology, 65, 131149. Morrow, P. C., & Crum, M. R. (1998). The effects of perceived and objective safety risk on employee outcomes. Journal of Vocational Behavior, 53, 300313. Motowidlo, S. J., Packard, J. S., & Manning, M. R. (1986). Occupational stress: Its causes and consequences. Journal of Applied Psychology, 71, 618629. Oates, W. (1971). Confessions of a workaholic: The facts about work addiction. New York: World. Organ, D. W. (1994). Personality and organizational citizenship behavior. Journal of Management, 20, 465478. Paterson, J. M., & ODriscoll, M. P. (1990). An empirical assessment of Kanungos (1982) concept and measure of job involvement. Applied Psychology: An International Review, 39, 293306. Porter, G. (1996). Organizational impact of workaholism: Suggestions for researching the negative outcomes of excessive work. Journal of Occupational Health Psychology, 1, 7084. Rizzo, J. R., House, R. J., & Lirtzman, S. I. (1970). Role conflict and ambiguity in complex organizations. Administrative Science Quarterly, 15, 150163. Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. Rosin, H., & Korabik, K. (1995). Organizational experiences and propensity to leave: A multivariate investigation of men and women managers. Journal of Vocational Behavior, 46, 116. Schaef, A. W., & Fassel, D. (1988). The addictive organization. San Francisco: Harper and Row. Schneer, J. A., & Chanin, M. N. (1987). Manifest needs as personality predispositions to conflicthandling behavior. Human Relations, 40, 575590. Scott, K. S., Moore, K. S., & Miceli, M. P. (1997). An exploration of the meaning and consequences of workaholism. Human Relations, 50, 287314. Seybold, K. C., & Salomone, P. R. (1994). Understanding workaholism: A review of causes and counseling approaches. Journal of Counseling and Development, 73(1), 49. Snyder, M., & Gangestad, S. (1986). On the nature of self-monitoring: Matters of assessment, matters of validity. Journal of Personality and Social Psychology, 51, 125139. Speed, A., & Gangestad, S. W. (1997). Romantic popularity and mate preferences: A peer-nomination study. Personality and Social Psychology Bulletin, 23, 928935. Spence, J. T., & Robbins, A. S. (1992). Workaholism: Definition, measurement, and preliminary results. Journal of Personality Assessment, 58, 160178. Steers, R. M., & Braunstein, D. N. (1976). A behaviorally-based measure of manifest needs in work settings. Journal of Vocational Behavior, 9, 251266.

Workaholism: Scale Development

111

Steiger, J. H. (1990). Structural model evaluation and modification: An interval estimation approach. Multivariate Behavioral Research, 25, 173180. Young, S. A., & Parker, C. P. (1999). Predicting collective climates: Assessing the role of shared work values, needs, employee interaction and work group membership. Journal of Organizational Behavior, 20, 11991218. Zaccaro, S. J., Foti, R. J., & Kenny, D. A. (1991). Self-monitoring and trait-based variance in leadership: An investigation of leader flexibility across multiple group situations. Journal of Applied Psychology, 76, 308315.

You might also like