Barry J. Babin & James S.

Boles

Employee Behavior in a Service

Environment: A Model and Test
of Potential Differences Between
Men and Women
The authors examine the attitudes and behaviors of employees who provide frontline service and address the
extent to which relationships vary among male and female employees. The overall model predicts effects of role
stress and work/nonwork conflict on customer-contact employees' job performance, job and life satisfaction, and
quitting intent. Results of structural equations modeling suggest an important role for work/nonwork conflict overall
as well as two areas of interesting variation across gender. Specifically, multisample structural equations analyses
suggest that role stress affects female service providers' job performance more negatively than it does males', and
that job satisfaction is related more highly to quitting intent among males. Overall, results suggest interesting sim-
ilarities and differences across gender.

R
esearchers increasingly are recognizing important and women, service organizations have an interest in
and unique demands and far-reaching consequences research that suggests how male and female employees
associated with the customer-management interface. might respond differently to workplace events. The potential
A significant body of literature addressing affective and for differing responses to like policies and supervisor
behavioral outcomes that result from boundary-spanning actions occurs when a person's job role does not override his
employees' stress has emerged (Brown and Peterson 1993, or her sex role. Previous studies investigating gender effects
1994; Lusch and Serpkenci 1990; Singh 1993). Neverthe- generally describe gender-related differences in levels of
less, stress-related marketing research focuses primarily on important organizational constructs, particularly role stress
industrial sales and purchasing positions as representatives andjob satisfaction (Busch and Bush 1978; Lefkowitz 1994;
of boundary-spanning occupations. Far more common in Schul and Wren 1992). However, organizational studies
number are customer-contact service providers, who occupy more often are conducted in male-dominant occupations
a critical role in heavily trafficked servicescapes including with little consideration of potential gender-based differ-
retail stores, hotels, tourist venues, restaurants, and other ences (Thomas and Ganster 1995). Therefore, managerial
service environments (Bitner 1992). Male and female front- thinking based on descriptive organizational research suf-
line service providers are charged with implementing strate- fers from a strong agentic bias.
gic marketing decisions at the exchange point. By carrying The research presented here expands current knowledge
out actions at the "critical moments of truth" at which cus- of employee behavior by addressing two relatively underre-
tomers and employees interact, these service providers searched issues. First, an overall theoretical model is tested
become the most salient and conspicuous indicators of a using a sample of nonmanagerial, frontline service
marketing organization's quality (Bitner, Booms, and Mohr providers. Considering the crucial role these employees play
1994, p. 95). in linking a firm with its customers and thus in building rela-
The boundary-spanning literature generally has over- tionships, there is great interest in understanding factors that
looked potentially important moderating factors such as affect their performance, satisfaction, and quitting inten-
gender (Singh, Verbeke, and Rhoads 1996). Because typical tions. Second, rather than examining differences in levels of
frontline service-providing jobs are occupied by both men key constructs across gender, our primary focus is on differ-
ences in relationships across gender among selected key
Barry J. Babin isAssociate Professor ofMarketing, Marketing Department, constructs. Even if men and women report similar levels of
University ofSouthern Mississippi. James S. Boles is Associate Professor a specific construct, there still might be differences in how a
of Marketing, Marketing Department, Georgia State University. The construct affects endogenous factors. This study specifically
authors contributed equally in the development of this research. This focuses on this potential moderator in a theoretical model
research was supported partially by the Office of Research and Spon- that examines role stress consequences. The model opera-
sored Programs at the University ofSouthern Mississippi and the Market- tionalizes role stress as two related constructs, role conflict
ing Round Table at Georgia State University. Theauthors acknowledge the
and role ambiguity, and predicts various direct and indirect
helpful comments of Danny Bellenger and Bill Darden inconstructing this
manuscript. They also acknowledge Larry Ross'shelp indata collection. effects on worklnonwork conflict, job performance, job sat-
isfaction, life satisfaction, and quitting intent.

Journal of Marketing
Vol. 62 (April 1998), 77-91 Employee Behavior in a Service Environment /77
 Service Provider Model Overview Previous research indicates that stress results in many
detrimental effects. Role ambiguity affects both perfor-
Figure 1 displays the proposed path model. We include these
mance and job satisfaction directly and negatively and
constructs because they are applied widely in organizational
increases work/nonwork conflict and quitting intent
research and maximize relevance in this particular context.
(Behrman and Perreault 1984; Good, Sisler, and Gentry
Given the extensive literature theoretically linking con-
1988). Role ambiguity is characterized by uncertainty as to
structs examined in this study (Behrman and Perreault 1984;
expected behavior in common job situations, and it reduces
Good, Sisler, and Gentry 1988; Hartline and Ferrell 1996), a
performance through diminished effort and delays in taking
detailed review is not presented. We highlight key rationale
action (Brown and Peterson 1994). Stress perpetuates nega-
and findings subsequently.
tive affect, lowers positive job appraisals, and results in
In a typical servicescape, service providers believe that
lower job satisfaction.
rules, regulations, and policies enforced by management and
often motivated by cost control are the primary reasons for Role stress also exerts influence through key facilitating
failing to satisfy customers completely (Bitner, Booms, and constructs. Both work/nonwork conflict (negatively) and
Tetreault 1990). However, the customer-contact employee, performance (positively) are related to job satisfaction
not the decision maker, is exposed to the customer's wrath. (Bagozzi 1978). The negative feelings associated with the
Service providers cope by making system "adjustments" to work/nonwork conflict eventually spill over and reduce job
avoid a negative service encounter (Bitner, Booms, and satisfaction. Previous empirical research suggests a modest
Mohr 1994; Bitner, Booms, and Tetreault 1990). Appropri- positive, direct relationship between performance and satis-
ate employee adjustments are consistent with empirical faction (Brown and Peterson 1994).
results that suggest a positive role conflict-performance Job satisfaction and life satisfaction are hypothesized as
relationship in customer-contact settings (Behrman and Per- affecting quitting intent directly (Good, Sisler, and Gentry
reault 1984; Dubinsky and Hartley 1986; Hartline and Fer- 1988; Netemeyer, Johnston, and Burton 1990). Low job sat-
rell 1996). isfaction perpetuates negative affective appraisals, which

FIGURE 1
Hypothesized Model of Customer-Contact Service Provider Behavior

Role Stress

Note: Hypothesized direction of relationships indicated by + or -. Dashed lines indicate relationships with hypothesizsed gender differences.

78/ Journal of Marketing, April 1998

represent threats to a worker's overall well-being and type of anxiety and its related consequences (Thomas and
increase his or her avoidance (quitting) motivations (Brown Ganster 1995).
and Leigh 1996). Life satisfaction is perceived as a positive Differences in self-reported measures of attitudes such
outcome of job satisfaction and a negative outcome of as job satisfaction, performance, and life satisfaction also
worklnonwork conflict (Adams, King, and King 1996; have been examined (Busch and Bush 1978; Schul, Rem-
Netemeyer, Boles, and McMurrian 1996). A desire to have ington, and Berl 1990). Female industrial salespeople report
both constructive work and nonwork lives generates trade- lower self-rated job performance than males (Schul and
offs that can be detrimental to overall life satisfaction. For Wren 1992). Women also report greater satisfaction with
example, extra time at work may mean that an employee pay and promotional opportunities and, in general, perva-
cannot maintain an effective exercise program, does not per- sively higher overall job satisfaction (Hodson 1989).
form well on an exam, or misses a family or social engage-
ment. Therefore, whereas job satisfaction's impact on Work and Sex Roles
quitting intent is direct, worklnonwork conflict's effect is Male and female employees can become socialized and
mediated by its impact on the employee's life situation, as behave similarly in their roles as police officers, psycholo-
reflected by life satisfaction. Overall, an important goal of gists, professors, chefs, and so on, rather than as men or
this research is to examine the validity of the proposed women (Chao et al. 1994). The work role and its prescrip-
model. Thus, the following model suggests a test of the tive attachments often override behavioral or attitudinal dif-
overall model: ferences attributable to gender. Despite this socialization,
there are instances in which sex role differences override
Hom: The path relationships shown in Figure I can be used to
reproduce the correlations among the constructs depicted. work roles, which suggests different reactions and behavior
between men and women. The strongest, most identifiable
and resilient sex role difference can be described in terms of
The Impact of Sex Roles on an agentic-cornmunal dimension (Eagly 1987; Iacobucci
and Ostrom 1993). Differences arise on the basis of male
Marketing Employees tendencies toward high assertiveness, task mastery, and
Previous Organizational Research on Gender individualism ("I can find it myself'), and female tendencies
Differences toward concern for people, devotion, and acquiescence
Investigating differences between men and women has not ("Let's ask for directions"). Personality research is generally
been an overly popular research area for many reasons, consistent with men showing relatively high levels of
including issue sensitivity (Eagly 1987). The lack of popu- exploratory (mastery) tendencies and women showing
larity contributes to the small number of organizational stud- greater passiveness (Pulkkinen 1996).
ies that examine how gender might alter relationships among Coworkers also have certain expectations about others
constructs. Most existing research compares amounts of con- that are influenced significantly by gender-based stereo-
structs expressed by female and male employees (Busch and types (Deaux 1985).1 Social cognition research suggests that
Bush 1978; Schul and Wren 1992; Siguaw and Honeycutt the most commonly associated characteristics of the typical
1995). For example, research indicates that female employ- female or "woman" stereotype are "emotional, weak, depen-
ees exhibit relatively high psychological and mental distress dent, passive, uncompetitive, and unconfident" (Fiske and
(Nelson and Quick 1985), whereas male employees report Stevens 1993, p. 179). Clearly, stereotypes contain miscon-
relatively high physical distress (lick and Mitz 1985). Other ceptions, but they do serve as prescriptive devices regarding
research suggests that women report greater emotional behavior and sometimes contain "kernels of truth" (Hoff-
exhaustion than do men, but male employees display higher man and Hurst 1990, p. 197). Role expectations can be
levels of coworker depersonalization (Himle, Jayaratne, and biased by stereotypes, and research suggests that con-
Thyness 1989). Prior studies also suggest that women expe- sumers' performance expectations vary correspondingly
rience lower levels of work-related role clarity than do men (Iacobucci and Ostrom 1993).
(e.g., Busch and Bush 1978). Despite progress, there is still a workplace power dis-
Women, consistent with female role expectations, report crepancy with correspondingly fewer numbers of women in
taking on relatively high levels of family responsibilities managerial positions. Female employees often face the
compared with men (Reifrnan, Biernat, and Lang 1991), dilemma of behaving consistently with the gender stereo-
which contributes to greater interference between work and type or more aggressively (an aggressive act is inconsistent
nonwork responsibilities (Greenhaus, Bedeian, and Mossh- with the female schema and therefore stands out and pro-
older 1987). Previous research also indicates gender-based vides a cue for contrast) and risk negative evaluations
differences with respect to consequences of worklnonwork because they are not behaving as expected of a woman. As
conflict (Duxbury and Higgins 1991; Gutek, Searle, and Fiske and Stevens (1993, p. 181) point out, "discrimination
Klepa 1991). Worklnonwork conflict occurs when a per- would result from not behaving like a woman should and, in
son's work role interferes with his or her family role the other case, from behaving too much like a woman." In
demands, personal responsibilities, and/or social life and is
accentuated by time pressure and job-related role stress I Technically, gender-based differences sometimes are distin-
(Boles and Babin 1996; Greenhaus and Beutell 1985). guished from sex-based differences in that they are more specific
Research demonstrates that employees with even the sim- and due to sociological as opposed to biological processes. Here,
plest family arrangements (i.e., young and single) suffer this the term "gender-based difference" is adopted.

Employee Behavior ina Service Environment /79

summary, male and female service providers might react Role Conflict and WorkINonwork Conflict
differently to phenomena present in the service encounter
Women increasingly have entered the workforce at all job
environment. This is particularly so for situations evoking
status levels. However, to the extent that women have made
agentic or communal qualities. Specific gender-based dif-
inroads in job markets, they still appear to bear dispropor-
ference hypotheses follow.
tionately the burden of family and homemaking responsibil-
ities (Gutek, Searle, and Klepa 1991). Since 1970, working
Role Stress and Performance
men report increasing the amount of time spent each week
There is ample evidence that suggests gender-based variation on family and household responsibilities from approxi-
in a person's reaction to stress and uncertainty (Cournoyer mately 81 to 83 minutes (Crosby and Jaskar 1993). So,
and Mahalik 1995). As discussed previously, strong gender- while women allocate more time in the paid workforce, men
based differences exist such that men are and are expected to have compensated by spending an extra two minutes a week
be relatively more aggressive and autonomous than women on family responsibilities. The female sex role and typal
(Eagly 1987; Hoffman and Hurst 1990). In the workplace, expectations are consistent with results that show greater
men, even in like positions, behave more authoritatively and communal and expressive properties, which include nurtur-
are more dominating, whereas women display more submis- ing of loved ones and a strong motivation for family main-
siveness and compliance (Berger, Rosenholtz, and Zelditch tenance (Deaux 1985). Although this discrepancy has
1980). This submissiveness and compliance creates a ten- obvious and familiar effects on married couples with chil-
dency for women to behave more consistently with voiced dren, evidence suggests it exists even in very simple family
organizational policies and rules (Eagly 1987), which makes arrangements (Hodson 1989).
it more difficult for them to make the needed adjustments to Research documents the interplay between work and
establish a positive stress-performance relationship (Bitner, nonwork stress. Persistent expectations for women to per-
Booms, and Mohr 1994). form family and home responsibilities may not leave room
When a confrontation occurs, a man is likely to experi- for the stress to dissipate (Gutek, Searle, and Klepa 1991).
ence feelings evoking approach rather than avoidance In contrast, the male sex role, which places less emphasis on
responses. Traditional female stereotypic (consistent with family obligations, allows a greater separation of conflict on
the sex role) behavior would lead to a greater avoidance of and off the job (Biernat and Wortman 1991). Furthermore,
conflict than would the stereotypic male behavior. In a to the extent that increased stress is associated with
boundary-spanning situation, positive conflict-performance increased hours at work, a significantly greater correlation
relationships are expected when a service provider uses spe- between hours at work and work/nonwork conflict is
cific coping strategies involving actions aimed at resolving, observed among women (Gutek, Searle, and Klepa 1991).
not avoiding, conflict (Brown and Peterson 1994). Many
studies showing a positive role conflict-job performance H2: The relationship between role stress and work/non work
relationship have involved boundary-spanning samples, conflict is moderated by gender, such that the overall pos-
itive role conflict-work/nonwork conflict relationship is
such as salespeople, that were traditionally populated dis- greater among women than among men and the positive
proportionately by men (cf. Behrman and Perreault 1984). role ambiguity-work/nonwork conflict relationship is
Furthermore, the masculine tendency toward mastery and greater among women than among men.
task proficiency promotes greater initiative when men are
confronted with uncertainty (Eagly, Makhijani, and Job Satisfaction and Life Satisfaction
Klonosky 1992).
Potential gender differences in the job satisfaction-life sat-
Evidence from the helping literature also supports a
isfaction relationship have probably received more atten-
more positive (negative) role stress performance relationship
tion than any other organizational relationship. A recent
among men (women). In many situations women tend to
meta-analysis suggests systematic variation in previous
help as much or more than men. However, meta-analysis of
findings (Tait, Padgett, and Baldwin 1989). Studies pub-
research on helping shows that the strongest moderator of
lished before (after) 1974 show greater (similar) job satis-
the helping relationship is risk. When risk is present, men
faction-life satisfaction relationships among men than
tend to help to a greater extent than women (Eagly 1987). In
among women. Traditionally, a person's nondomestic occu-
the service-providing environment, in which both customers
pation is a more (less) important component of the male
and managers are present, potential deviations from expecta-
(female) self-concept and sex role (Biernat and Wortman
tions are seen as risky and create stress (Ramaswami 1996).
1991; Hoffman and Hurst 1990; Josephs, Markus, and
Therefore, to the extent that the customer is seen as needing
Tafarodi 1992). Men find more self-gratification in their
help, and service provider performance is ultimately mea-
work; therefore, a stronger relationship might be expected
sured by relationships established with customers, men may
between job and life satisfaction. In the interest of compar-
cope with stress more positively than women.
ing results with those of workers studied previously and
HI: The relationship between role stress and job performance examining the relative strengths of sex roles versus work
is moderated by gender, such that the overall positive role roles in forming this relationship, the following hypothesis
conflict-performance relationship is greater (more posi- is offered:
tive) among men than among women and the negative role
ambiguity-performance relationship is greater (smaller in H3: The positive relationship between job satisfaction and life
absolute value, less negative) among men than among satisfaction is moderated by gender, such that it is stronger
women. among men than among women.

80I Journal of Marketing, April 1998

Job Satisfaction and Quitting Intent scales have been used extensivly in marketing and organi-
Gender differences in the job satisfaction-quitting intent zational research (e.g., Brown and Peterson 1994; Michaels
relationship also might be expected. A potential rationale and Dixon 1994). An exhaustive measurement analysis sup-
again involves the aggressive (noncompliant) nature of the ports the scales' validity in representing job-related role
male sex role (Eagly 1987) and women's tendancy to be stress as two related constructs (Netemeyer, Johnston, and
more resigned to their fate than are males, both in general Burton 1990).
(Pulkkinen 1996) and on the job (Reifman, Biernat, and The work/nonwork conflict measure consists of five
Lang 1991). A male employee is more apt to take an questions from Burke, Weir, and DuWors' (1979) study as
extreme action such as quitting than a female in the same adapted by Parasuraman and colleagues (1989). This mea-
situation. sure addresses a variety of issues related to areas in which
Statistical evidence also shows a differential in the ben- work- and nonwork-related issues can conflict, including
efits of time unemployed. On average, men benefit finan- relationships with friends, mental and physical states at
cially from a period of unemployment as evidenced by a home, and participation in home activities.
wage-gain rate nearly twice as high as that for women (Ben- Ten five-point Likert items assessed overall job satisfac-
ham 1993). In other words, men usually take a job earning tion of respondents (Brayfield and Rothe 1951). Similar to
more money than they did on their previous job. This effect other recent attempts (Singh, Verbeke, and Rhoads 1996),
is not observed among women experiencing the same level the items reflect overall satisfaction and not any specific
of unemployment. Heightening this effect is the high dimension of satisfaction. Overall life satisfaction was mea-
turnover rate common to service industries, which gives sured by Quinn and Shepard's (1974) Quality of Life Scale.
employees considerable experience in the effects of unem- Performance was assessed through a seven-item self-report
ployment. Therefore, a woman might believe that it will be measure similar to that reported by Singh, Verbeke, and
more difficult to find a comparable or better job if she leaves Rhoads (1996). These questions focus on a respondent's
her present job. view of his or her performance relative to coworkers. Five-
In addition, despite increases in gender equity, the norm point Likert scales were used as indicators. This measure
is still that a married women is given second priority in fam- was adapted from similar self-reported measures, used in
ily job decisions. A married woman remains more likely to previous marketing studies, to reflect the specific require-
follow her husband to a new location than vice versa. There- ments of a waitstaff position.
fore, she may leave a job despite being satisfied. All of this Quitting intent was assessed using items developed by
evidence points to a stronger relationship between job satis- Bluedorn (1982). Respondents were asked how likely they
faction and quitting intent among men than among women. would be to leave their job within given time frames.
Responses were reported on a one to seven scale anchored
H4 : The negative relationship between job satisfaction and by "I = terrible chance" and "7 = excellent chance." All
quitting intent is moderated by gender, such that it is measures were scored so that higher numbers reflect corre-
stronger (more negative) among men than among women. spondingly greater amounts of the construct. We provide
construct descriptions and summaries in Table I.
Research Methodology
Sample Results
A sample of employees who provide retail service offered Measurement Results
data for analysis. Specifically, food servers at relatively Overall measurement quality was assessed using confirma-
upscale, full-service restaurants in a large metropolitan area tory factor analysis (Anderson and Gerbing 1992). Although
were interviewed with drop-off questionnaires. The restau- measurement quality is sometimes assessed factor by factor,
rants included a wide range of locations around the city and each multiple-item indicator was considered simultaneously
a wide variety of menu types. Completed questionnaires to provide for the fullest test of convergent and discriminant
were returned directly to us by a postage-paid envelope. validity. Initial analyses suggested five items with low fac-
Approximately 500 questionnaires were distributed. tor loadings (below .50) that were dropped from further
Questionnaires were returned by 331 respondents. Of those, analyses (see Table 2).
328 had complete information, which represent a 65.6% All loadings exceed .5, and each indicator t-value
usable response rate. The cooperation of management and exceeds 10.0 (p < .001). Coefficient a exceeds .8 for each
distribution during employee meetings helped the response scale. The overall fit supports the measurement model. The
rate. Demographically, 43.3% of the respondents were men, X2 fit statistic is 1232.6 with 758 degrees of freedom (p <
and the typical respondent was 26 years of age with approx- .001). The root mean squared error (RMSE) is .05, the com-
imately six years' industry work experience. The sample is parative fit index (CFl) is .93, the adjusted goodness-of-fit
similar in profile to that of a previous study examining hotel index (AGFI) is .83, the parsimony normed fit index is .7 I,
service employees (Hartline and Ferrell 1996). and the X2/dfratio is 1.63. All support the overall measure-
ment quality given a large sample and number of indicators
Measures (Gerbing and Anderson 1992). Furthermore, the variance
Role stress was measured using Rizzo, House, and Lirtz- extracted in each measure exceeds the respective correlation
man's (1970) role conflict and role ambiguity scales. These estimate between factors, which provides evidence of dis-

Employee Behavior in a Service Environment /81

 TABLE 1
Summary of Overall Model Constructs

Construct Description Key Citations

Quitting Intent The likelihood that a person will leave his or her job within the Bluedorn (1982); Good,
foreseeable future. Quitting intent is a function of job Sisler, and Gentry (1988);
satisfaction and an employee's overall life satisfaction, among Johnston et al. (1990)
other things.

Life Satisfaction The degree to which people judge the quality of their lives Adams, King, and King
favorably. It can be equated with a general degree of (1996); Netemeyer, Boles,
happiness. Life satisfaction is generally viewed as a function of and McMurrian (1996)
job satisfaction and other personal considerations.

Job Satisfaction A positive emotional state that results from employees' Bagozzi (1978); Brown and
appraisal of their job situation. High performance is appraised Peterson (1994); Good,
positively and results in higher satisfaction. Stress-related Sisler, and Gentry (1988);
factors generally lower job satisfaction. Singh, Verbeke, and Rhoads
(1996)

Work/Nonwork The degree to which role responsibilities from the work and Boles and Babin (1996);
Conflict nonwork domains are incompatible. That is, fulfilling Burke, Weir, and DuWors
responsibilities in the nonwork role is made more difficult by (1979); Parasuraman et al.
participation at work. For example, overtime at work may (1989)
interfere with a part-time employee's ability to prepare for an
important marketing exam. Work-related stress spills over and
causes work/nonwork conflict.

Job Performance The level of productivity of an individual employee, relative to Busch and Bush (1978);
his or her peers, on several job-related behaviors and Kohli (1985); Singh,
outcomes. Job performance is affected by work-related Verbeke, and Rhoads (1996)
variables including role stress and work/nonwork conflict.
Relatively high performance generally leads to higher job
satisfaction.

Role Conflict The degree to which work expectations and work requirements Behrman and Perreault
of two or more persons are incompatible. For example, a (1984); Brown and Peterson
restaurant might have a policy of no splitting of entrees. A (1994); Good, Sisler, and
waitstaff member might face a customer who requests that he Gentry (1988); Michaels and
and a family member be allowed to share an entree. ThUS, Dixon (1994); Rizzo, House,
management's expectations conflicts with the customer's. and Lirtzman (1970)

Role Ambiguity The degree of uncertainty about one's job including uncertainty Behrman and Perreault
regarding management's expectations. For example, improper (1984); Brown and Peterson
training may result in employees not knowing the expected (1994); Good, Sisler, and
response to frequently occurring job events. Gentry (1988); Michaels and
Dixon (1994); Rizzo, House,
and Lirtzman (1970)

criminant validity (Fornell and Larcker 1981). Therefore, Factor structure invariance was tested by comparing
the measures are adequate for further analysis (see the results of a confirmatory model fitting separate models for
Appendix for items). men and women. Initially, model coefficients were freed,
such that separate loading estimates were computed for each
Factor Structure Invariance subsample. Next, the model was retested adding the con-
The research objectives require testing theoretical models straint that A, the matrix of factor loadings, remains invari-
on the overall sample and on split samples examining rela- ant across samples. A comparison of the two fits enables an
tionships among only male and female respondents, respec- assessment of factor structure invariance. The overall X2 for
tively. Therefore, construct validity must be exhibited in the the two sample model is 2213 with 1516 degrees of freedom
measurement model, tested over the entire sample and in (p < .001; CFI = .90) for the "totally free" model and 2247.5
each subsample. Therefore, an examination of the factor with 1557 degrees offreedom for the constrained model. The
structure across gender was conducted (Bandolos and Ben- X2 difference statistic between these two models is 34.5 with
son 1990; Byrne 1988). 41 degrees of freedom and is nonsignificant (p > .1), which

82/ Journal of Marketing, April 1998

 TABLE 2
Standardized Measurement Coefficients and T-Values Resulting from Confirmatory Factor Analysis a

Construct
Roll Stress

Item Role Role WorkJ Job Life Quitting

Abbreviation Conflict Ambiguity Performance Nonwork Satisfaction Satisfaction Intent
RC1 .59 (10.4)b
RC2 .60 (10.4)
RC3 .64 (11.5)
RC4 .65 (11.9)
RC5 .75 (14.4)
RC6 .64 (11.4)
RA1 .68 (12.8)
RA2 .69 (13.3)
RA3 .75 (14.6)
RM .80 (16.0)
RA5 .70 (13.1)
JP1 .69 (13.3)
JP2 .76 (15.2)
JP3 .79 (16.1)
JP4 .63 (11.9)
JP5 .71 (13.8)
JP6 .60 (10.7)
JP7 .74 (14.6)
WNW1 .78 (15.4)
WNW2 .65 (12.0)
WNW3 .60 (11.1)
WNW4 .75 (14.6)
WNW5 .59 (10.1)
JS1 .84 (18.4)
JS2 .69 (13.6)
JS3 .73 (14.8)
JS4 .72 (14.7)
JS5 .80 (16.9)
JS6 .73 (14.9)
JS7 .78 (16.3)
JS8 .80 (16.8)
JS9 .68 (13.5)
LS1 .73 (14.6)
LS2 .77 (16.1)
LS3 .82 (17.6)
LS4 .78 (16.3)
LS5 .67 (13.3)
LS6 .88 (19.6)
LS7 .84 (13.5)
011 .90 (14.4)
012 .74 (12.2)
Variance
Extracted .42 .53 .50 .46 .57 .62 .68
a .80 .84 .89 .80 .92 .92 .80
ax2 = 1232.6, 758 degrees of freedom (p < .001), CFI = .93, parsimony normed fit index (PNFI) = .71, RMSE = .05.
bT-values shown in parentheses. All are significant (p < .001).

provides evidence that the measurement model depicted in of freedom (p < .01). A significant difference provides pre-
Table 2 holds across both samples. As is expected from this liminary evidence that gender-based differences among con-
result, all factor loadings are highly significant, and reliabil- struct relationships might exist.
ity estimates remain above .8 in each gender group.
In addition, the model was examined with the added Differences in Means
constraint that <1>, the matrix of interfactor correlations, Table 3 displays construct means by gender. Although no
remains invariant across subsamples. The X2 difference hypotheses were proposed as to mean-level differences, we
resulting from a comparison of this model, with the model present them for comparative purposes. Results are based on
constraining only factor loadings, was 42.1 with 21 degrees two-tailed t-tests. In general, few differences are found. T-

Employee Behavior in a Service Environment /83

tests for equality of means across samples indicates a sig- HI' HI predicts that role stress has a more negative (pos-
nificant difference in work/nonwork conflict (tdf= 303 =2.37; itive) impact on female (male) service providers' job perfor-
p < .05) and in role conflict (tdf= 317 = 3.14; P < .01), such mance. This effect should be evidenced in both the role
that men report relatively high levels of each compared with conflict-job performance and the role ambiguity-job per-
women. In addition, women report marginally higher job formance relationships. For role conflict, this prediction
satisfaction than men (tdf= 306 = 1.91; p < .10). No other sig- suggests a stronger positive coefficient for the male sample.
nificant differences are indicated. Path estimates displayed in Table 3 are consistent with this
prediction. Among men, the role conflict-job performance
SuucturalModelResu/~ path is positive and significant (YI,I =.38; p < .01), and the
Because of the large number of indicators included in this corresponding path estimate among women is negative and
model and the supportive measurement results, summated insignificant (YI,1 = -.06).
indicators of each construct were used in the structural analy- For role ambiguity, HI predicts a weaker negative rela-
ses reported subsequently. Technically, this procedure tionship between role ambiguity and job performance
involves constraining measurement coefficients to the square among male service providers. Results are consistent with
root of a scale's reliability and the corresponding error coef- this prediction as evidenced by a negative path estimate
ficients to one minus scale reliability (Kenney 1979). This smaller in magnitude among the male sample (Y1,2 = -.19;
allows for a more parsimonious presentation of results. p < .05) than among the female sample (Y1,2 = -046; p < .01).
A further examination of H I was conducted using nested
Hom (overall model results). The hypothesized model was
tested across the combined sample (both men and women; n structural equations models. The multi sample model
= 321). The resulting X2 is 11.2 with 7 degrees of freedom (p
described previously was refit, with an additional constraint
> .10; CFI = .99; AGFI = .96; RMSE = .026), which suggests that the paths corresponding to the stress-performance (YI,I
that the hypothesized model fits the data. In Table 4, we pre- and Y1,2) relationship remain equal across the male and
sent the resulting standardized parameter estimates. female customer-contact employee samples. The resulting
In general, estimates are consistent with expectations, X2 is 30.8 with 16 degrees of freedom (p = .014) and the
because all direct paths are significant (p < .05) and in the constrained paths are .19 (p < .05) and -.39 (p < .001) for
expected direction, with the exception of the role the role conflict and role ambiguity to performance paths,
ambiguity-job satisfaction (Y3,3 = -.03, n.s.) and the life respectively. The difference in fits between this and the
satisfaction-quitting intentions scale (~5,4 = -.10, n.s.). The totally free model (X2 = 16.3 with 2 degrees of freedom; p <
strongest direct relationships suggested are from role ambi- .001) supports gender as a moderator of role stress's effect
guity to job performance (YI,2 = -AI, P < .001), work/non- on job performance. Overall, results support HI'
work conflict to job satisfaction (~3,2 = -.55, P < .001), and H 2. H2 suggests that role stress affects female service
job satisfaction to quitting intent (~5,3 = -044,P < .001). The providers' work/nonwork conflict more severely than it
results are supportive of Hom (see Figure 2). does male service providers'. This translates specifically
Split sample analyses. Multisample analyses were con- into a positive coefficient in the female sample stronger in
ducted to examine potential differences in relationships magnitude for both the role conflict and role ambiguity rela-
across men and women. A model was tested that enables all tionships. Unlike the strong support shown for H" the
hypothesized parameters to be estimated freely for both men results here are equivocal. As H2 predicts, the role con-
and women. Separate parameter estimates resulting from flict-work/nonwork conflict estimate is slightly stronger
this analysis are shown in separate columns of Table 4. The among female service providers (Y2,1 = .23) compared with
resulting X2 is 14.5 with 14 degrees of freedom (p = .39), male service providers (Y2,1 = .14). However, the difference
which suggests adequate fit. in the role ambiguity-work/nonwork conflict relationship

TABLE 3
Correlation Estimates (C'b) and Construct Means

Role Stress Mean Values

JP WNW JS LS 01 RC RA Men Women

Job Performance 1.00 28.5 (4.6)a 27.3 (4.6)

Work/Nonwork Conflict -.09 1.00 15.1 (3.3) 14.3 (3.3)
Job Satisfaction .14 -.65 1.00 31.6 (7.7) 32.3 (7.9)
Life Satisfaction .19 -.52 .45 1.00 31.2 (7.9) 32.1 (7.7)
Quitting Intent -.14 .34 -.49 -.30 1.00 8.1 (3.9) 7.8 (4.0)
Role Conflict .06 .31 -.45 -.23 .09 1.00 18.3 (4.8) 16.6 (5.4)
Role Ambiguity -.26 .31 -.37 -.19 .26 .58 1.00 10.6 (3.4) 10.0 (4.4)
Variance Extracted .42 .53 .50 .46 .57 .62 .68
aStandard deviationsare shown in parentheses.

84/ Journal ofMarketing, April 1998

 TABLE 4
Standardized Structural Path Estimates Across the FUll, Male, and Female Samples

Path from
Role Conflict Role Ambiguity Job Performance Work/Nonwork Job Satisfaction Life Satisfaction

to Full Men Women Full Men Women Full Men Women Full Men Women Full Men Women Full Men Women

JP .24 .38 -.06 -.41 -.19 -.46

t-value 2.6 2.6 .41 -4.7 -1.8 -3.2

WNW .18 .14 .23 .20 .28 .14

t-value 2.0 1.3 1.4 2.3 2.5 .92

JS -.27 -.32 -.33 -.03 -.17 .09 .09 .21 .06 -.55 -.45 -.58
t-value -3.6 -3.4 -2.7 -.37 -1.8 .70 1.7 2.6 .78 -9.6 -5.2 -7.9

LS .16 .13 .17 -.38 -.40 -.35 .18 .12 .22

t-value 2.6 1.6 2.3 -4.7 -3.4 -3.2 2.2 1.1 2.2

Quit -.44 -.62 -.28 -.10 -.08 -.13

m
3 t-value -6.6 -5.3 -3.3 -1.5 -.91 -1.4
"'CI

X2 14.5 5.0 9.5

f df 14 7 7
aJ .39 .20
(II P .66
f tori! (J. =.01 = 2.3.
...0" tOril (J. =.05 = 1.6 .
3' toril (J. =.10 = 1.3.
l:\l

f
2"
m
~,

~
I-
 FIGURE 2
Standardized Sturctural Path Coefficients for the Overall, Male, and Female Samples

Role Stress

/
/ re..~
f''Y -.03/-.17/-.09

overall: X2 = 14.5, df = 14, P = .39

male: X2 = 5.0, df = 7, p = .66
female: X2 = 9.5, df = 7, P = .20

Note: The path estimates are presented for the overall sample first, followed by the male sample, with the coefficient estimated for the female
sample appearing third. For significance levels, see Table 4.

conflicts with predictions. The estimate among women (Y2.1 (p < .001), whereas among women the estimate is -.28 (p <
= .14) is smaller than it is among men (Y2.1 = .28). The X2 .01). Moreover, the X2 difference (1 df) between the totally
that results from constraining these paths to be equal across free model and a model adding a path equality constraint is
samples is 15.4 with 16 degrees of freedom and does not 4.1 (p < .05). Overall, H4 is supported.
indicate a significantly worse fit than the totally free model Other results. Given the general interest in the degree to
(X 2 = .9 with 2 degrees of freedom; n.s.). Therefore, there is
which male and female employee behavior might differ and
only partial support for this hypothesis in the form of a
an argument that findings showing no differences are equally
greater role conflict-worklnonwork conflict relationship
as important as those showing differences (Lefkowitz 1994),
among female service providers.
a look at the overall pattern of relationships between men
H 3. H3 predicts that male service providers will display and women is warranted. Path estimates resulting from mul-
a stronger positive relationship between job and life satis- tisample analyses shown in Table 4 exhibit more relatively
faction than will female service providers. Neither the path small (.10 or less) than large differences. Therefore, the
estimates (~4,3 = .12, n.s., for men and ~4,3 =.22, p < .05, for observed invariance in correlation estimates noted previ-
women) nor the X2 difference (.4, 1 df) that results from con- ously is concentrated among a few relationships.
straining this path across samples support H3. Results sug- Among nonhypothesized differences in relationships,
gest that female service providers account for the significant only two exceed .10. The worklnonwork conflict-job satis-
positive coefficient observed in the overall sample (~4,3 = faction relationship path estimate is -.55 (p < .001) overall,
.18, P < .05). Whereas the relationship is significant and pos- -.45 (p < .001) for men, -.58 (p < .001) for women, and is
itive for women, it is not significant for men. directionally consistent with the notion that job satisfaction
H 4. H4 predicts that the negative relationship between is damaged more by worklnonwork conflict among women.
job satisfaction and quitting intent is greater in magnitude Given similar average respondent hours worked per week,
among male service providers. Path estimates are consistent this difference is consistent with a gender role explanation
with this prediction. Among men, this path estimate is -.62 of worklnonwork conflict (Gutek, Searle, and Klepa 1991).

86/ Journal of Marketing, April 1998

In addition, the job performance-job satisfaction estimate is Gender-Based Differences in Relationships
.09 (p < .05) overall, .21 for men (p < .05), and only .06 Schul and Wren (1992) examine differences in important
(n.s.) for women. The observed variation complements attitudinal and behavioral constructs across gender in an
meta-analytic research into the perfonnance-satisfaction industrial sales setting and conclude that there were too few
relationship and offers a potential explanation for differing differences to recommend highly different managerial poli-
results (Brown and Peterson 1994). This result is also con- cies for men and women. Here, four hypotheses regarding
sistent with a male orientation toward mastery. gender-based differences in relationships were hypothesized
specifically. These hypotheses were based largely on social
role theory that contrasted work versus gender-based role
Discussion expectations. Similar to the previous study (Schul and Wren
We offer two important contributions. First, an organiza- 1992), not all predictions found empirical support. How-
tional model of customer-contact service providers' per- ever, two key relationships with important implications for
ceived role stress and related consequences, including job service quality and service provider well-being exhibited
and life satisfaction, work/nonwork conflict, performance, significant and nontrivial differences.
and quitting intent, is tested. This model builds on the rela- First, service provider role stress affects job performance
tively scant literature that deal with service provider/cus- differently among men than among women. Results suggest
tomer-contact employee behavior. Second, potential that female service providers' performance is affected more
gender-based differences in relationships between con- negatively by increased role conflict or role ambiguity. The
structs constituting the overall theoretical model are exam- observed path estimate between role conflict and job perfor-
ined. Specific attention is paid to gender-based differences mance is significant, nontrivial, and positive among men, but
in the outcomes of role stress on the job. Differences in the like path among women is nonsignificant. Ambiguity's
these and similar relationships indicate that gender should effect is negative overall, but the observed path between role
be considered a potential moderator. ambiguity and job performance is significantly more nega-
tive among women than among men. These results are con-
Overall Results sistent with sex role characteristics, which suggests a more
aggressive reaction to stress among men.
Structural equations analysis supports the hypothesized Previous research suggests variation in the effects of the
model. Path estimates suggest that service provider role stress-performance relationship among boundary spanners.
stress affects customer-contact service providers' job perfor- Although a plurality of studies report a positive relationship
mance, work/nonwork conflict, and job satisfaction directly (Brown and Peterson 1994), this finding should be recon-
and influences life satisfaction and quitting intent indirectly. sidered because most organizational behavioral research has
However, it appears that stress-related effects can be disag- been conducted using predominantly male samples. By not
gregated into productive and counterproductive compo- considering gender as a moderator, previous studies can
nents. A meta-analysis of boundary spanners in general report an attenuated relationship for men and an overstated
commented on the conflicting findings with respect to this and potentially misleading relationship for women.
relationship by discussing salespersons' capability to cope Second, job satisfaction affects quitting intent differ-
with stress by avoiding confrontation (Brown and Peterson ently among female and male service providers. Although
1994). In a service-providing situation, a customer sits at the relationship is significant and negative for both, the path
your table or comes to your counter, making escape quite estimates suggest a stronger relationship for men than for
difficult. Therefore, results presented here support the idea women. Therefore, it may be more difficult to keep a less
that conflict is dealt with most productively through con- than satisfied group of male service providers employed
fronting rather than avoiding the situation. than a like-minded group of female service providers. The
Results also point to the important role played by the stronger relationship between job satisfaction and quitting
work/nonwork interface. Increased stress on the job, mani- intent observed among men also means that work/nonwork
conflict has a greater indirect impact on quitting intent
fested as either role conflict or ambiguity, induces greater
among men than among women (.28 versus .16). Stress-
work/nonwork conflict as evidenced by significant, positive
related constructs and selected organizational outcomes
path coefficients. Although work/nonwork conflict is nor-
similar to those considered here explain male service
mally addressed in more professional domains with employ- providers' quitting intent better than they do that of female
ees who generally are older and married with children, in service providers. Thus, female service providers might
this sample of frontline service providers, work/nonwork leave their jobs for reasons other than dissatisfaction. This
conflict is a major contributor to job dissatisfaction, life dis- might be due partially to the limited opportunities for mobil-
satisfaction, and, indirectly, higher quitting intent. Although ity that women experience.
we might expect work/nonwork to be less of a factor among Contrary to traditional role expectations, men did not
employees who appear to have a "simpler" nonwork living exhibit a stronger job satisfaction-life satisfaction relation-
arrangement, the results suggest otherwise. Results from a ship. In contrast, a slightly more positive and significant
study of retail managers suggest a far weaker work/non- relationship was observed among female service providers.
work-job satisfaction relationship than that reported here Evidently, women today identify with their work, even
(cf. Good, Page, and Young 1996). in nonprofessional settings, to a degree that makes its

Employee Behavior ina Service Environment /87

importance to their overall life satisfaction virtually indis- the effective management of turnover is, to some degree, a
tinguishable from its importance to men. Businesses accom- function of helping service providers manage their
modating the work/non work interface can expect higher work/nonwork conflict.
levels of employee life satisfaction as well as job satisfac- Because service-providing employees are relatively
tion. Also, the relationship between role stress and young, single, and more often part-time than employees in
work/non work conflict did not differ significantly across other workplaces, managers might not recognize potentially
women and men. Both groups indicate a strong relationship deleterious effects of work/nonwork conflict. Common
between work-related role stress and perceptions of conflict restaurant, hotel, and retail practices, which include irregu-
between work and nonwork responsibilities. lar schedules, limited weekend time off, and altering sched-
ules with little or no notice, affect the work/nonwork
Managerial Implications interface. To the extent that these situations can be mini-
Overall model. The effective management of employee mized, the payoff is decreased turnover.
satisfaction, turnover, and service quality is essential to suc- Gender-based differences. Managers must consider the
cess in a service industry. The overall model of service gender effects found here in interpreting the role stress-
provider behavior suggests important roles for role stress performance relationship. Results indicate that women are
and work/nonwork conflict in understanding variance in affected more severely by increased stress. Therefore, in
turnover and service quality. Several interesting findings managing role conflict and role ambiguity, these differential
involve the productive and nonproductive reactions to work- effects might need to be considered. For example, female
related role stress. employees in high stress, customer-contact environments
The service-providing environment presents significant might need more concrete guidance about what they can and
friction between customer and organizational goals (e.g., cannot do to remedy conflict between managers' expecta-
incompatible requests, doing things that please the boss but tions and customer demands.
not the customer, and so on). It is the effective management Once again, these results can be pertinent to the debate
of this friction by service providers that sometimes can on process versus output controls. For example, if the ratio-
change a potentially disastrous service encounter to one that nale that men respond aggressively to conflict by taking lib-
creates high customer satisfaction and helps build a long- erties with stated policies and rules is valid, the results
lasting relationship (Bitner, Booms, and Tetreault 1990). suggest a differential effect of output versus process control
The positive role conflict-job performance relationship sug- on male and female employees. As no operationalization of
gests that an absence of stress perceptions among service control methods is presented in this study, this is clearly a
providers could lower overall job performance and customer research area worthy of further attention.
perceptions of service quality. Although a manager might try The differential results observed in the job satisfaction-
to reduce conflict through greater communication of proce- turnover relationship also have implications for practice. A
dures and the implications of deviating from those proce- straightforward implication is that steps taken to improve
dures (Reardon and Enis 1990), if the result is reduced job satisfaction will reduce turnover more effectively among
perceptions of initiative to resolve conflicts, lower service men than among women. That is, managerial policies might
quality could result. Increased flexibility is likely to improve have more effect on male than on female quitting intent. The
service quality but increase stress (Hartline and Ferrell stronger relationship among men might suggest that women
1996). In contrast, role ambiguity, indicated by attitudes will tolerate more negative working conditions on the job
such as uncertainty regarding authority, seems to have only before they quit. Upper management concerned with acting
counterproductive outcomes. ethically should take extra precaution to reduce or eliminate
Service managers, as a result, might consider acknowl- these negative work conditions.
edging the inherent customer-contact stress elements (e.g.,
incompatible requests, accepted by one and not the other) Limitations
and making clear the degree to which employees can "bend Studies of employee behavior always are subject to attack
rules" in performing their job duties. Clearly, this question on the basis of the constructs selected for study. Ideally,
ties into the current debate on the effects of process versus other constructs would be included. An operationalization of
output-based marketing controls (Lusch and Jaworski 1991; the control environment would be a worthy addition and
Singh, Verbeke, and Rhoads 1996). Rather than building a would extend the contribution offered here. Also, a self-
single type of control environment, service management report job performance measure was used here. Although
might consider a blend in which employees are motivated previous research shows considerable correspondence
by certain outcomes but also are given clear guidance on between self-report and other performance measures
how much deviation from accepted procedures will be tol- (Churchill et al. (985), the relationships examined might be
erated and/or rewarded. distorted because all measures are self-reports. From the
Overall results suggest an important role for work/non- customer's standpoint, amounts tipped would be good indi-
work conflict, which directly lowers job and life satisfaction cators of job performance. Evidence suggests that male and
and indirectly increases quitting intent. Furthermore, female consumers differ in their judgments of male and
work/non work conflict, to a large extent, facilitates the non- female service provider performance (Iacobucci and Ostrom
productive effects of role stress by mediating relationships (993). Further research might consider more closely the
between role ambiguity and important outcomes. Therefore, matching process between consumer and employee gender

881 Journal ofMarketing, April 1998

and its affect on performance perceptions. Furthermore, ing environments involving the dyadic interchange between
other affective constructs are worthy of consideration and customers and employees inevitably involve conflict that
might be relevant to potential gender versus work role dif- has beneficial and detrimental consequences. Furthermore,
ferences. Some possibilities for further consideration the study shows that men and women, though reacting sim-
include commitment and burnout. ilarly to many workplace constructs, react differently to role
Our sample provides service in full-service restaurants. stress and job satisfaction. Most organizational research has
However, a comparison of the results presented here with been conducted on wholly or predominantly male samples
those from other marketing contexts is worthwhile. For exam- and occupations; therefore, many of the relationships
ple, had the sample also involved marketing positions consid- demonstrated might be male phenomena. This study pro-
ered more professional, such as many business-to-business vides empirical evidence that at least some of these rela-
sales settings, further analyses could have examined level of tionships might vary between male and female service
professionalism or customer respect as further moderators. providers. Perhaps the results will help marketing managers
reexamine the treatment of service providers overall and of
Conclusion women in particular. At the least, the results provide evi-
We present results and implications relevant to the effective dence useful in developing theory related to managing cus-
management of customer contact positions. Service provid- tomer-contact service provider relationships.

APPENDIX
Description of Construct Item Indicators
(Abbre- (Abbre-
viation) Scale Items viation) Scale Items
Job Performance (JP) LS1. interesting-boring
Relative to other workers here, I ... (1 = "Strongly Disagree" LS2. enjoyable-miserable
to 5 = "Strongly Agree"): LS3. worthwhile-useless
LS4. full-empty
JP1. am a top performer.
LS5. hopeful-discouraging
JP2. average higher sales per check than other servers.
LS6. rewarding-disappointing
JP3. am in the top 10 percent of servers here.
LS7. friendly-lonely
JP4. get along better with customers than do others.
JP5. know more about menu items.
Quitting Intent (QI)
JP6. know what my customers expect.
Respondents were asked to rate their chances of ... (seven-
JP7. get better tips than most.
point scale ranging from 1 = "Excellent Chance" to 7= "Ter-
rible Chance") (Bluedorn 1982):
Work/Nonwork Conflict (WNW)
The impact your current job has on (insert phrase) is ...? Q11. Quitting this job in the next three months.*
("Strong negative impact" = 1 to "Strong positive impact" = Q12. Quitting this job sometime in the next year.*
5) (Burke, Weir, and DuWors 1979):
Role Stress
WNW1. your mental and physical state away from work.
Five-point Likert (1 = "Strongly Disagree" to 5 = "Strongly
WNW2. your participation in home activities.
Agree") (Rizzo, House, and Lirtzman 1970)
WNW3. concern for your health and/or safety.
WNW4. your personal development. Conflict (RC)
WNW5. your weekend, vacation time, and social life. RC1. I receive an assignment without the manpower to
complete it.
Job Satisfaction (JS) RC2. I sometimes have to bend a rule or policy in
Five-point Likert scale (1 = "Strongly Disagree" to 5 = order to carry out my job.
"Strongly Agree") RC3. I receive incompatible requests from two or more
people.
JS1. I consider my job unpleasant.*
RC4. 1do things that are apt to be accepted by one
JS2. I am often bored with my job.*
person and not accepted by others.
JS3. I feel fairly well-satisfied with my present job.
RC5. I receive assignments with inadequate resources
JS4. Most of the time, I have to force myself to go to
and materials to execute them.
work.*
RC6. I work on unnecessary things.
JS5. I definitely dislike my work.*
JS6. Most days, I am enthusiastic about my work. Ambiguity (RA)
JS7. My job is pretty uninteresting.* RA1. 1feel certain about how much authority I have.*
JS8. I find real enjoyment in my work. RA2. There are clear, planned goals and objectives for
JS9. I am disappointed I ever took this job.* my job.*
RA3. I know what my responsibilities are.*
Life Satisfaction (LS) RM. I know exactly what is expected of me.*
Respondents marked the blank (seven-point semantic dif- RA5. The explanations are clear as to what I have to do.*
ferential scored from 7 to 1) that described best how he or
she saw his or her life at that particular point in time (Quinn
and Shepard 1974).
Note: Starred items were reversed scaled prior to analyses, so higher scores indicate higher levels of constructs.

Employee Behavior in a Service Environment /89

REFERENCES
Adams, Gary A., Lynda A. King, and Daniel W. King (1996), Chao, Georgia T., Anne M. O'Leary-Kelly, Samantha Wolf,
"Relationships of Job and Family Involvement, Family Social Howard J. Klein, and Philip D. Gardner (1994), "Organiza-
Support, and Work-Family Conflict with Job and Life Satisfac- tional Socialization: Its Content and Consequences," Journal of
tion," Journal ofApplied Psychology, 81 (August), 411-20. Applied Psychology, 79 (October), 730--43.
Anderson, James C. and David W. Gerbing (1992), "Assumptions Churchill, Gilbert A., Jr., Neil M. Ford, Steve W. Hartley, and
of the Two-Step Approach to Latent Variable Modeling," Soci- Orville C. Walker Jr. (1985), "The Determinants of Salesperson
ological Methods and Research, 20 (February), 321-33. Performance: A Meta Analysis," Journal of Marketing
Bandalos, Deborah and Jeri Benson (1990), "Testing the Factor Research, 22 (May), \03-18.
Structure Invariance of a Computer Attitude Scale over Two Cournoyer, Robert J. and James R. Mahalik (1995), "Cross-
Grouping Conditions," Educational and Psychological Mea- Sectional Study of Gender Role Conflict Examining College-
surement, 50 (Spring), 49-60. Aged and Middle-Aged Men," Journal of Counseling Psychol-
Bagozzi, Richard P. (1978), "Salesforce Performance and Satisfac- ogy, 42 (January), ll-19.
tion as a Function of Individual Difference, Interpersonal, and Crosby, Faye J. and Karen L. Jaskar (1993), "Women and Men at
Situational Factors," Journal of Marketing Research, 15 Home and at Work: Realities and Illusions," in Gender Issues
(November), 517-31. in Contemporary Society, Stuart Oskamp and Mark Costanzo,
Behrman, Douglas, H. and William D. Perreault Jr. (1984), "A eds. Newbury Park, CA: Sage Publications, 143-72.
Role Stress Model of the Performance and Satisfaction of Deaux, K. (1985), "Sex and Gender," Annual Review of Psychol-
Industrial Salespersons," Journal ofMarketing, 48 (Fall), 9-21. ogy, 36, 49-81.
Benham, Harry C. (1993), "Unemployment Durations and Post- Dubinsky, Alan J. and Steven W. Hartley (1986), "A Path-Analytic
Unemployment Wages: Differences by Race and Gender," The Study of a Model of Salesperson Performance," Journal of the
Journal of Economics, 19 (Spring), 31-37. Academy of Marketing Science, 4 (Spring), 36--46.
Berger, J., S.J. Rosenholtz, and M. Zelditch Jr. (1980), ''The Eco- Duxbury, Linda Elizabeth and Christopher Alan Higgins (1991),
nomic Emergence of Women," Annual Review of Sociology, 6, "Gender Differences in Work-Family Conflict," Journal of
479-508. Applied Psychology, 76 (February), 60-74.
Biernat, Monica and Camille B. Wortman (1991), "Sharing of Eagly, Alice H. (1987), Sex Differences in Social Behavior: A
Home Responsibilities Between Professionally Employed Social-Role Interpretation. Hillsdale, NJ: Lawrence Erlbaum
Women and Their Husbands," Journal of Personality and
Associates.
Social Psychology, 60 (June), 844-60. - - - , Mona G. Makhijani, and Bruce Klonsky (1992), "Gender
Bitner, Mary Jo (1992), "Servicescapes: The Impact of Physical
and Leadership Style: A Meta Analysis," Psychological Bul-
Surroundings on Customers and Employees," Journal of Mar-
letin, III (January), 3-22.
keting, 56 (April), 57-71.
Fiske, Susan T. and Laura E. Stevens (1993), "What's So Special
- - - , Bernard H. Booms, and Lois A. Mohr (1994), "Critical
About Sex? Gender Stereotyping and Discrimination," in Gen-
Service Encounters: The Employee's View," Journal of Mar-
der Issues in Contemporary Society, S. Oskamp and M.
keting, 58 (October), 95-\06.
Costanzo, eds. Newbury Park, CA: Sage Publications, 173-96.
- - - , - - - , and Mary Stanfield Tetreault (1990), "The Ser-
Fornell, Claes and David Larcker (1981), "Evaluating Structural
vice Encounter: Diagnosing Favorable and Unfavorable Inci-
Equations Models with Unobserved Variables and Measure-
dents," Journal of Marketing, 54 (January), 71-84.
ment Error," Journal of Marketing Research, 18 (February),
Bluedorn, Allen C. (1982), "A Unified Model of Turnover from
Organizations," Human Relations, 35 (February), 135-53. 39-50.
Boles, James S. and Barry J. Babin (1996), "On the Front Lines: Gerbing, David W. and James C. Anderson (1992), "Monte Carlo
Stress, Conflict, and the Customer Service Provider," Journal Evaluations of Goodness of Fit Indices for Structural Equations
of Business Research, 37 (September), 41-50. Models," Sociological Methods and Research, 21 (November),
Brayfield, A.H. and H.F. Rothe (1951), "An Index of Job Satisfac- 132-60.
tion," Journal ofApplied Psychology, 35 (October), 307-1 I. Good, Linda K., Grovalynn F. Sisler, and James W. Gentry
Brown, Steven P. and Thomas Leigh (1996), "A New Look at Psy- (1988), "Antecedents of Turnover Intentions Among Retail
chological Climate and Its Relationship to Job Involvement, Management Personnel," Journal of Retailing, 64 (Fall),
Effort, and Performance," Journal of Applied Psychology, 81 295-314.
(August), 358-68. - - - , Thomas J. Page, and Clifford E. Young (1996), "Assess-
- - - and Robert A. Peterson (1993), "Antecedents and Conse- ing Hierarchical Differences in Job-Related Attitudes and
quences of Salesperson Job Satisfaction: Meta-Analysis and Turnover Among Retail Managers," Journal of the Academy of
Assessment of Causal Effects," Journal ofMarketing Research, Marketing Science, 24 (Spring), 148-56.
30 (February), 63-77. • Greenhaus, Jeffrey H., Arthur G. Bedeian, and Kevin W. Mossh-
- - - and - - - (1994), "The Effect of Effort on Sales Perfor- older (1987), "Work Experiences, Job Performance, and Feel-
mance and Job Satisfaction," Journal of Marketing, 58 (April), ings of Personal and Family Well-Being," Journal of
70-80. Vocational Behavior, 31 (October), 200-215.
Burke, Ronald J., Tamara Weir, and Richard E. DuWors Jr. (1979), - - - and Nicholas J. Beutell (1985), "Sources of Conflict
"Type A Behavior of Administrators and Wives' Reports of Between Work and Family Roles," Academy of Management
Marital Satisfaction and Well-Being," Journal of Applied Psy- Review, \0 (January), 200-215.
chology, 64 (February), 57-65. Gutek, Barbara A., Sabrina Searle, and Lilian Klepa (1991),
Busch, Paul and Ronald F. Bush (1978), "Women Contrasted to "Rational Versus Gender Role Explanations for Work-Family
Men in the Industrial Sales force: Job Satisfaction, Values, Role Conflict," Journal ofApplied Psychology, 76 (4), 560-68.
Clarity, Performance, and Propensity to Leave," Journal of Hartline, Michael D. and O. C. Ferrell (1996), "The Management
Marketing Research, 15 (August), 438--48. of Customer-Contact Service Employees: An Empirical Inves-
Byrne, B. M. (1988), "Measuring Adolescent Self-Concept: Facto- tigation," Journal of Marketing, 60 (October), 52-70.
rial Validity and Equivalency of the SDQIII Across Gender," Himle, David P., Srinika Jayaratne, and Paul Thyness (1989), ''The
Multivariate Behavioral Research, 23 (July), 361-75. Effects of Emotional Support on Burnout, Work Stress and

90I Journal of Marketing, April 1998

 Mental Health Among Norwegian and American Social Work- Parasuraman, Saroj, Jeffrey H. Greenhaus, Samuel Rabinowitz,
ers," Journal of Social Science Research, 13 (January), 27-45. Arthur G. Bedeian, and Kevin W. Mossholder (1989), "Work
Hodson, Randy (1989), "Gender Differences in Job Satisfaction: and Family Variables as Mediators of the Relationship Between
Why Aren't More Women Dissatisfied?" The Sociological Wives' Employment and Husbands' Well-Being," Academy of
Quarterly, 30 (3), 385-99. Management Journal, 32 (March), 185-201.
Hoffman, Curt and Nancy Hurst (1990), "Gender Stereotypes: Per- Pulkkinen, Lea (1996), "Female and Male Personality Styles: A
ception or Rationalization," Journal of Personality and Social Typological and Developmental Analysis," Journal of Person-
Psychology, 58 (February), 197-208 ality and Social Psychology, 70 (June), 1288-1306.
Iacobucci, Dawn and Amy Ostrom (1993), "Gender Differences in Quinn, R. P. and L. 1. Shepard (1974), The 1972-73 Quality of
the Impact of Core and Relational Aspects of Services on the Employment Survey. Ann Arbor: Institute for Social Research,
Evaluation of Service Encounters," Journal of Consumer Psy- University of Michigan,.
chology, 2 (3), 257-86. Ramaswami, Sridhar N. (1996), "Marketing Controls and Dys-
Jick, Todd D. and Linda F. Mitz (1985), "Sex Differences in Work functional Employee Behaviors: A Test of Traditional and Con-
Stress," Academy of Management Review, 10 (3), 408-20. tingency Theory Postulates," Journal of Marketing, 60 (April),
Johnston, Mark W., Charles M. Futrell, A. Parasuraman, and 105-20.
William C. Black (1990), "A Longitudinal Assessment of the Reardon, Kathleen K. and Ben Enis (1990), "Establishing a Com-
Impact of Selected Organizational Influences on Salespeople's pany-Wide Customer Orientation Through Persuasive Internal
Organizational Commitment During Early Employment," Jour- Marketing," Management Communications Quarterly, 3 (Feb-
nal of Marketing Research, 27 (August), 333-44. ruary), 376-87.
Josephs, Robert A., Hazel Rose Markus, and Romin W. Tafarodi Reifman, Alan, Monica Biernat, and Eric L. Lang (1991), "Stress,
(1992), "Gender and Self-Esteem," Journal of Personality and Social Support, and Health in Married Professional Women
Social Psychology, 63 (3), 391-402. with Small Children," Psychology of Women Quarterly, 15,
Kenney, David A. (1979), Correlation and Causation. New York: 431-45.
John Wiley & Sons. Rizzo, John R., Robert 1. House, and Sidney I. Lirtzman (1970),
Kohli, Ajay (1985), "Some Unexplored Supervisory Behaviors and "Role Conflict and Ambiguity in Complex Organizations,"
Their Influence on Salespeople's Role Clarity, Specific Self- Administrative Science Quarterly, 15 (June), 150-63.
Esteem, Job Satisfaction, and Motivation," Journal of Market- Schul, Patrick L., Steven Remington, and Robert L. Berl (1990),
ing Research, 22 (November), 424-33. "Assessing Gender Differences in Relationships Between
Lefkowitz, Joel (1994), "Sex-Related Differences in Job Attitudes Supervisory Behaviors and Job-Related Outcomes in the Indus-
and Dispositional Variables: Now You See Them, Now You trial Salesforce," Journal of Personal Selling and Sales Man-
Don't," Academy of Management Journal, 37 (April), 323-49. agement, 10 (Summer), 1-16.
Lusch, Robert F. and Bernard J. Jaworski (199\), "Management - - - and Brent M. Wren (1992), "The Emerging Role of
Controls, Role Stress, and Retail Store Manager Performance," Women in Industrial Selling: A Decade of Change," Journal of
Journal of Retailing, 67 (Winter), 397-419. Marketing, 56 (July), 38-54.
- - - and Ray A. Serpkenci (1990), "Personal Differences, Job Siguaw, Judy A. and Earl D. Honeycutt (1995), "An Examination
Tension, Job Outcomes, and Store Performance: A Study of of Gender Differences in Selling Behaviors and Job Attitudes,"
Retail Store Managers," Journal of Marketing, 54 (January), Industrial Marketing Management, 24 (January), 45-52.
85-101. Singh, Jagdip (1993), "Boundary Role Ambiguity: Facets, Deter-
Michaels, Ronald E. and Andrea L. Dixon (1994), "Sellers and minants, and Impacts," Journal of Marketing, 57 (April),
Buyers on the Boundary: Potential Moderators of Role Stress- 11-31.
Job Outcome Relationships," Journal of the Academy of Mar- - - - , William Verbeke, and Gary K. Rhoads (1996), "Do Orga-
keting Science, 22 (Winter), 62-73. nizational Practices Matter in Role Stress Processes? A Study of
Nelson, D. L. and J. C. Quick (1985), "Professional Women: Are Direct and Moderating Effects for Marketing-Oriented Bound-
Distress and Disease Inevitable?" Academy of Management ary Spanners," Journal of Marketing, 60 (July), 69-86.
Review, 10,206-18. Tait, M., M. Y. Padgett, and T. T. Baldwin (1989), "Job and Life
Netemeyer, Richard G., James S. Boles, and Robert McMurrian Satisfaction: A Reevaluation of the Strength of the Relationship
(1996), "Development and Validation of Work-Family Conflict and Gender Effects as a Function of the Date of the Study,"
and Family-Work Conflict Scales," Journal of Applied Psy- Journal ofApplied Psychology, 74 (3), 502-507.
chology, 81 (August), 400-410. Thomas, Linda T. and Daniel C. Ganster (1995), "Impact of Fam-
- - - , Mark W. Johnston, and Scott Burton (1990), "Analysis ily-Supportive Work Variables on Work-Family Conflict and
of Role Conflict and Ambiguity in a Structural Equations Strain: A Control Perspective," Journal ofApplied Psychology,
Framework," Journal of Applied Psychology, 75 (April), 80 (I), 6-15.
148-57.

Employee Behavior ina Service Environment /91