You are on page 1of 16

Attitudinal and behavioral consequences of work-

family conflict and family-work conflict

Does gender matter?

Abstract

Purpose – The purpose of this paper is to develop and test a conceptual model to examine the effects of
work-family conflict, family-work conflict, and emotional exhaustion on job performance and turnover
intentions. The paper also aims to investigate the role of gender as a moderator of the posited
relationships.

Design/methodology/approach – A sample of frontline hotel employees in Turkey serves as the study


setting. Data were collected via self-administered questionnaires. A total of 723 usable responses were
obtained.

Findings – The results show that employees facing conflicts originating from their work (family) and
family (work) roles become emotionally exhausted. These two forms of interrole conflicts are also
significant predictors of frontline employees' turnover intentions. Gender moderates several of the
relationships proposed in this paper.

Practical implications – Turkish hotels will benefit from establishing a family-supportive work
environment to lessen the detrimental impact of conflicts in the work-family interface on frontline
employees' emotional exhaustion and job outcomes. A dual (i.e. gender-specific) approach appears to
hold promise in managing frontline employees.

Originality/value – When these results are compared to the results of studies conducted in western
countries, a number of similarities become evident. These similarities broadly suggest that research
findings derived from western countries are generalizable into a culturally different setting, and support
the premise that as traditional gender roles continue to expand and change, a convergence of findings in
work-family research takes place cross-culturally.

Article Type:
 

Research paper

Keyword(s):
 

Family; Role conflicts; Employee attitudes; Gender; Hotels; Turkey.

Journal:
 

International Journal of Service Industry Management


Volume:
 

19

Number:
 

Year:
 

2008

pp:
 

7-31

Copyright ©
 

Emerald Group Publishing Limited

ISSN:
 

0956-4233

Introduction

In an era of intense competitive pressures, dwindling resources and discerning customers, service
organizations including hotels, the setting of this study, realize that attaining customer satisfaction through
delivery of quality services is a key to their survival and success. Accordingly, hotel executives engage in
multi-pronged actions ranging from innovative marketing and operational strategies to pointed competitive
analyses to better serve their customers and improve their profitability (Kashyap and Bojanic, 2000; Yasin
and Yavas, 2001; Yavas and Babakus, 2005). Astute hotel executives also recognize that no strategy
aimed at motivating, satisfying and retaining external customers can be considered complete unless it
includes programs for reaching and winning over internal customers. To such executives, retention of
motivated, satisfied and committed frontline employees who can consistently deliver high-service quality
is as important to business success as customer satisfaction and retention (Bowen and Ford,
2004; Kotler et al., 2006).

This is not surprising as frontline employees in service organizations, whether they actually render the
service or simply interact with customers face-to-face or voice-to-voice, are the main actors in the delivery
of service quality (Bettencourt and Brown, 2003; Bettencourt and Gwinner, 1996). However, frontline
employees are often underpaid, typically work long hours, irregular schedules and carry heavy workloads
(Babin and Boles, 1998; Singh, 2000). These coupled with dramatic changes in today's social (e.g. entry
of more women to the workforce and rise in the number of dual-career couples; increased responsibilities
to care for older, infirm parents/relatives) and economic conditions (e.g. downsizing resulting in “lean and
mean” organizations necessitating the surviving employees to work even longer hours) give rise to
heightened conflicts between work and family roles (Halpern, 2005; Posig and Kickul, 2004).

The difficulties frontline employees experience in balancing the demands of their multiple roles result in
poor morale and can impact their performance and commitment. Such employees can become
emotionally exhausted and think of leaving their organization (Boleset al., 1997; Karatepe, 2006). These
adverse consequences can indeed become quite costly. Emotional exhaustion undermines frontline
employees' performance and hinders effective customer service (Babakus et al., 1999; Cropanzano et al.,
2003; Wright and Cropanzano, 1998). Employees with intentions to leave may reduce their productivity
prior to quitting (Halpern, 2005). Furthermore, the costs of actual turnovers (e.g. recruitment and training
of additional staff, overtime payments to existing staff to alleviate shortages, disrupted service, increased
turnover among remaining staff who feel pressured and overworked) can reach exorbitant proportions
(Franket al., 2004; Hendrie, 2004).

Purpose

Against this backdrop, the purpose of the present study is twofold. First, it develops and tests a model of
frontline employees' interrole conflicts (i.e. work-family conflict and family-work conflict) and their impacts
on emotional exhaustion, job performance and turnover intentions, and between emotional exhaustion
and job performance and turnover intentions. Second, it examines the potential moderating role of gender
on these relationships. We test our model by using data collected from frontline hotel employees in
Turkey.

A study addressing these issues is relevant and significant. First, despite the recognition that interrole
conflicts can have detrimental effects on job outcomes, the bulk of prior studies examine the separate
influences of work-family conflict (Babin and Boles, 1998; Boleset al., 1997; Carlson and Perrewé
1999; Parasuraman and Simmers, 2001) and to a lesser extent the impact of family-work conflict
(Behson, 2002; Butler and Skattebo, 2004; Parker and Griffin, 2002) on job outcomes. Few studies
(Netemeyer et al., 2004; Posig and Kickul, 2004) consider the effects of both work-family conflict and
family-work conflict simultaneously on emotional exhaustion and job outcomes.

Second, despite the evidence that interrole conflicts may vary by gender (Boles et al., 2003; Peeters et
al., 2005), empirical research on the moderating role of gender on the relationships between conflicts in
the work-family interface and emotional exhaustion and job outcomes is very sparse (Eby et al.,
2005; Poelmans et al., 2005). Third, the preponderance of empirical research on the issues surrounding
work-family conflict and family-work conflict has been conducted in such developed countries as the USA
(Boles et al., 2003; Frone et al., 1997; Netemeyer et al., 2005), Finland (Kinnunen and Mauno,
1998; Mauno and Kinnunen, 1999), and Australia (Elloy and Smith, 2003). In fact, as Barnett and Hyde
(2001) forcefully state, much of what we know on the topic is based on studies of white, middle-class,
urban Americans. Since, issues of work and family are closely associated with a society's culture and
gender roles (Aryee et al., 1999; Aycan and Eskin, 2005), extension of the research stream to other
societies is imperative for richer insights (Netemeyer et al., 2004) and to allow for informed judgments
regarding the generalizability of western findings.

Besides filling in these informational voids and serving as a frame of reference for future research, the
findings of the study may prove useful to managers. An understanding of presence (or absence) of male-
female differences in the relationships between interrole conflicts, emotional exhaustion, job performance
and turnover intentions is crucial for managers in determining if an undifferentiated or dual (i.e. gender-
specific) approach is warranted in managing frontline employees (Moncrief et al., 2000).

In the next section, we present our conceptual model and the relevant literature leading to our specific
research hypotheses. This is followed by discussions of the method and results of the empirical study we
conducted in Turkey, which bridges east and west geographically and culturally. We conclude the paper
with the implications of the results and avenues for future research.
Conceptual framework and hypotheses

Figure 1 shows the conceptual model guiding our study. Based on extant literature, we contend that
interrole conflicts emanating from the demands of the two universal domains of adult life, work and family,
lead to emotional exhaustion. Furthermore, we posit that employees experiencing work-family and family-
work conflicts become poor performers and think of leaving their organization. Our model also proposes
that emotional exhaustion results in poor job performance and a higher propensity to leave the
organization. Hence, the key underlying premise of our model is that work-family conflict and family-work
conflict impact the two critical job outcomes of performance and turnover intentions both directly as well
as indirectly through the mediating role of emotional exhaustion.

In addition, we contend that there may be gender-based differences in the relationships depicted in our
model. Thus, we examine the moderating role of gender on the relationships between work-family conflict
and family-work conflict, and emotional exhaustion and job outcomes. We also examine the role of
gender as a moderator of the relationships between emotional exhaustion and job outcomes. As shown
in Figure 1, we include age, education, tenure, marital status, and the number of children as control
variables in our conceptual model since they may influence the key constructs and confound the
relationships posited in the model.

Consequences of work-family conflict and family-work conflict

Work-family conflict and family-work conflict are distinct, but conceptually related forms of interrole conflict
(Frone et al., 1992;Netemeyer et al., 1996). Work-family conflict refers to “a form of interrole conflict in
which the general demands of, time devoted to, and strain created by the job interfere with performing
family-related responsibilities” and family-work conflict refers to “a form of interrole conflict in which the
general demands of, time devoted to, and strain created by the family interfere with performing work-
related responsibilities” (Netemeyer et al., 1996, p. 401).

Both forms of conflict basically result from an individual's attempts to meet an overabundance of demands
emanating from the home/family and work domains in which the individual operates (Boles et al., 2001).
The demands coming from one domain make performance of roles in the other domain more difficult. In
our model, we contend that three consequences of work-family and family-work conflicts are emotional
exhaustion, poor job performance and higher turnover intentions.

Emotional exhaustion

Emotional exhaustion is the first stage of the burnout syndrome (Cordes and Dougherty, 1993; Maslach
and Jackson, 1981) and it occurs when an individual faces seemingly overwhelming demands on his/her
time and energy. Depletion of emotional resources and a lack of energy characterize emotional
exhaustion (Gaines and Jermier, 1983). Precepts of three well-endorsed theoretical frameworks (interrole
conflict theory, identity theory, and conservation of resources – COR theory) suggest that work-family and
family-work conflicts can lead to emotional exhaustion.

First, according to the interrole conflict theory, since work-family and family-work conflicts arise from
opposing pressures due to participation in different roles (Greenhaus and Beutell, 1985), individuals are
likely to experience emotional exhaustion when they try to meet the demands of work and family roles.
Second, the identity theory suggests that individuals hold multiple role identities and invest themselves in
maintaining the identities that are salient to them (Thoits, 1991). Under these circumstances, work-family
conflict or family-work conflict prevents individuals from fulfilling the demands of their work or family role,
and lead to heightened emotional exhaustion. Third, the COR theory suggests that individuals seek to
acquire, maintain and preserve certain resources (e.g. time, energy) (Hobfoll, 1989). Since, individuals
are likely to lose such resources while juggling both work (family) and family (work) roles, the COR theory
leads to the conclusion that conflicts arising from work (family) interfaces should result in emotional
exhaustion.
Emotional exhaustion appears to be prevalent among employees in people-oriented jobs (Babakus et al.,
1999; Karatepe, 2006) such as our study setting, and prior empirical research indicates that employees
who experience elevated levels of work-family conflict are more likely to be emotionally exhausted
(Bacharach et al., 1991; Boles et al., 1997; Mauno and Kinnunen, 1999). Likewise, in a recent
study, Posig and Kickul (2004) report that family-work conflict as well leads to emotional exhaustion.
Based on the preceding discussion and findings, we propose the following hypotheses:

H1a. Work-family conflict is positively related to frontline employees' emotional exhaustion.

H1b. Family-work conflict is positively related to frontline employees' emotional exhaustion.

Job performance

In this study, job performance is defined as “the level of productivity of an individual employee, relative to
his or her peers, on several job-related behaviors and outcomes” (Babin and Boles, 1998, p. 82). Work
and family are the two important domains of adult life. Problems associated with one domain (e.g. work)
spillover to the other domain (e.g. family) (Williams and Alliger, 1994), and detract from the limited
resources (e.g. time, energy) people have in fulfilling their multiple roles (Hobfoll, 1989). Empirical
research reveals that work-family conflict has a detrimental impact on performance (Aryee,
1992; Frone et al., 1997; Netemeyer et al., 2004). Anecdotal evidence by the Family and Work Institute as
well indicates that employees who cannot balance their work demands with home and family
responsibilities experience negative spillovers resulting in decreased job performance (Netemeyer et al.,
2003). There is also evidence indicating that family-work conflict decreases employees' work-related
performance (Frone et al., 1997; Netemeyer et al., 2004). Accordingly, we propose the following
hypotheses:

H2a. Work-family conflict is negatively related to frontline employees' job performance.

H2b. Family-work conflict is negatively related to frontline employees' job performance.

Turnover intentions

The COR theory contends that once employees recognize that they may not be able to cope with
difficulties arising from work-family conflict, they attempt to preserve their scarce resources by thinking
about leaving their current organization (Grandey and Cropanzano, 1999). Indeed, Allen et al.'s
(2000) meta-analytic inquiry suggests that turnover intentions is the job outcome most closely associated
with work-family conflict. This is a conclusion supported by more recent research (Anderson et al.,
2002; Boyar et al., 2003). Although limited in number, there is also empirical support that family-work
conflict and turnover intentions are related (Boyar et al., 2003). It appears that individuals who are more
sensitive to family life may be willing to meet their family demands at the expense of losing their role in
the workplace (Armour, 2002). In light of the aforementioned findings and discussion, we propose the
following hypotheses:

H3a. Work-family conflict is positively related to frontline employees' turnover intentions.

H3b. Family-work conflict is positively related to frontline employees' turnover intentions.

Consequences of emotional exhaustion

Emotional exhaustion is not only a serious outcome in itself resulting in depletion of one's valuable
resources as discussed in the COR theory, but it is also a significant determinant of such job outcomes as
job performance and turnover intentions. Individuals who no longer possess the adequate resources to
cope with emotional exhaustion display decreased job performance and a higher propensity to leave the
organization. For instance, Babakus et al. (1999) reported that salespeople experiencing higher emotional
exhaustion demonstrated lower job performance. This finding is also supported by the results
of Cropanzano et al.'s (2003) and Wright and Hobfoll's (2004) recent studies.

Extant research also shows that emotional exhaustion is a critical determinant of turnover intentions
(Boles et al., 1997; Cropanzano et al., 2003). A more recent Turkish study reached the same conclusion
and indicated that emotional exhaustion exerts a strong positive impact on frontline employees' intentions
to leave their banks (Karatepe, 2006). Thus, we posit that:

H4. Emotional exhaustion is negatively related to frontline employees' job performance.

H5. Emotional exhaustion is positively related to frontline employees' turnover intentions.

Gender-based differences

Work-family conflict and family-work conflict – emotional exhaustion relationships

Owing to their role as primary caretaker of their households and the nearly universal demands of
motherhood (Wood and Eagly, 2002), women value more and assign a higher priority to their family roles
than their work roles. On the other hand, men attach higher priorities to their job-related responsibilities
(Martins et al., 2002). In an era of excessive job demands, irregular and incompatible schedules, and long
work hours (Spinks, 2004), because of the higher priority they place on family roles, women employees
experience greater family-work conflict (Wayne et al., 2004). Also experiencing greater difficulty in coping
with conflicts emanating from work and non-work domains, women are more likely to suffer from
emotional exhaustion. There is some empirical evidence to support this premise. For instance, Posig and
Kickul (2004) found that the relationship between family-work conflict and emotional exhaustion was
stronger among female employees compared with male employees. Demerouti et
al. (2005) demonstrated that the relationship between partners' ratings of work-family conflict and
exhaustion was higher among women than men. Thus, we hypothesize that:

H6a. The positive relationship between work-family conflict and emotional exhaustion is stronger among
female frontline employees than male frontline employees.

H6b. The positive relationship between family-work conflict and emotional exhaustion is stronger among
female frontline employees than male frontline employees.

Work-family conflict and family-work conflict – job outcome relationships

A general observation surfacing from studies examining the role of gender in managerial and marketing
contexts (Babin and Boles, 1998; Eagly et al., 1995; Iacobucci and Ostrom, 1993; Martins et al., 2002) is
that men are more task- or goal-oriented (agentic) and women are more relationship-oriented
(communal). This distinction between the two sexes impacts the types of things each gender values in
various aspects of their lives including their jobs. Accordingly, one would expect interrole conflicts to affect
the job performances of relationship-valuing female employees more negatively in interaction-oriented
frontline jobs where employees are expected to deal with customer requests and complaints through long
work hours. Demands placed on their time and energy from opposing domains are more likely to prevent
women employees from meeting management's performance expectations (Scott, 1997).

In today's global work environment, there is an increasing involvement of women in the workforce
(McElwain et al., 2005). However, working wives and husbands do not equally share home and/or family
responsibilities. Working wives are not only expected to shoulder a greater portion of family
responsibilities but they are also expected to be supportive of their husbands' job activities (Bedeian et
al., 1988; Noor, 2003). Facing higher levels of family demands, women are susceptible to higher levels of
family-work conflict (Aryee et al., 1999; Boles et al., 2003). Not surprisingly, a number of working wives
prefer to quit their jobs in corporate life to start their own small businesses where control over their
working hours allows them to spend more time in fulfilling family responsibilities (Eagle et al., 1997). The
above discussion leads to the following hypotheses:

H6c. The negative relationship between work-family conflict and job performance is stronger among
female frontline employees than male frontline employees.

H6d. The negative relationship between family-work conflict and job performance is stronger among
female frontline employees than male frontline employees.

H6e. The positive relationship between work-family conflict and turnover intentions is stronger among
female frontline employees than male frontline employees.

H6f. The positive relationship between family-work conflict and turnover intentions is stronger among
female frontline employees than male frontline employees.

Emotional exhaustion – job outcome relationships

The COR theory suggests that depletion of emotional and other resources is a characteristic of emotional
exhaustion (Hobfoll, 1989;Wright and Hobfoll, 2004), and research shows that women experience higher
levels of emotional exhaustion than men (Gaines and Jermier, 1983). Once individuals do not have the
necessary resources to perform their roles in multiple domains, they are forced to make tough choices.
Women have a more difficult time in trading off work for family and easing off their family responsibilities
at the expense of work (Posig and Kickul, 2004). Thus, less time devoted to work activities adversely
impacts women's performance in the workplace (Noor, 2003). Also to regain the resources (primarily
time) lost to work responsibilities, women are more likely to exhibit turnover intentions by thinking of
altogether quitting their jobs or by seeking new jobs that would better fit their expectations (Eagle et al.,
1997). The above discussion prompts our final set of hypotheses:

H6g. The negative relationship between emotional exhaustion and job performance is stronger among
female frontline employees than male frontline employees.

H6h. The positive relationship between emotional exhaustion and turnover intentions is stronger among
female frontline employees than male frontline employees.

Control variables

While their findings are mixed, several studies suggest that various demographic characteristics can
impact the key constructs in our study and may confound the relationships depicted in Figure 1 (Allen,
2001; Bekker et al., 2005; Noor, 2003; Stoeva et al., 2002; van Vegchel et al., 2004; Wayne et al., 2004).
Hence, to allow for better delineation of the relationships proposed in our model and to provide a more
rigorous test of the theoretical linkages, in our study we included several control variables. These were
age, education, tenure, marital status, and the number of children.

Methodology

Sample

Data for the study were collected from the frontline employees (e.g. food servers, front desk agents,
concierges, and bartenders) of three-, four-, and five-star hotels in Ankara, Turkey's capital. Information
we received from the Governorship of the City and the Directorate of Tourism indicated that, at the time of
the study, 5 five-star hotels, 10 four-star hotels, and 27 three-star hotels were operating in Ankara.
Managements of all these hotels were contacted prior to data collection and permission was granted by
all the five-star hotels, 9 four-star hotels and 23 three-star hotels. According to the managements of the
hotels, these hotels collectively employed 1,316 frontline employees. Questionnaires were then
distributed to all of these employees by the research team. Employees were given assurance of
confidentiality (that only the aggregate results would be shared with the managements of hotels) and
were requested to fill out the questionnaires in a self-administered manner. By the cut-off date for data
collection, a total of 723 usable questionnaires were retrieved by the research team, yielding a response
rate of 54.9 percent.

About 46 percent of the respondents were between the ages of 18-27, 40 percent between the ages of
28-37 and the rest were older than 37. The sample was fairly balanced in gender (53 percent male and 47
percent female) and marital status (52 percent married and 48 percent single/divorced). About 31 percent
of the respondents had graduated from two-year colleges and about 32 percent from four-year colleges.
Almost 2 percent of the respondents had graduate degrees. Approximately, 30 percent of the
respondents had secondary and high school education. The rest had primary school education. Three-
fourths of the respondents had tenures of five years or less and the rest had been with their hotel for six
or more years. While 59 percent of the respondents had no children, 19 percent had one child, 17 percent
two, 4 percent three and the rest more than three children.

Measurement

Multiple-item scales obtained from the relevant literature were used to operationalize the study constructs.
Specifically, five items each from Netemeyer et al. (1996) and Boles et al. (2001) were employed to
measure work-family conflict and family-work conflict. Emotional exhaustion was operationalized via eight
items from Maslach and Jackson (1981). Five items were adapted from Babin and Boles (1998)to
measure job performance. And three items were adapted from Boshoff and Allen (2000) to measure
turnover intentions. Responses to each of these items were elicited on five-point scales ranging from 5 = 
strongly agree to 1 = strongly disagree. Higher scores indicated higher work-family conflict, family-work
conflict, emotional exhaustion, job performance, and turnover intentions.

Age, education, tenure and the number of children were measured via five-point scales. Higher scores
indicated older age, better education, longer tenure, and more children. Marital status was coded as a
dichotomous variable (0 = single/divorced and 1 = married).

The survey instrument was initially prepared in English and then translated into Turkish via the back-
translation method (McGorry, 2000). To ensure that the item contents were cross-linguistically
comparable and generated the same meaning, two faculty members of a Turkish university fluent in both
languages further checked the questionnaire. Prior to administering in the field, the questionnaire was
pre-tested with a pilot sample of 30 frontline employees and no changes in the wording of the questions
were deemed necessary.

Results

Measurement results

The measures were initially subjected to exploratory factor analysis (principal components with oblique
rotation) and reliability assessment using the entire sample as well as the female and male sample data
separately. In each case, five-factor solutions with eigenvalues greater than 1.0 emerged and accounted
for 60, 63 and 59 percent of the variance in the combined, female and male sample data. The
overwhelming majority of the items loaded heavily on their respective underlying factors in all three factor
analyses. Coefficient α ranged from 0.74 to 0.88 for the combined sample, 0.68-0.89 for the female
sample, and 0.71-0.87 for the male sample. The list of items, sources of the scales, and scale reliabilities
are presented in the Appendix.
For a more rigorous psychometric assessment and to address measurement invariance across genders,
we used LISREL 8.51 (Joreskog and Sorbom, 1993) to conduct two-group simultaneous confirmatory
factor analyses following Fornell and Larcker (1981) andHair et al. (2006). An initial five-factor base
model, which allowed all model parameters to be estimated freely in both groups, failed to provide a
proper solution due to negative error variance estimates for two indicators. As a remedy, we partially
aggregated scale items by randomly splitting items in each scale into two sets and used the average of
each set as a composite indicator of its underlying construct (Bagozzi and Heatherton, 1994). Since,
turnover intentions scale had only three items, they were kept intact. The five-factor base model using the
composite indicators converged with a proper solution and showed a good fit to the data across groups
(χ 692=211.44, RMSEA = 0.076, NFI = 0.95, NNFI = 0.95, CFI = 0.97).

Next, by constraining factor loadings to be equal across groups, we tested for full metric invariance and
obtained a χ 75 2 value of 231.72, which indicates that full metric invariance is untenable due to significant
deterioration in model fit at the 0.05 level. However, following Hair et al. (2006) we were able to establish
partial metric invariance. This required freeing only one item in the turnover scale (It would not take much
to make me leave this hotel) and constraining all other indicator loadings to be equal across groups. The
resultingχ 74 2 value of 224.08 (compared to the base model χ 69 2=211.44) showed that model
deterioration was not significant, suggesting that the measures exhibited partial metric invariance. Finally,
when we imposed an additional constraint, the equivalence of covariances of the underlying constructs,
the resulting χ 89 2 value was 259.87. The difference between this test statistic and the result from the
partial metric invariance test (χ 74 2=224.08) showed a significant deterioration in the model fit, and
suggested a moderator role for gender.

After establishing partial metric invariance, we examined the average variance extracted (AVE) and
shared variance (Φ2) values for each underlying construct across groups to assess convergent and
discriminant validities of the measures (Fornell and Larcker, 1981). The AVEs ranged from 0.59 (turnover
intentions) to 0.79 (emotional exhaustion), and the shared variances (Φ2) ranged from a low of 0.001
(between job performance and turnover intentions) to a high of 0.58 (between work-family and family-
work conflict). These results collectively provide evidence of convergent and discriminant validity. Table
I provides correlations, means and standard deviations of the composite indicators of the model
constructs and control variables for female and male respondents.

Tests of the model and research hypotheses

Overall research mode

We first examined model fit and tested the overall research hypotheses (H1 through H5) using the
combined sample covariance matrix as input to LISREL 8.51 (Joreskog and Sorbom, 1993). The results
in Table II indicate that the model fits the data well (χ 64 2=172.77, RMSEA = 0.049, NFI = 0.97, NNFI = 
0.96, CFI = 0.98) and accounts for 23 percent of the variance in emotional exhaustion, 22 percent in job
performance, and 49 percent in turnover intentions.

A closer examination of the results for the combined sample in Table II reveals that work-family conflict
and family-work conflict both have significant (p<0.05) positive impacts on emotional exhaustion,
supporting H1a and H1b. The impact of work-family conflict on job performance is significant. But, the
influence is positive (opposite to the expected sign). Therefore, H2a is not supported by the data. On the
other hand, family-work conflict has a significant negative impact on job performance. This is consistent
with H2b. Both work-family conflict and family-work conflict, as predicted, have significant positive impacts
on turnover intentions. Thus, H3a and H3b are supported. While emotional exhaustion has a negative
impact on performance as predicted by H4, this effect is not significant at the 0.05 level. Hence, H4 is not
supported. The effect of emotional exhaustion on turnover intentions is positive and significant, which
provides support for H5. In terms of the control variables, the combined sample results show that married
frontline employees report better job performance and education has a negative effect on turnover
intentions. The control variables collectively explain only 1 percent of the variance in work-family and 1
percent in family-work conflict.
Gender as a moderator

Next, we tested the moderating effects of gender on the theoretical linkages in Figure 1 by performing a
two-group simultaneous analysis using the sample covariance matrices as input to LISREL 8.51
(Joreskog and Sorbom, 1993). Initially, with the exception of measurement invariance constraints, no
equality constraint was imposed on the corresponding structural model parameters across female and
male samples. This “base” model generated a χ 2 value of 281.43 (df = 134), and served as benchmark
for testing the moderator hypotheses. Parameters estimates, t-values and model fit statistics for the two-
group base model are presented in Table IIunder the “two-group analysis” columns.

Using the χ 2 value of 281.43 (df = 134) generated by the base model as benchmark, we examined each
moderator hypothesis by removing the equality constraint for a particular path across genders and
conducting a χ 2 difference test with 1 df. A moderator hypothesis would be supported if the
resulting χ 2 suggested a significant model fit deterioration over the base result. For instance, to test H6a,
the path from work-family conflict to emotional exhaustion was constrained to be equal across genders.
The resulting χ 2value of 282.55 (df = 135) indicates that model fit does not deteriorate significantly. Thus,
gender does not moderate the relationship between work-family conflict and emotional exhaustion,
and H6a is not supported. Likewise, to test H6d, we constrained the effect of family-work conflict on job
performance to be equal across genders. The χ 2 value of 287.02 (df = 135) we obtained indicates a
significant difference in the magnitude of this path across genders. Thus, H6d is supported. Gender
moderates the relationship between family-work conflict and job performance and, as hypothesized, the
relationship is stronger among female employees.

Following the preceding approach, we tested each of the remaining moderator hypotheses. We found
that, as in the case of work-family conflict, gender did not moderate the relationships between family-work
conflict and emotional exhaustion. Thus, H6b cannot be supported. The same is true for H6g, which
posited that the relationship between emotional exhaustion and job performance would be stronger
among female employees. Our results show that gender does not moderate the emotional exhaustion –
job performance relationship. Thus, H6g is not tenable. While gender moderates the relationships
between work-family conflict and job performance (H6c), the effect of work-family conflict on job
performance is positive and contrary to the expected sign. A similar situation exists with respect to the
relationship posited between family-work conflict and turnover intentions (H6f). Gender moderates the
relationship between family-work conflict and turnover intentions. However, the relationship is significantly
stronger for males than females. On the bases of these results H6c and H6f are not supported by the
data.

Of the remaining hypotheses, our results support that the relationships between work-family conflict and
turnover intentions (H6e) and between emotional exhaustion and turnover intentions (H6h) are, as
predicted, stronger among female employees. Thus, in addition toH6d, H6e and H6h receive support from
the data. A closer inspection of the results presented in Table II shows that gender also moderates the
effects of some control variables on the two job outcomes included in our model. Education has a
significant positive effect on male employees' job performance, but its effect is not significant for female
employees. Married female frontline employees report significantly better job performance, while marital
status does not have a significant impact on male frontline employees' job performance. Finally, the
number of children has a significant positive impact on male frontline employees' turnover intentions, but
its effect on females is not significant. A summary of the study's overall and moderator hypotheses test
results are presented in Table III.

Finally, while our focus in this study was on gender's moderating role on the relationships shown in Figure
1, we also examined its potential direct effects on the study constructs (work-family conflict, family-work
conflict, emotional exhaustion, job performance, and turnover intentions). As the results of multivariate
analysis of variance (MANOVA) summarized in Table IV indicate, female frontline employees experience
significantly higher levels of work-family and family-work conflicts than their male counterparts at 0.05 or
better level of significance. However, there are no significant differences between male and female
frontline employees with respect to emotional exhaustion, job performance and turnover intentions.
Discussion

By using a sample of frontline hotel employees in Turkey as its setting, this study developed and tested a
model to investigate the effects of work-family conflict and family-work conflict, and emotional exhaustion
on two organizationally valued job outcomes, job performance and turnover intentions. Also examined in
the study was the role of gender as a moderator of the relationships in the model. Several observations
emerge from our findings.

First, the results clearly suggest that our overall model is viable. Indeed, of the eight overall hypotheses
we tested, six received support from the data. Consonant with the results reported in other studies
(Bacharach et al., 1991; Boles et al., 1997; Mauno and Kinnunen, 1999), our study shows that employees
facing conflicts originating from their work (family) and family (work) roles become emotionally exhausted.
Likewise, similar to the results reported by Boyar et al. (2003), our results demonstrate that both work-
family conflict and family-work conflict are significant predictors of frontline employees' turnover
intentions. Our results also mirror image Karatepe's (2006)findings and show that emotional exhaustion is
at the root of frontline employees' turnover intentions.

One unexpected result in our study pertains to the relationship between work-family conflict and job
performance. Contrary to our predictions, work-family conflict seems to trigger frontline employees' job
performance. One plausible explanation for this unexpected finding may be based on van Dyne et al.'s
(2002) argument that people who experience interpersonal conflict and tension at work tend to focus on
their work activities more to shield themselves from further tension and to be able to reach higher levels of
performance. An alternative explanation to this unexpected finding can be offered in light of the
insignificant role of emotional exhaustion in mediating the relationship between work-family conflict and
job performance. Two types of countervailing effects may be operating on work-family conflict – job
performance relationship at the same time. These are a eustress-type (positive/functional) effect and a
distress-type (negative/dysfunctional) effect (Singh et al., 1994). These opposite effects may act
simultaneously to reduce the total effect of work-family conflict on job performance. Conceivably, here the
eustress-type (positive/functional) may be the more dominant component in affecting job performance. It
is possible that the distress-type (negative/dysfunctional) effect of work-family conflict on job performance
does not kick in until emotional exhaustion experienced by frontline employees exceeds a certain
threshold. Whatever the explanation may be, certainly the relationship between work-family conflict and
job performance is more complicated than expected and deserves further attention.

Second, when the results are viewed from the perspective of the simultaneous effects of the two forms of
interrole conflict, it appears that both work-family conflict and family-work conflict have detrimental
impacts on emotional exhaustion and turnover intentions. However, the effects of work-family conflict and
family-work conflict on job performance are different. As noted before, while work-family conflict depicts a
positive relationship with job performance, family-work conflict has a detrimental impact on job
performance. These results corroborate that work-family and family-work conflicts are distinct but
conceptually related concepts (Netemeyer et al., 1996). In many cases, not being mutually exclusive, they
have spillover effects and exert the same type of impact on an outcome. Yet, in other cases, as
demonstrated in a study of interrelationships of work-family conflict and family-work conflict with work
satisfaction (Boles et al., 2001), work-family conflict and family-work conflict do not necessarily yield the
same impact on the outcome measure (i.e. work satisfaction).

An explanation for such a differential impact comes from Netemeyer et al. (2005) who argue that when
employees are cognizant of the potential for a particular form of conflict to affect the outcome (e.g. job
performance), they may engage in processes and behaviors that partially compensate for the effect.
Given this explanation, it may be surmised that employees surveyed here are more cognizant of the
potential impact of work-family conflict on their job performance. But not being equally cognizant of the
potential for family-work conflict to affect their job performance, they may not engage in cognitive
processes or behaviors that could reduce the effect.
Third, unlike the bulk of the studies which examined interrole conflicts and related issues in developed
countries, our study was conducted in Turkey. When our results are compared to the results of studies
conducted in the North American – European axis, a number of similarities are apparent. For example,
our results pertaining to the positive relationship between work-family conflict and emotional exhaustion
are congruent with Boles et al.'s (1997) research in the USA, and Mauno and Kinnunen's (1999) study in
Finland. Likewise, our findings concerning the effect of family-work conflict on emotional exhaustion
parallels the results reported in Posig and Kickul's (2004) US-based study. Similarly, our results
pertaining to the impact of family-work conflict on job performance is consonant with Frone et al.'s
(1997) study in the USA. Also our findings regarding the effects of work-family conflict and family-work
conflict on turnover intentions are consistent with those of Boyar et al. (2003) who conducted their study
in the USA. These results, on one hand, suggest that research findings derived from western countries
are generalizable into a different cultural setting and, on the other hand, lend credence to Aycan and
Eskin's (2005) observation (based on Barnett and Hyde's (2001) expansionist theory) that as traditional
gender roles continue to expand and change globally, a convergence of findings in work-family research
takes place cross-culturally.

Fourth, on the basis of our results, the answer to the fundamental question we asked, “Does gender
matter?” appears to be yes. Indeed, of the eight gender-related linkages we proposed in our study, five
proved to be significant. Thus, at the aggregate level, gender does moderate a majority of the
relationships in our model. However, in two cases, the differences between male and female employees
are contrary to our predictions. Hence, three of our hypotheses receive support from our data while the
other two are not tenable.

We offer a speculative explanation as to why the positive relationships between work-family conflict and
job performance and between family-work conflict and turnover intentions are weaker among female
employees. These may be rooted in the fast pace of change in the work and family roles of men and
especially women in Turkey. As Yavas et al. (1999) write, while the Turkish society is male-dominated
and males are the primary breadwinners, changes enacted in the Turkish Civil Law in the 1990s have
accelerated the transition in the status of Turkish women from that of traditional housewife to equal
partner. The changes on the legislative front have picked up momentum in the last couple of years as
Turkey enters formal negotiations with the European Union to be a full-fledged member and makes its
laws compatible with those of member European countries.

Perhaps, even more important is the fact that these days Turkish women are economically active. Many
are no longer “stay-at-home” moms waiting for their husbands to return to the nest. As Aycan and Eskin
(2005) discuss, Turkish women especially in urban metropolitan areas (such as Ankara, the location of
our study) join the workforce in increasing numbers. With the increasing involvement of women in the
workforce, cultural values and norms with respect to gender roles are undergoing a rapid change in
Turkey. Recently, women are more involved in work outside the home and men, slowly but surely,
assume more of the household responsibilities and tasks that were traditionally considered as women's.
In such a milieu, Turkish men and women alike are trying to adapt to the modern gender role norms
(Aycan and Eskin, 2005) and a “blurring” of traditional gender role distinctions is taking place.

It appears that this trend toward the “blurring” of traditional gender roles and the Turkish women's desire
to succeed in the full-time workforce, despite experiencing significantly higher conflict between work and
family roles, may have caused the unexpected findings. It is also likely that, as discussed in the gender
role theory, despite being more prone to the dysfunctional effects of work-family conflict than men
(Grandey et al., 2005), women expect such an interference between work and family domains more so
than men. It should also be remembered that frontline service jobs, as interaction-oriented positions, fit
better the relationship-valuing and nurturing nature of the female gender (Putrevu, 2001). Furthermore, as
a testament to the context-specific changes taking place in role expectations for men and women, women
nowadays value certain masculine-stereotyped job attributes as highly or more highly than men
(Konrad et al., 2000). In addition, in today's uncertain work environment, women may be more likely than
men to seek job security (Konrad et al., 2000). Consequently, compared with men, women may be in a
better position to cope with and control the dysfunctional effects of conflicts emanating from work and
family domains on their job performance and turnover intentions.

Implications

Based on our overall results, it appears that Turkish hotels will benefit from establishing a family-
supportive work environment for their employees to lessen (and hopefully eliminate) the negative impact
of conflicts emanating from the work-family interface on their emotional exhaustion and job outcomes. In
this context, on-site childcare services for employees with younger children, after-school programs for
those with older children, for instance, can pay dividends.

However, it should be pointed out that given the gender-specific findings of this study, “one-size fits all”
approach may be problematic. For example, our results demonstrate that the impact of work-family
conflict on turnover intentions is stronger among female employees, while the effect of family-work conflict
on turnover intentions is higher for male employees. Thus, management can organize gender-specific
small-group meetings and sessions to identify, for instance, the main sources of work-family conflict
among female and family-work conflict among male employees. In a same-gender environment,
employees can freely voice their concerns and openly discuss the drivers of their interrole conflicts.

Once the root causes of conflicts are identified, then management can take the necessary actions. For
instance, gender-specific programs may be offered to educate employees on changing cultural norms,
gender roles, job sharing responsibilities at home, etc. Given that gender consciousness at home is the
final frontier in the quest for gender equality in work-family relationships, programs/presentations to
promote and reinforce this idea in Turkey that is in a period of transformation is particularly important
(Aycan and Eskin, 2005). Likewise, employees can be educated on the crucial role of spousal and family
support in alleviating interrole conflicts and coping with emotional exhaustion. Management can
complement such actions by also training the employees in supervisory roles and educating them in
actions that would be helpful in dealing effectively with the interrole conflicts that female and male
employees experience.

Limitations and future research directions

Although this study expands our knowledge base, viable prospects for further research remain. First, the
cross-sectional design of our study does not permit us to make causal inferences. Future studies
employing longitudinal designs would be helpful in establishing causal relationships. Concurrently, some
unexpected findings in our study underscore the need for qualitative studies to gain deeper and richer
insights into how women and men experience and cope with conflicts in the work-family interface.
Second, in this study, data from single-informants (self-report data from employees) were used to
measure all the variables. Such data are prone to common-method variance (Doty and Glick, 1998). To
minimize common method-variance, future studies should use multiple-informants, and for instance,
should measure frontline employees' job performance on the basis of their supervisors' assessment.
Third, to cross-validate our results and broaden the database for further generalizations, replication
studies among other samples of frontline employees in Turkey are needed. Fourth, inclusion of other
easily obtainable variables such as the work status of each spouse in a married couple (dual-earner/dual-
career vs single-earner/single-career) and availability of various types of support (e.g. extended family
support, spousal support) as control factors might lead to finer insights as such characteristics may affect
various relationships posited in our model (Aycan and Eskin, 2005; Elloy and Smith, 2003).

In conclusion, interrole conflicts experienced by frontline employees have significant consequences. At a


time of a changing work environment where gender roles all over the world continuously evolve, the
management of frontline employees to attain positive job outcomes will be a challenge. Hence, the issues
we addressed in our study should remain as a research priority.
Figure 1Research model

Table ICorrelations, means and standard deviations of latent variable indicators and control variables
Table IITests of the research model and hypotheses
Table IIISummary of hypothesis test results

You might also like