Causal model of stress and coping: Women in management

Joui1992,

rnal of Counseling Psychology12, Vol. 39, No. 2, 227-239

Copyright 1992 by the American Psychological Association Inc.0O22-O167/92/$3.0O

Causal Model of Stress and Coping: Women in ManagementBonita C. Long and Sharon E. Kahn

Department of Counselling Psychology, University of British ColumbiaVancouver, British Columbia, Canada

Robert W. SchutzSchool of Physical Education and Recreation, University of British Columbia

Vancouver, British Columbia, Canada

A model of managerial women's stress was tested (N = 249) with structural equation modeling.The model was developed from Lazarus's (1966) theoretical framework of stress/coping andincorporated 3 causal antecedent constructs (Demographics, Sex Role Attitudes, Agentic Traits),4 mediating constructs (Environment, Appraisals, Engagement Coping, Disengagement Coping),and 3 outcomes (Work Performance, Distress, Satisfaction). The final model, found to be mostplausible in the sample population, accounted for 56% of the total variance among the constructs.Lazarus's theory of psychological stress, which postulates a central role for cognitive appraisalsand coping, was supported. In addition, agentic traits and sex role attitudes had both direct andindirect effects on outcome variables. Implications for career development theory and counseling,as well as limitations of the study, are discussed.

Clients often call on counseling psychologists to help themreduce stress and improve well-being. Empirical evidencelinks many sources of occupational stress to psychologicaland behavioral outcomes. For example, employees who per-form continuous repetitive tasks may find that their job stresseventually becomes their long-term career stress. Similarly,career events, such as the loss of a desired promotion, mayresult in stress and negative outcomes that reach far beyondthe discrete career event (cf. Latack, 1989). Unfortunately forcounseling psychologists who are interested in the psychoso-cial factors that mediate the stress/well-being relationship,much of the stress research is inadequate (for a review seeMatheny, Aycock, Pugh, Curlette, & Silver Cannella, 1986).Psychosocial variables are rarely studied in combination withone another, and despite the large amount of research onfactors related to occupational stress, the lack of an integrativetheory to describe relationships among variables makes itdifficult to determine the relative strength of variables (cf.Fassinger, 1987). This second problem is particularly truewith regard to the study of women's employment experience.Most stress research derives from observations of White,middle-class men and thus ignores variables specific to the

Completion of this article was made possible in part by the SocialScience and Humanities Research Council Strategic Grant 482-87-006. We wish to thank Gerald Haag for assistance with the analysisand Colleen Haney, Lynne Nikolychuk, and the many student assis-tants for their help with the data collection.

A summary of this work was presented in June 1989 to the"Women's Career Development" symposium at the annual conven-tion of the Canadian Psychologist Association, Halifax, Nova Scotia,Canada.

Correspondence concerning this article should be addressed toBonita C. Long, Department of Counselling Psychology, 210-5780Toronto Road, University of British Columbia, Vancouver, BritishColumbia, Canada V6T 1L2.

understanding of women's well-being (Eisler, Skidmore, &Ward, 1988).

Women employed in male-dominated careers cope withparticular employment stressors, for example, gender-rolestereotypes and occupational sex discrimination (Terborg,1985). By applying a model of stress and coping to women inmanagement positions, we hope to understand women's em-ployment experiences and the effects of women's appraisalsand coping responses on their well-being. Therefore, this studyis intended to test directly, and if necessary to modify, thetheoretical framework of psychological stress and coping firstproposed by Richard Lazarus (1966). Lazarus and Folkman's(1984) stress theory defines psychological stress as a relation-ship between person and environment that is appraised bythe person as taxing or exceeding his or her resources andendangering the person's well-being. The theory "identifiestwo processes, cognitive appraisal and coping, as critical me-diators of stressful person-environment relationships andtheir immediate and long-term outcomes" (Folkman, Laza-rus, Gruen, & DeLongis, 1986, p. 572). Thus, appraisals ofthe stimulus itself and the choice of and effectiveness of one'scoping actions are crucial to the stress process.

Folkman and Lazarus (1988) suggested that antecedentconditions, such as motivations (e.g., values, commitments,and goals), beliefs about oneself, and recognition of personalresources for coping, moderate the stress process because theyinteract with other conditions to produce an outcome. Copingprocesses, however, are considered mediators because copingdetermines which of the various long-term effects occur (e.g.,morale, health, psychological functioning). Mediating proc-esses, hypothesized to change the relationship between theantecedent and the outcome variable, are generated in thestressful encounter. Thus, mediating processes, or mediators,are distinguished from moderators. The framework thatguides the selection of variables in this study consists of (a)antecedent variables, (b) processes that mediate the person-environment encounter, and (c) long-term outcomes.

227

228 B. LONG, S. KAHN, AND R. SCHUTZ

To provide a much needed link to career theory (Heppner,Cook, Strozier, & Heppner, 1991; Latack, 1989), we basedour choice of variables not only on the stress/coping literatureand in-depth interviews with professional women (B. C. Long,1988) but also on aspects of career development theory thatare relevant to women's nontraditional careers. Career theoryhas identified individual differences as well as social andenvironmental forces that are associated with career commit-ment and choice of nontraditional careers for women (Betz& Fitzgerald, 1987). Furthermore, appraisal and coping proc-esses women engage in when struggling with the demands ofnontraditional occupations have not been linked to the pre-dictor variables that are associated with women's career ad-justment (cf. Heppner et al., 1991).

We focused on three types of causal antecedent variables:(a) demographics, occupational roles, and family roles; (b) sexrole attitudes; and (c) agentic traits. Higher occupationalstatus, income, and education have been related to greaterproblem-focused coping (e.g., Pearlin & Schooler, 1978). Paidwork is generally related to better mental health for marriedwomen with children (Repetti, Matthews, & Waldron, 1989),whereas marriage and motherhood are negatively associatedwith strong career orientations and nontraditional choices(Hock, Morgan, & Hock, 1985).

Career theory and empirical evidence indicate that sex roleattitudes, which may modify the meaning of work stress, playan important role in determining career orientation andchoice (Betz & Fitzgerald, 1987; Fassinger, 1985, 1990). Forexample, women espousing the traditional feminine rolemight appraise situations that require aggressive behavior asmore threatening and thus might be less adept at coping withthese events than women who hold more egalitarian sex roleattitudes. Although egalitarian sex role attitudes are related toa stronger career orientation for women, Bernard (1981)contends that women who step outside the socially ascribedroles of wife and mother, by placing careers before theirfamilies, experience emotional turmoil and stress.

A personality characteristic important to both career theoryand the stress process is one's sense of agency or efficacy(Bandura, 1986; Betz & Hackett, 1987). An optimistic senseof personal efficacy has been identified as a precursor tohuman attainment and positive well-being (Bandura, 1986).Studies have suggested that low perceived self-efficacy may bedetrimental to appraisals and perceptions of effective copingwith stress (e.g., Holahan & Holahan, 1987). Optimism, theexpectation that the future occurrence of desirable outcomesis high, irrespective of one's ability to control those outcomes,has been related to greater problem-focused coping and toless avoidance or disengagement coping (Scheier, Weintraub,& Carver, 1986). In addition, the sex-typed trait, instrumen-tality (Bern, 1981), is a particularly important measure ofagency because of its association with managerial success(Jackinowski, 1987) and its prediction of a strong careerorientation (Fassinger, 1990). Instrumentality has been asso-ciated with more problem-focused and less emotion-focusedcoping in response to stressful situations (Nezu & Nezu, 1987),and it has a strong relationship with high levels of adjustment(Feather, 1985; B. C. Long, 1989). Finally, a proactive health-enhancing coping style (e.g., being physically active, engaging

in leisure activities) is related to specific appraisals and copingbehaviors (Bruning & Frew, 1987). Thus, we examined lifecircumstances (e.g., family and work roles, education andincome), sex role attitudes, and agentic traits (self-efficacy,optimism, instrumentality, proactive health behaviors) asmoderators of the stress process because these constructsrepresent means by which women may experience (a) differ-ential exposure to work stressors, (b) differential availabilityof resources, and (c) differential perceptions of the meaningsof employment stress.

Mediators of the person-environment interaction includeenvironmental constraints and resources, appraisals, and cop-ing strategies (Lazarus & Folkman, 1984). Substantial evi-dence indicates that perceptions of the constraints and re-sources of the work environment (e.g., high job demands, lowcontrol) are related to mental strain and job dissatisfaction(Karasek, Baker, Marxer, Ahlbom, & Theorell, 1981). How-ever, emotional support provided by coworkers and supervi-sors has been shown, in general, to lower occupational stressand improve physical and mental health (for a review seeCohen & Wills, 1985). In addition, aspects of the environ-ment, such as daily hassles (irritating, frustrating, distressingincidents that occur in everyday transactions; Lazarus, 1981)and the type of stressor event may influence appraisals andchoice of coping strategies. Evidence indicates that daily has-sles are associated with several negative outcomes (Kanner,Coyne, Schaefer, & Lazarus, 1981) and that stressor eventsthat are impersonal are related to greater problem-solvingcoping (Pearlin & Schooler, 1978), presumably because theyare more amenable to change than interpersonal events.

Evidence indicates that appraisals of "what is at stake" areassociated with different types of coping (Folkman, Lazarus,Dunkel-Schetter, DeLongis, & Gruen, 1986). In addition,appraisals of the stressor's importance and controllability maypredict specific types of coping strategies (B. C. Long, 1990;Parkes, 1986). Coping refers to a person's cognitive andbehavioral efforts to manage demands that are appraised astaxing or as exceeding their resources (Lazarus & Folkman,1984). Some forms of coping are consistently found to berelated to positive outcomes (e.g., coping directed towardaltering the situation, cognitive reappraisal of an encounter),whereas coping that does not change the event directly relatesto more negative outcomes (Billings & Moos, 1981; B. C.Long, 1988; Parasuraman & Cleek, 1984). Previous investi-gations have shown that people generally use multiple formsof coping in virtually every type of encounter (Folkman &Lazarus, 1980, 1985). Thus, in addition to appraisals andcoping strategies, we included environmental constraints andresources (work demands, control, and support), as well asdaily hassles and type of stressor, as constructs expected tomediate antecedents and outcomes in the stress model.

In the Lazarus and Folkman (1984) model, stress responsesare described in terms of immediate and long-term effects onphysiological, psychological, and social levels. Outcomes suchas job and life satisfaction and psychosomatic functioning areof particular interest. In addition, perceived ability to carryout the requirements of a job is an important outcome becauseit may be affected by the ways in which one copes withoccupational stress (Davidson & Cooper, 1984).

CAUSAL MODEL OF MANAGERIAL WOMEN'S STRESS 229

The present study was designed to apply the method ofstructural equation modeling to the study of factors thatinfluence women's experience of stress in the workplace. Themodel, presented in Figure 1, contains three independent(exogenous) latent variables and seven dependent (endoge-nous) latent variables. In the figure the circles represent latentvariables and the squares represent indicators (i.e., observedvariables). The double-headed arrows reflect the correlationsamong the exogenous factors, and the single-headed arrowsindicate the direction of the predicted relations among thelatent variables (in some cases there are reciprocal relations,hence, the two arrows). We hypothesized that the four Time2 endogenous variables mediate the relationships between the

causal antecedents (Demographics, Sex Role Attitudes, andAgentic Traits) and outcomes (Distress, Satisfaction, andWork Performance). We did not hypothesize direct pathsfrom Appraisals and Environment to outcomes at Time 3because, according to theory (Lazarus & Folkman, 1984), themediating effects of Disengagement Coping and EngagementCoping account for these relationships. In addition, we hy-pothesized that there are reciprocally determined relationshipsbetween pairs of outcomes and between pairs of mediators.For simplicity we indicate only paths of hypothesized theo-retical significance, not all expected direct paths, because weare most interested in the mediating processes of appraisalsand coping as postulated by Lazarus and Folkman (1984).

EXOGENOUSTime 1

ENDOGENOUSTime 2 Time 3

Figure 1. Full structural equation model of women's stress and coping. (DEM = Demographics; AGE= age; POS = months in position; EMP = years employed; MAR = marital status; PAR = parentalstatus; JLE = job level; INC = income; EDUC = education; SRA = Sex Role Attitudes; AWS =Attitudes Toward Women Scale; FEM = feminism; AGT = Agentic Traits; INS = instrumentality;LOT = Life Orientation Test; GSE = General Self-Efficacy Scale; PRC = preventive coping; ENV =Environment; PGD = personal growth dimension (work demands); SMD = system maintenancedimension (work demands); REL = relationship dimension (work support); HAS = daily hassles; STR= stressor type; APP = Appraisals; SES = threat to self-esteem; RES = loss of respect; GOA = threat togoal attainment; FIN = strain finances; IMP = episode importance; UPS = episode upsetting; CON =perceived control; DEN = Disengagement Coping; ENG = Engagement Coping; DIS = Distress; DEP= depression; ANX = anxiety; SOM = somatic symptoms; SAT = Satisfaction; LSA = Life SatisfactionScale; JSA = Job Satisfaction Scale; HJS = Hoppock Job Satisfaction Scale; WPE = Work Performance.)

230 B. LONG, S. KAHN, AND R. SCHUTZ

For example, we expected sex role attitudes and agentic traitsto directly affect satisfaction. Structural equation modelingwith LISREL provides modification indices that indicate thepossible omission of these direct paths. We examined thesemodification indices in an exploratory fashion when sup-ported by a plausible alternative theory. Thus, we examinedboth direct and indirect effects of the antecedent and mediat-ing variables.

We assessed the exogenous variables at Time 1, 1 monthprior to assessment of the stress appraisals, coping, and envi-ronment variables, which in turn, were assessed 1 month prior(Time 2) to the outcome variables (Time 3; Distress, Satisfac-tion, and Work Performance). The rationale for this design isthat it addresses some of the methodological pitfalls in seekingcausal links between stress and well-being. Given that theappropriate time period within which one can expect stressand coping to affect psychosomatic health and other measuresof well-being is controversial (cf. Kessler, 1983), we chose the1-month period as a reasonable compromise because (a) theconsequences of occupational stress and coping may taketime to impact; and (b) we wished to determine whethersituational stress/coping processes at Time 2 were sustained,in which case they would be related to psychosomatic healthand satisfaction at Time 3.

hold incomes of less than 40,000 Canadian dollars per annum; 45%had incomes between 40,000 and 80,000 Canadian dollars; and 25.5 %had incomes greater than 80,000 Canadian dollars.

In response to advertisements and networking, potential respond-ents contacted us by telephone. Participation was voluntary, and theconfidentiality of all individual data was guaranteed. Those womenwho met our criteria for inclusion were contacted by trained inter-viewers who arranged meetings (1 month apart), during which theyadministered structured questionnaires. At the first meeting, ques-tionnaires included an informed consent statement and cover letter,a demographic questionnaire regarding experience and status, andother exogenous measures. At the second and third meetings, stand-ardized self-report instruments were administered that assessed therespondent's experience of occupational stress and coping responses,perceptions of the work environment, daily hassles, psychosomaticwell-being, job/life satisfaction, and work performance. The stressand coping questionnaire was administered first to minimize theeffects of fatigue and to obtain as accurate an indication as possible.Interviewers were trained to facilitate the respondent's identificationof an appropriate work-related stressor, which was necessary for thecompletion of the stress and coping questionnaire. Although anidentical set of instruments was administered at the second and thirdmeetings, we used only the outcome variable data from the thirdinterview for this analysis. Completed questionnaires were returnedin sealed envelopes to ensure confidentiality and the freedom tochoose whether to participate.

Method

Subjects and Procedure

Participants were 294 female managers from both large and smallorganizations. The women were recruited through the media and bynetworking. Women who were full-time managers employed in non-traditional occupations (e.g., education and nursing were excluded)were invited to participate in a study of occupational stress andcoping. Criteria for nontraditional occupations were determined byconsulting Canada Census data (<35% of Canadian employees arewomen). Two independent raters classified occupations separatelywith 70% agreement. When raters did not agree about the occupa-tion codes, they reexamined the codes jointly until they reached aconsensus. The major categories were the following: officials andadministrators unique to government (5%), other managers andadministrators (59%), and occupations related to management andadministration (36%). We excluded 27 women from our originalsample on the basis of these ratings. An additional 18 respondentswho either did not report occupational stressors (n = 4) or whodropped out at a later date (n = 14; e.g., moved, resigned, or died)were removed from this analysis.

Respondents were full-time managers (34% entry level, 41% mid-dle level, and 26% executive level). The mean age of the respondentswas 38.8 years (SD = 7.7, range = 22-66); 57.0% were married,22.5% were single, and 20.5% were divorced/separated/widowed. Ofthese women, 56% had no children, 30% had two or more children,and 14% had one child. Of the women who had children, 66% hadchildren in school and 23% had children of preschool age. Sixty-twopercent of the women had completed postsecondary education. Thewomen had been in the work force an average of 16.1 years (SD =7.0, range = 2-40); the average organizational tenure was 8 years (SD= 6.2, range 0-35); and the average job tenure was slightly less than4 years (M = 46.5 months, SD = 42.7). The majority of the womenwere employed in organizations of less than 200 employees (44.6%)or greater than 1,000 employees (34.1%). Thirty percent had house-

Measures

Personal and job demographics. Each manager supplied data onage, months in position, total years employed, marital status, parentalstatus (number of children), job level, household income, and edu-cation (see Table 1 for demographic levels).

Sex-role attitudes. Two indicators were used to assess sex roleattitudes. The first indicator was a short form (15 items) of theAttitudes Toward Women Scale (Spence & Helmreich, 1978). Daugh-erty and Dambrot (1986) reported a test-retest reliability of .86 over3 months. Higher scores on the Attitudes Toward Women Scalereflect more liberal attitudes. The second sex role attitudes indicatorwas the following: summed responses to 2 items regarding the use ofthe title Ms. and the self-referent labeling of "feminist" (see Fassinger,1985, 1990). Higher scores indicate greater preference for using thetitle Ms. and for the use of the feminist label (after Smith & Self,1981).

Agentic traits. Four indicators were used to assess agentic person-ality traits. The first indicator was a measure of instrumentalitycontaining the masculine items of the Bern Sex-Role Inventory (Bern,1981), which assess the extent to which persons ascribe masculine(instrumental) qualities to themselves. Test-retest reliabilities rangefrom .76 to .94 (Bern, 1981). The second indicator was the LifeOrientation Test, a measure of optimism assessed in terms of gener-alized expectations of the occurrence of good outcomes in one's life(Scheier & Carver, 1985). Scheier and Carver reported a 4-weektest-retest reliability of .79. General self-Efficacy Scale, a measuredeveloped by Sherer etal.(1982) assesses general expectancies of self-efficacy. Long and Haney (1988) reported a 1-year test-retest relia-bility of .76. The fourth indicator, preventive coping consists of 6items representing ways to promote one's well-being and to reducethe likelihood of anticipated or potential problems (Wong & Reker,1983).

Environment. Five indicators were used to measure aspects of theenvironment: three work environment indices, daily hassles, and typeof stressor. The Work Environment Scale (Moos, 1981) assesses threecommonly identified work domains: relationship dimensions, per-

CAUSAL MODEL OF MANAGERIAL WOMEN'S STRESS 231

sonal growth or goal orientation dimensions, and system maintenanceand change dimensions. We developed two composite measures ofwork demands by summing two personal growth or goal orientationdimension subscales (autonomy and work pressures) and two systemmaintenance and change subscales (clarity and control). Autonomyand clarity subscales were reverse scored so that all subscales werescored in the same direction. The relationship dimensions included acomposite score of three subscales: involvement, cohesion, and su-pervisor support. One month test-retest reliabilities for the subscalesranged from .69 to .83 (Moos, 1981).

The fourth indicator, daily hassles, can be conceptualized as sourcesof repetitive personal frustration that, in and of themselves, may bequite innocuous. We used the Hassles Scale (Kanner et al., 1981), a117-item questionnaire in which respondents are instructed to indi-cate the occurrence of any items that have hassled them in the pastmonth (e.g., social activities, the environment, practical considera-tions, and finances). Work- and health-related items (23 items) werenot scored because they were redundant with items on other scales.Thus, the cumulative severity score was based on 94 items. Test-retest reliability (6 months) has been reported as .79 for frequencyand .48 for intensity (DeLongis, Coyne, Dakof, Folkman, & Lazarus,1982).

The fifth indicator consisted of occupational stressors. In responseto the Ways of Coping Checklist (Folkman, Lazarus, Dunkel-Schetter,DeLongis, & Gruen, 1986), respondents recorded the primary occu-pational stressor that had occurred during the previous month. Work-ing separately, two independent raters classified episodes on the basisof content into six predetermined categories. Agreement between thetwo raters was 81%. When the two raters did not agree, stressors werereexamined jointly and a consensus decision was reached. For theanalysis, interpersonal conflicts (60.2%) were collapsed into onecategory (coded 0), and staff shortages/overload and time pressures(39.8%) were collapsed into a second category (coded 1). This wasdone because there is some evidence that interpersonal conflicts atwork are particularly distressing (Rook, 1984).

As a validity check to determine whether the identified stressorwas perceived as harm/loss, threat, or challenge, participants wereasked to indicate "the primary emotion experienced as a result of theevent." Threat emotions (angry, disgusted, frustrated, disappointed)were reported by 40%; 50% reported harm emotions (worried, fearful,anxious, tense); 9% reported loss emotions (loss, depression, guilt);and only 1 person indicated benefit emotions (eager, pleased, exhila-rated; according to Folkman & Lazarus, 1988).

Appraisals. The several indicators used for the appraisal constructwere selected from items reported by Folkman and Lazarus (1980)and Parkes (1986) and from a review of the literature. We chose itemsdeemed particularly relevant to the work setting. First, primaryappraisal (what was at stake) was measured with 8 items previouslyused by Lazarus and Folkman (1984). A self-esteem "stake," identi-fied through factor analysis (Folkman & Lazarus, 1980), contains 5of the items, such as "losing affection of someone important to you,""losing your self-respect," and "appearing incompetent." The remain-ing three stakes include "losing respect for someone else," "notachieving an important goal at your job or in your work," and "astrain on your financial resources." In addition, we assessed on 5-point scales appraisals of the degree of episode importance, howupsetting the episode was, and how much control the respondent feltshe had in dealing with the situation.

Coping strategies. Coping strategies were assessed with a revisedversion of the Ways of Coping Checklist that includes items specificto the work environment. Item modifications have been described byB. C. Long (1990). For this study, the 42-item coping scale wassubmitted to exploratory factor analyses to define the scale. A two-factor solution showed a clean factor structure for both Varimax andOblimin rotations. The two factors, Disengagement Coping and

Engagement Coping, were used as single indicators of coping. Sixitems with low loadings (<.30) were excluded, and 19 and 14 itemswere defined to measure the Disengagement Coping and EngagementCoping factors, respectively (factor loadings are available from BonitaC. Long). The directions request that the respondent focus on theprimary occupational stressor that occurred during the previousmonth and respond to each coping strategy according to the degreeto which it was used to deal with the stressor.

Distress. We used three indicators of distress: anxiety, depression,and somatic symptoms. The Symptom Check List-90-R (SCL-90-R;Derogatis, 1977) assesses the extent of psychological symptoms andsomatic health status. To analyze distress, we used the Depression(13 items), Anxiety (10 items), and Somatic Distress (12 items)subscales. Derogatis and Melisaratos (1983) reported test-retest reli-abilities over 2 weeks for the Somatic Distress subscale (.68), Depres-sions subscale (.84), and Anxiety subscale (.79).

Satisfaction. We assessed job satisfaction with two scales. On thefirst scale, respondents indicated how much they had enjoyed theirwork during the past 2 weeks (Quinn & Staines, 1979). The secondscale, the Hoppock Job Satisfaction Scale (McNichols, Stahl, &Manley, 1978), is a 4-item scale that assesses the respondents' satis-faction with their present job and how they feel they compare withothers in job satisfaction. Life satisfaction was assessed on an 8-itemscale adopted from the Life Satisfaction Scale (Warr, Cook, & Wall,1979). The items reflect satisfaction with lifestyle and with personallife.

Work performance. Davidson and Cooper's (1984) 16-item workperformance scale was used as a single indicator of one's perceivedability to carry out the requirements of managerial tasks. The totalscore is derived by summing the responses and dividing by thenumber of applicable items.

Response set. Because measures of sex role attitudes are highlysusceptible to systematic distortion in the direction of norms of sexualequality, we administered an index of cognitive response style todemonstrate discriminant validity of the Attitudes Toward WomenScale. We used the Repression-Sensitization Scale-Short Form (43items; Byrne, 1964) because it has been recommended as the bestmeasure of self-deception (defensiveness) for use in self-report studies(Linden, Paulhus, & Dobson, 1986).

Data Analysis

Structural equation modeling overcomes the problem of identify-ing valid and unconfounded measures of the explanatory constructs,a limitation of previous efforts to test theoretically informed models.Hypothesized structural models are supported if the overall fit of themodel to the observed data is adequate and if the relevant structuralcoefficients between latent variables are statistically significant and inthe predicted direction. However, several alternative models can begenerated, with each providing an adequate fit to the model. Thus, aplausible model depends more on theoretical soundness than statis-tical significance. Furthermore, the data never confirm a model butonly fail to disconfirm it (Cliff, 1983).

We conducted data analyses, using complete data for all 249respondents, with the computer programs PRELIS (Joreskog & Sor-bom, 1988) and LISREL 7 (Joreskog & Sorbom, 1989). PRELIS wasused for prescreening, to calculate data transformations, and togenerate the covariance matrix that was subsequently analyzed byLISREL. Following the recommendations of Joreskog and Sorbom(1989), we analyzed the covariance matrix. Analysis of the correlationmatrix can produce an incorrect chi-square (and other related good-ness-of-fit measures) and give incorrect standard errors. We followedthe two-step approach recommended by Anderson and Gerbing(1988): First the measurement model was confirmed; then the struc-

232 B. LONG, S. KAHN, AND R. SCHUTZ

tural model was tested. We used a maximum likelihood estimationprocedure to test the model. One loading on each latent variable wasset to a value of 1.0 to establish a common metric (J. S. Long, 1983).

A number of the variables were measured on an ordinal scale.Additionally, five of the variables exhibited non-normality (skewnessor kurtosis values >2.0). Consequently, we believed it was desirableto use a method of parameter estimation that was not based onmultivariate normal assumptions (see Bollen, 1989). The LISREL 7method for dealing with non-normal data, the weighted least squaresestimation method, is based on Browne's (1984) asymptotically dis-tribution-free generalized least squares approach. PRELIS generatesthe asymptotic covariance matrix required for weighted least squaresanalysis by LISREL. However, valid asymptotic estimates of vari-ances and covariances require very large sample sizes (approximately1,000 for our model containing 26 observed variables); thus theweighted least squares method was not suitable for the analysis of thisdata set (Joreskog & Sorbom, 1989). With respect to the analysis ofordinal data, the use of polychoric correlations is often recommended(e.g., Bollen, 1989). However, maximum likelihood estimation meth-ods may not yield correct standard errors or chi-square values withsuch a correlation matrix (Joreskog & Sorbom, 1989). Consequently,because our data were a mix of continuous and ordinal measures,this option was not appropriate. Given that neither the weighted leastsquares procedure nor polychoric/polyserial correlations were appro-priate, we applied transformations to those variables exhibiting largekurtosis or positive skewness (>1.5). Square root transformationswere applied to the four variables meeting this condition (daily hassles,SCL-90-R Depression subscale, SCL-90-R Anxiety subscale, andSCL-90-R Somatic Distress subscale), reducing both skewness andkurtosis to less than 1.0 in all four measures. The Attitudes TowardWomen Scale was the only variable remaining with a skewness orkurtosis value greater than 1.5 (skewness = -1.4, kurtosis = 3.2), andthese levels could not be substantially reduced with transformations.However, overall the data appear not to deviate from an assumeddistribution of multivariate normal, as indicated by Mardia's (1985)measure of multivariate kurtosis of 1.02 (p = . 15).

Results

Table 1 presents the means, standard deviations, ranges,and Cronbach's alphas of the measured variables in the initialmodel. As can be seen from the table, the sample was generallycharacterized by a strong sense of personal efficacy and rela-tively egalitarian sex role attitudes. In comparison withDaugherty and Dambrot's (1986) sample of 43 female collegestudents, their mothers, and their grandmothers (Ms = 31.7,29.1, 22.1, respectively; SDs = 6.1, 8.0, 9.0, respectively), thewomen in this study were considerably more egalitarian (M= 38.7, range = 14 to 45). In terms of satisfaction, the womenin this study reported similar levels of life satisfaction assecretaries in Kahn, Long, and Peterson's (1989) study butslightly greater job satisfaction (Ms = 42, 18.1, respectively).With regard to psychosomatic distress, these women reportedgreater anxiety and depression than the norm group of womenbut similar somatic distress (Ms = 3.0, 4.7, and 4.3, respec-tively; Derogatis, Rickels, & Rock, 1976).

The correlation matrix of all variables used for testing themodel is presented in Table 2 (although the covariance matrixwas analyzed, the correlations are presented here for interpre-tation purposes). In addition, correlations were calculatedbetween the Repression-Sensitization Scale and both theAttitudes Toward Women Scale and the Feminist measure(rs = .00 and .12, respectively), indicating that a systematic

Table 1Means, Standard Deviations, and Reliabilities (Cronbach'sAlpha) of Observed Variables in the Initial Model

Observed variable

Age (years)Months in positionYears employedParental status

(no. of children)Marital status8

Job level"Income0

Education11

Attitudes Toward WomenScale (egalitarian)

FeminismInstrumentalityLife Orientation Test

(optimism)General Self-Efficacy ScalePreventive copingRelationship dimension

(work support)Personal growth dimension

(work demands)System maintenance

dimension(work demands)

Daily hasslesStressor type'Appraisals

Threat to self-esteemThreat to goal

attainmentStrain financesLoss of respect

Perceived controlEpisode upsettingEpisode importanceDisengagement copingEngagement copingDepressionAnxietySomatic symptomsLife Satisfaction ScaleJob Satisfaction Scale'Hoppock Job Satisfaction

ScaleWork performance

(satisfaction)

Range———

0-4

0-450-6

20-140

0-3217-1190-18

0-27

0-18

0-180-282

—

5-25

1-51-51-51-51-51-50-570-420-130-100-128-561-10

4-28

1-5

M

38.846.516.1

1.91.43.04.53.1

38.73.3

105.1

23.395.511.1

20.3

8.9

8.132.70.4

10.1

3.51.62.53.13.34.2

13.320.97.54.04.2

41.56.5

20.8

2.2

SD

7.742.77.0

1.20.51.71.51.2

4.91.7

12.5

4.612.33.4

5.4

3.0

3.024.00.5

4.2

1.51.21.51.11.00.87.68.07.64.95.26.92.1

3.6

0.5

a

———

—

—

.81

.66

.88

.84

.86

.73

.87

.62

.80

.93—

.68

—

—

——.82.81.86.82.83.80

.86

.90Note. N = 249. High scores indicate higher levels of the characteristicas defined by the labels. Dashes indicate not applicable.' 1 = married; 2 = not married. b 1 = supervisor, 2 = entry level; 3= middle level; 4 = executive. c 1 = <$15,000; 2 = $15,000-$25,000; 3 = $26,000-40,000; 4 = $41,000-$60,000; 5 = $61,000-$80,000; 6 = $81,000-$ 100,000; 7 = >$ 100,000. d 1 = <Highschool; 2 = high school; 3 = college (2 years postsecondary education);4 = University; 5 = post University. ' 0 = interpersonal stressors; 1= other stressors. ' single-item measure.

response bias with regard to self-deception does not exist inthese data.

Measurement Model

We tested the hypothesized measurement model in a con-firmatory factor analysis to establish valid and unconfounded

CAUSAL MODEL OF MANAGERIAL WOMEN'S STRESS 233

aon

33

<a.UJQ

gw1

113

1

IFe

I

X5

°OO <N

o\ — © >o—• <N <N CNI I I

I I I

o — o o — o

I I I

(N<NCTNt^0*nfNO\0

I I I I I

-H— ( N © — I O O N - CNCN©<NCN(N"J-I I I I I I I I I

infnm-MfSOM-«O. . . . - .I I I I I I I I I I

oo-oo«5-o

t i l l

f S f N t I v ^ ( Soooo-«

© © <NI I I I

)(S««O6OO« — <O O < N © T J - — i t ^ VO ^ t X - « ©© © — o — © o — o © o

I I I I I I I

\ r f ^^ O <N <

> r^i —-• ^" r*^ ^^ O> f^i fO "^ ON t*^i — o o * -> o rs — -^ ^^ '— < •"T7-7TT°--T

J g :

i-ro-ii V

' • a

"a

r-e;•i

234 B. LONG, S. KAHN, AND R. SCHUTZ

indicators of the stress and coping process. We evaluated theoverall fit of this and all subsequent models using a numberof indices: the x

2/df ratio (Q), the LISREL goodness-of-fitindex (GFI), Bollen's (1989) incremental fit index (Z)2), andthe root mean square residual (RMSR). Q values of less than2.0 were interpreted as suggesting a plausible model (Carmines& Mclver, 1981). The LISREL GFI is influenced by samplesize, and thus it is difficult to set evaluative standards for itsinterpretation; however, values above .90 are generally con-sidered good, and values above .85 are considered acceptable.Choosing a suitable alternative to the GFI index is difficultbecause there currently exists considerable debate concerningthe relative merits of a large number of indices (e.g., Bentler,1990; Marsh, Balla, & McDonald, 1988; Mulaik et al., 1989).We chose to use Bollen's D2, an incremental fit index similarto Bentler and Bonett's (1980) normed fit index, because itlessens the effect of sample size and also takes into accountthe model size. Bollen indicated that D2 values above .80 maysuggest a reasonable model, depending on the circumstances.Modifications of the model were considered on the basis of(a) modification indices and t values from the LISREL outputand (b) the correlations between variables; all modificationswere made within the constraints of the theoretical basis ofthe original model.

Initial tests of the proposed measurement model identifieda number of variables that failed to load (<30) on thehypothesized factors (i.e., latent variables). The majority ofthese measures also failed to exhibit a relationship with otherfactors (as indicated by low modification indices). Conse-quently, nine measures were removed from the model: age,months in position, years employed, job level, education,stressor type, threat to self-esteem, strain financial resources,and importance. As a result of these deletions, Demographicswas renamed Status. We made two additional revisions. First,Work Performance was strongly related to measures of Satis-faction. An examination of the scale items suggested thatsatisfaction with work performance was being measured,rather than perceptions of an individual's capability in per-forming her job functions. Thus Work Performance wasincluded as an indicator of the Satisfaction latent variable(with a subsequent loading of -.57). Second, Daily Hasslesshowed large modification indices (MI) with Satisfaction (MI= 16.6) and Distress (MI = 84.5), but because these factorswere measured at a different time (1 month later) it was nottheoretically justifiable to allow Daily Hassles to load on thesefactors. On the basis of this and the fact that the loading ofDaily Hassles on Environment was only .32, Daily Hassleswas removed as an indicator of the latent variable Environ-ment. By removing Daily Hassles, a much better fitting modelwas obtained: the x2/df ratio of the differences between thetwo models was 118.2/22 =*= 5.37, p < .01, indicating that themodel with Daily Hassles removed as an indicator of Envi-ronment is a better model for these data. However, becauseof the hypothesized theoretical importance of Daily Hassles,we decided that it should be retained in the model, but as asingle-item factor in the structural model. The revised meas-urement model, consisting of the seven factors with two ormore indicators, provided an adequate fit to the observeddata. Goodness-of-fit indices were the following: x2/df ratio

of 1.97, x2 (214, N = 249) = 422, GFI = .87, D2 = .86, andRMSR = .08. Additionally, all measurement coefficients ofthe model were significant \t > 2.0) and greater than .30. Thismodel was used in testing the structural model.

Structural Model

Initial attempts to fit the structural model to the sampledata were unsuccessful; thus we undertook a series of modi-fications. Modifications were restricted to those that wereconsistent with previous theory and empirical studies andwere accepted only if the resultant change in the x2/df ratiowas significant (p < .01). Detailed analysis of each subsequentmodified model in this article was not possible, and only thefinal model is presented (detailed information on all modelsis available from Bonita C. Long). The major modificationswere the removal of the direct beta paths (a) from Disengage-ment Coping and Engagement Coping constructs to the out-come, Satisfaction and (b) from Engagement Coping to theoutcome, Distress. These parameter estimates were small andnonsignificant (t < 1.0). Also, the proposed reciprocal rela-tionships between the three pairs of latent variables, Apprais-als-Environment, Disengagement Coping-Engagement Cop-ing, and Distress-Satisfaction were not supported. Minormodifications that led to the final model were the following:The beta path from Environment to Satisfaction (MI = 15.9)was freed, and the theta-delta off-diagonal values (i.e., corre-lating the error terms) between the feminism measure andmarital status and between income and instrumentality (Mis> 10) were freed. The error terms for these two pairs ofvariables were allowed to correlate because there is evidencethat single women are more likely to espouse more egalitariansex role beliefs (Smith & Self, 1981). Furthermore, there isevidence that managers with high instrumental traits are morelikely to be promoted and to earn higher incomes, in com-parison with managers with low instrumental traits (Jacki-nowski, 1987).

Figure 2 depicts the final model. The coefficients reportedare the LISREL "completely standardized" parameter esti-mates (i.e., both the observed and the latent variables havebeen standardized). The fit information for the final model ispresented in Table 3. Although the chi-square value is signif-icant, x2 (278) = 509, p < .001, which indicates that thismodel is not plausible in the population, other fit informationsuggests an acceptable model. The GFI (.87) and Bollen's D2

(.88) are both greater than .85. The Q value (1.97) is below2.0, and the RMSR for the correlations is .074. The coefficientof determination for the structural equations indicates thatapproximately 56% of the total variance in the model wasaccounted for. Finally, the squared multiple correlations forDistress (.49) and Satisfaction (.50) suggest that the causalantecedent and mediating latent variables explain approxi-mately 50% of the variance in the outcome variables.

No further modifications were made. The somewhat lowsquared multiple correlations for the structural equations forEngagement Coping (.17), Environment (. 19), and Daily Has-sles (.22) suggest that further efforts to improve the model fitmay not be entirely justifiable. Future tests of the model must

CAUSAL MODEL OF MANAGERIAL WOMEN'S STRESS 235

Figure 2. Model 4: Final model of managerial women's stress, coping, and well-being. (Significantpaths and standardized LISREL estimates are indicated. STA = Status; MAR = marital status; PAR =parental status; INC = income; SRA = Sex Role Attitudes; AWS = Attitudes Toward Women Scale;FEM = feminism; AGT = Agentic Traits; INS = instrumentality; LOT = Life Orientation Test; GSE= General Self-Efficacy Scale; PRC = preventive coping; ENV = Environment; PGD = personal growthdimension (work demands); SMD = system maintenance dimension (work demands); REL = relation-ship dimension (work support); APP = Appraisals; RES = loss of respect; GOA = threat to goalattainment; UPS = episode upsetting; CON = perceived control; DEN = Disengagement Coping; ENG= Engagement Coping; HAS = Daily Hassles; DIS = Distress; DEP = depression; ANX = anxiety;SOM = somatic symptoms; LSA = Life Satisfaction Scale; SAT = Satisfaction; JSA = Job SatisfactionScale; HJS = Hoppock Job Satisfaction Scale; WPE = work performance.)

focus on improvements of these factors. Therefore, Figure 2is presented as the most plausible of the models tested.

Discussion

The results of the final model tested indicate that Lazarus's(1966; Lazarus & Folkman, 1984) theory of psychologicalstress and coping is useful in understanding managerial wom-en's responses to occupational stress. These results extend theliterature on stress and coping as well as that on careerdevelopment. Because the appraisal of work stress and copingefforts are central to the experience of daily hassles andpsychosomatic health, work stress experienced by women innontraditional careers is an important career issue. Further-more, sex role attitudes and agentic traits, characteristics thatare associated with the development of a nontraditional career

orientation, also predict the appraisal of work stress andcoping efforts.

To focus first on the role of appraisals, the results indicatethat women managers who maintain traditional lifestyles(married with children) and traditional beliefs appraise occu-pational stressors as less threatening. Career theory wouldsuggest that these women are less invested in their careers(Betz & Fitzgerald, 1987). Consequently, they may have lessat stake in stressful events at work when the stressors areperceived to be related to occupational goals or achievements.

The finding that marital and parental status relates toappraisals is consistent with the literature that indicates thatmarriage buffers some women's psychological distress (e.g.,Gore & Mangione, 1983; Repetti et al., 1989). Thoits (1983)suggested that multiple roles can actually enhance well-beingdue to the multiple sources of reward stemming from each,

236 B. LONG, S. KAHN, AND R. SCHUTZ

Table 3Goodness-of-Fit Model Fit and Structural Relations Indicesfor the Causal Model

Index

Overall model fitX2 (278, JV =249)X2ldfLISREL goodness of fitBollen's goodness of fitRoot mean square residual

Structural relationsCoefficient of determination (x variables)Coefficient of determination (equations)Squared multiple correlation

AppraisalEnvironmentDisengagement CopingEngagement CopingDaily HasslesSatisfactionDistress

Value

509.331.83.87.88.074a

.99

.56

.29

.19

.48

.17

.22

.50

.491 Calculated from the analyses of the correlation matrix (all otherindices, calculated from an analysis of the covariance matrix, wereidentical for correlation and covariance matrix analyses).

and Verbrugge and Madans (1985) found that the healthiestwomen have multiple roles: a job, a husband, and oftenchildren. Thus, there is no support for the notion of roleoverload for managerial women with high incomes. It maybe that these women have the financial resources to purchaseservices that reduce role overload.

As Bernard (1981) suggested, there may be a cost to ac-cepting feminist ideology. The relationship between sex roleattitudes and appraisals appears to disadvantage egalitarianmanagers with regard to occupational stress. Egalitarian man-agers may find that their expectations conflict with male-dominated policies and discriminatory practices and that thisresults in appraisals of less control, as well as greater emotionalreactions, in some work situations. Although the path coeffi-cient from sex role attitudes to appraisals is significant, it isnot a very strong relationship, (-.16) perhaps because theattitude scales assessed attitudes toward general sex role ar-rangements, rather than attitudes toward specific sex rolearrangements at work. Future research should examine sexrole attitudes that relate specifically to the work context, andas Fassinger (1990) suggested, future research may redefinetraditional and nontraditional roles, as married career womenwith children increasingly become the norm for the femalepopulation.

Although agentic traits were expected to predict appraisals,agency only indirectly influenced appraisals through the vari-able work environment. For managerial women an unfavor-able work environment (i.e., high demands, low support) andlow self-efficacy traits result in negatively appraised work-related stressor events. Thus, to understand the meaningsattributed to a particular work stressor (i.e., appraisals) andsubsequent coping efforts, it is necessary to understand thework context (both barriers and supports), as well as one'spersonal sense of agency. Career development theory hasfound the concept of agency useful in understanding women'scareer orientation (Betz & Hackett, 1987). It may also be

useful in predicting managerial women's work adjustmentbecause agency was found to relate to perceptions of the workenvironment, to job satisfaction, and most importantly toengagement coping.

A central concept in the Lazarus and Folkman (1984)model is coping efforts. In the final model, positive appraisalsrelate to less frequent use of disengagement coping, which inturn influences engagement coping. These results are consist-ent with the findings of Folkman, Lazarus, Dunkel-Schetter,DeLongis, and Gruen (1986) that planful problem solvingand emotional self-control strategies were most often usedwhen the stressor involved a work goal. Lazarus and Folkman(1984) concluded that managing emotional responses may bea necessary precursor to dealing with a problem more directly.Thus, finding ways to disengage (e.g., through emotional self-control) may be as important as task accomplishment formanagerial women coping with occupational stress.

Engagement coping is also influenced by agentic traits, thatis, women who have stronger agentic traits are likely to usemore engagement coping. It is interesting to note that agencydirectly influences engagement coping but not disengagementcoping. Thus, women with high confidence in their generalcoping skills are likely to persist in engagement coping efforts(Bandura, 1986). However, persistence in the face of anunresolved stressor event, may lead to increased subjectivedistress (e.g., daily hassles) because of longer exposure to thestressor. Given that greater agency predicts greater engage-ment coping, which in turn predicts more severe hassles, thereis some evidence that increased distress may occur in high-efficacy women who persist in coping. Litt (1988) and Ban-dura, O'Leary, Taylor, Gauthier, and Gossard (1987) pro-vided evidence that increased distress does in fact occur inhigh-efficacy individuals who persist.

To clarify these relationships, we reanalyzed the finalmodel, with coping entered as a single variable scored as arelative score (Vitaliano, Maiuro, Russo, & Becker, 1987). Inthis analysis, the path coefficients from relative engagementcoping to daily hassles and to distress only approach signifi-cance (standardized coefficient from engagement coping todaily hassles = - .11 and from engagement coping to distress,—.08); the greater the ratio of engagement to total coping, theless severe the daily hassles. Thus, when relative engagementcoping is used, the relationships among situation-specific cop-ing responses, concurrent hassles, and psychological distressexperienced 1 month later are weakened. Furthermore, thepath coefficient from appraisals to relative engagement copingchanged in both magnitude and direction (from -.70 to .35).Thus, more positive appraisals are related to more relativeengagement coping. Although as seen earlier the total amountof engagement coping is correlated with more severe dailyhassles, the use of engagement strategies in combination withsome disengagement techniques is related to less severe has-sles. This implies that the way in which one copes with stressmay facilitate positive adaptation, whereas how much onecopes may reflect one's level of distress (Forsythe & Compas,1987).

Although appraisals and coping strategies are related (eitherdirectly or indirectly) to the outcomes distress and satisfactionmeasured 1 month later, egalitarian attitudes, perceptions of

CAUSAL MODEL OF MANAGERIAL WOMEN'S STRESS 237

demanding but nonsupportive work environments, and lessefficacy are associated with greater general dissatisfaction.Greater dissatisfaction has been postulated to result fromunfulfilled needs and discrepancies between demands and thebelief system of the individual (Cox, 1978). Furthermore, ifmanagerial women are dissatisfied, they are also likely toexperience psychosomatic distress (poor health). These resultsare similar to those of other studies that have found that workenvironments (Karasek et al., 1981), as well as agentic traits,are related to various indicators of well-being (B. C. Long,1989). In future studies, researchers might determine whetheregalitarian women's career orientation changes over time withcontinuing job and life dissatisfaction.

The problems of self-report methodology and conceptualoverlap between independent and dependent variables wassubstantially lessened with the design of our study (withassessments 1 month apart). However, caution is warrantedin inferring causal relations from the data, given that, in theabsence of experimental control, all correlation-based studiesare vulnerable to alternative causal explanations and influ-ences of unmeasured variables. The efforts to keep measuresas short and concise as possible led to a number of variableswith poor psychometric properties, measures that eventuallyhad to be dropped from the model being tested (e.g., some ofthe appraisal indicators). It is difficult to determine whetherthe disappearance of these variables is due to psychometricinadequacy or lack of significance of the constructs in thisparticular sample.

A modification to the theoretical model involving dailyhassles also warrants discussion. The difficulty of determiningwhere hassles should be placed in Lazarus's (1984) theoreticalmodel has been addressed by Dohrenwend and Shrout (1985).They argued that hassles are determined by environmentalcircumstances and reactions to them by personal dispositions.Lazarus (1984), however, shifted from a stimulus-centeredview of hassles, with hassles as antecedents of appraisal andcoping, to hassles as a consequence of appraisal. Lazarus andFolkman (1984) postulated that "whereas daily hassles wouldbe normally considered to fuel the coping process, they arealso an outcome of coping" (p. 313). The relation postulatedby Dohrenwend and Shrout (1985) and Lazarus is supportedin this analysis in that personal resources (e.g., agentic traits)were negatively related to daily hassles (-.34), but engagementcoping strategies (.18) and disengagement coping strategies(.27) were positively related to daily hassles. Furthermore, theeffects of appraisals and both engagement coping and disen-gagement coping on the outcomes are mediated by dailyhassles. Although disengagement coping has a weak but sig-nificant direct effect on distress (.19), the direct effect of dailyhassles is strong (.47), and the indirect effect of disengagementcoping on distress mediated by daily hassles (.13) is alsosignificant. This suggests that, for a managerial woman, thereis a strong relationship between everyday irritants and well-being 1 month later.

The sample came from a normal population and consistedpredominantly of White female managers from a large met-ropolitan community. Thus, the results may not be general-izable to female managers of other geographic settings, organ-izational levels, or clinical populations. Replications with

more refined instruments and larger, more diverse popula-tions are needed, as such replications would aid in discrimi-nating between the theoretical and psychometric weaknessesof our study. Finally, it must be remembered that the resultsof this investigation only failed to disconfirm the model; allconclusions must be approached tentatively.

Despite the limitations to our study, this investigationprovides some promising leads for counseling interventionsfor managerial women and informs models of career theory.From the complexity of the model, it is apparent that thereare many intervention options. The most obvious is to de-velop more effective coping efforts. Individualized cognitive-behavioral interventions that facilitate the development ofadaptive appraisals and coping strategies (Matheny et al.,1986; Meichenbaum, 1977) would be a useful component ofa stress-management program for these women. However,particular attention should be paid to disengagement copingstrategies. Although both types of coping are useful, depend-ence on disengagement coping may lead to greater distress.Furthermore, because a strong sense of agency may result inexcessive persistence in coping in situations that cannot bechanged, problem-solving strategies that clearly assess one'soptions for change might be useful. Initial assessments of sexrole attitudes and agentic traits will provide information usefulto the counselor and the client in revealing potential frustra-tions and difficulties in overcoming the stressors of a nontra-ditional career. The results also suggest that the work contextshould not be ignored as a major contributor to well-being.Thus, the development of mentoring systems and other meth-ods that facilitate supportive work environments (e.g., throughassessments of the work climate) would reduce stress anddissatisfaction.

Sex role attitudes, marital and parental roles, and agentictraits have been associated with women's choice and com-mitment to nontraditional occupations. However, these char-acteristics have both negative and positive components. Thepresent study suggests that profeminist attitudes toward wom-en's roles may not always have facilitative effects on women'scareer development. Egalitarian sex role attitudes lead to moredistressing stressor appraisals (as does remaining single andchildless), as well as greater job dissatisfaction, possibly be-cause managers with such attitudes have more at stake inparticular types of work-related stressor events. However, anexamination of these attitudes may facilitate a client's con-frontation with sex role social barriers (Betz & Fitzgerald,1987). Thus, interventions addressing occupational stress andcareer adjustment needs of managerial women should includeconsideration of nonoccupational roles, internal psychologicalfactors, and structural components of the work environment,as well as the social forces that stem from sex role stereotypesand discrimination.

References

Anderson, J. C, & Gerbing, D. W. (1988). Structural equationmodeling in practice: A review and recommended two-step ap-proach. Psychological Bulletin, 103, 411-423.

Bandura, A. (1986). Social foundation of thought and action: A socialcognitive theory. Englewood Cliffs, NJ: Prentice-Hall.

Bandura, A., O'Leary, A., Taylor, C. B., Gauthier, J., & Gossard, D.

238 B. LONG, S. KAHN, AND R. SCHUTZ

(1987). Perceived self-efficacy and pain control: Opioid and nono-pioid mediators. Journal of Personality and Social Psychology, 53,563-571.

Bern, S. L. (1981). Bern Sex Role Inventory: A professional manual.Palo Alto, CA: Counsulting Psychologists Press.

Bentler, P. M. (1990). Comparative fit indexes in structural models.Psychological Bulletin, 107, 238-246.

Bentler, P. M., & Bonett, D. G. (1980). Significance tests and good-ness-of-fit in the analysis of covariance structures. PsychologicalBulletin, 88, 588-606.

Bernard, J. (1981). The female world New York: Free Press.Betz, N. E., & Fitzgerald, L. F. (1987). The career psychology of

women. San Diego, CA: Academic Press.Betz, N. E., & Hackett, G. (1987). Concept of agency in educational

and career development. Journal of Counseling Psychology, 34,299-308.

Billings, A. G., & Moos, R. H. (1981). The role of coping responsesand social resources in attenuating the stress of life events. Journalof Behavioral Medicine, 4, 139-157.

Bollen, K. A. (1989). Structural equations with latent variables. NewYork: Wiley.

Browne, M. W. (1984). Asymptotic distribution free methods inanalysis of covariance structures. British Journal of Mathematicaland Statistical Psychology, 37, 62-83.

Bruning, N. S., & Frew, D. R. (1987). Effects of exercise, relaxation,and management skills training on physiological stress indicators:A field experiment. Journal of Applied Psychology, 72, 512-521.

Byrne, D. (1964). Repression-sensitization as a dimension of person-ality. In B. A. Maher (Ed.), Progress in experimental personalityresearch (Vol. 1; pp. 169-220). San Diego, CA: Academic Press.

Carmines, E., & Mclver, J. (1981). Analyzing models with unobservedvariables: Analysis of covariance structures. In G. Bohrnstedt & E.Borgatta (Eds.), Social measurement: Current issues (pp. 65-115).Beverly Hills: Sage.

Cliff, N. (1983). Some cautions concerning the applications of causalmethods. Multivariate Behavioral Research, 18, 115-126.

Cohen, S., & Wills, T. (1985). Stress, social support, and the bufferinghypothesis. Psychological Bulletin, 98, 310-377.

Cox, T. (1978). Stress. London: Macmillan.Daugherty, C. G., & Dambrot, F. H. (1986). Reliability of the

Attitudes Toward Women Scale. Educational and PsychologicalMeasurement, 46, 449-453.

Davidson, M., & Cooper, C. L. (1984). Occupational stress in femalemanagers: A comparative study. Journal of Management Studies,21, 185-205.

DeLongis, A., Coyne, J. G, Dakof, G., Folkman, S., & Lazarus, R.S. (1982). Relationship of daily hassles, uplifts, and major lifeevents to health status. Health Psychology, 1, 119-136.

Derogatis, L. R. (1977). SCL-90: Administration, scoring and proce-dure manual-I. Baltimore: Johns Hopkins University Press.

Derogatis, L. R., & Melisaratos, N. (1983). The Brief SymptomInventory: An introductory report. Psychological Medicine, 13,595-605.

Derogatis, R., Rickels, K., & Rock, A. (1976). The SCL-90 and theMMPI: A step in the validation of a new self-report scale. BritishJournal of Psychiatry, 128, 280-289.

Dohrenwend, B. P., & Shrout, P. E. (1985). "Hassles" in the concep-tualization and measurement of life stress variables. AmericanPsychologist, 40, 780-785.

Eisler, R. M., Skidmore, J. R., & Ward, C. H. (1988). Masculinegender-role stress: Predictor of anger, anxiety and health-risk be-haviors. Journal of Personality Assessment, 52, 133-141.

Fassinger, R. E. (1985). A causal model of college women's careerchoice. Journal of Vocational Behavior, 27, 123-153.

Fassinger, R. E. (1987). Use of structural equation modeling in

counseling psychology research. Journal of Counseling Psychology,34, 425-436.

Fassinger, R. E. (1990). Causal models of career choice in two samplesof college women. Journal of Vocational Behavior, 36, 225-248.

Feather, N. T. (1985). Masculinity, femininity, self-esteem, and sub-clinical depression. Sex Roles, 12, 491-500.

Folkman, S., & Lazarus, R. S. (1980). An analysis of coping inmiddle-aged community samples. Journal of Health and SocialBehavior, 27,219-239.

Folkman, S., & Lazarus, R. S. (1985). If it changes it must be aprocess: Study of emotion and coping during three stages of acollege examination. Journal of Personality and Social Psychology,48, 150-170.

Folkman, S., & Lazarus, R. S. (1988). Coping as a mediator ofemotion. Journal of Personality and Social Psychology, 54, 466-475.

Folkman, S., Lazarus, R. S., Dunkel-Schetter, C, DeLongis, A., &Gruen, R. (1986). The dynamics of a stressful encounter: Cognitiveappraisal, coping, and encounter outcomes. Journal of Personalityand Social Psychology, 50, 992-1003.

Folkman, S., Lazarus, R. S., Gruen, R., & DeLongis, A. (1986).Appraisal, coping, health status, and psychological symptoms. Jour-nal of Personality and Social Psychology, 50, 571-579.

Forsythe, C, & Compas, B. E. (1987). Interaction of cognitive ap-praisals of stressful events and coping: Testing the goodness of fithypothesis. Cognitive Therapy and Research, 11, 473-485.

Gore, S., & Mangione, T. W. (1983). Social roles, sex roles andpsychological distress: Additive and interactive models of sex dif-ferences. Journal of Health and Social Behavior, 24, 300-312.

Heppner, P. P., Cook, S. W., Strozier, A. L., & Heppner, M. J. (1991).An investigation of coping styles and gender differences with farm-ers in career transition. Journal of Counseling Psychology, 38, 167-174.

Hock, E., Morgan, K., & Hock, M. (1985). Employment decisionsmade by mothers of infants. Psychology of Women Quarterly, 9,383-402.

Holahan, C, & Holahan, C. J. (1987). Self-efficacy, social support,and depression in aging: A longitudinal analysis. Journal of Ger-ontology, 42(1), 65-68.

Jackinowski, C. M. (1987). Androgyny in a male-dominated field:The relationship of sex-typed traits to performance and satisfactionin engineering. Sex Roles, 17, 529-547.

Joreskog, K. G., & Sorbom, D. (1988). PRELIS: A preprocessor forLISREL (2nd ed.), Mooresville, IN: Scientific Software.

Joreskog, K. G., & Sorbom, D. (1989). LISREL 7: User's referenceguide. Mooresville, IN: Scientific Software.

Kahn, S. E., Long, B. C, & Peterson, C. (1989). Marital and parentalstatus and quality of work life of female clerical workers. CanadianJournal of Counselling, 23, 174-183.

Kanner, A., Coyne, J., Schaefer, C, & Lazarus, R. (1981). Compari-son of two modes of stress measurement: Daily hassles and upliftsversus major life events. Journal of Behavioral Medicine, 4, 1-39.

Karasek, R. A., Baker, D., Marxer, F., Ahlbom, & Theorell, T. (1981).Job decision latitude, job demands, and cardiovascular disease: Aprospective study of Swedish men. American Journal of PublicHealth, 71, 694-705.

Kessler, R. (1983). Methodological issues in the study of psychosocialstress. In H. B. Kaplan (Ed.), Psychosocial stress (pp. 267-341).San Diego, CA: Academic.

Latack, J. C. (1989). Work, stress, and careers: A preventive approachto maintaining organizational health. In M. B. Arthurs, D. T. Hall,& B. S. Lawrence (Eds.), A handbook of career theory (pp. 252-275). London: Cambridge University Press.

Lazarus, R. S. (1966). Psychological stress and the coping procedure.New York: McGraw-Hill.

CAUSAL MODEL OF MANAGERIAL WOMEN'S STRESS 239

Lazarus, R. S. (1981, July). Little hassles can be hazardous to health.Psychology Today, pp. 58-62.

Lazarus, R. S. (1984). Puzzles in the study of daily hassles. Journalof Behavioral Medicine, 7, 375-389.

Lazarus, R. S., & Folkman, S. (1984). Stress, coping, and appraisal.New York: Springer.

Linden, W., Paulhus, D. L., & Dobson, K. S. (1986). Effects ofresponse styles on the report of psychological and somatic distress.Journal of Consulting and Clinical Psychology, 54, 309-313.

Litt, M. D. (1988). Self-efficacy and perceived control: Cognitivemediators of pain tolerance. Journal of Personality and SocialPsychology, 54, 149-160.

Long, B. C. (1988). Work-related stress and coping strategies ofprofessional women. Journal of Employment Counseling, 25, 37-44.

Long, B. C. (1989). Sex role orientation, coping strategies, and self-efficacy of women in traditional and nontraditional occupations.Psychology of Women Quarterly, 13, 307-324.

Long, B. C. (1990). Relation between coping strategies, sex-typedtraits, and environmental characteristics: A comparison of maleand female managers. Journal of Counseling Psychology, 37, 185-194.

Long, B. C, & Haney, C. J. (1988). Coping strategies for workingwomen: Aerobic exercise and relaxation interventions. BehaviorTherapy, 19, 75-83.

Long, J. S. (1983). Confirmatory factor analysis: A preface to LISREL.Beverly Hills, CA: Sage.

Mardia, K. V. (1985). Mardia's test of multinormality. In S. Kotz &N. L. Johnson (Eds.), Encyclopedia of statistical sciences (Vol. 5,pp. 217-221). New York: Wiley.

Marsh, H. W., Balla, J. R., & McDonald, R. P. (1988). Goodness-of-fit indices in confirmatory factor analysis: The effect of sample size.Psychological Bulletin, 103, 391-410.

Matheny, K. B., Aycock, D. W., Pugh, J. L., Curlette, W. L., & SilverCannella, K. A. (1986). Stress coping: A qualitative and quantitativesynthesis with implications for treatment. The Counseling Psychol-ogist, 14, 499-549.

McNichols, C, Stahl, M., & Manley, T. (1978). A validation ofHoppock's job satisfaction measure. Academy of ManagementJournal, 21, 737-742.

Meichenbaum, D. (1977). Cognitive behavioral modification: Anintegrative approach. New York: Plenum Press.

Moos, R. (1981). Work Environment Scale Manual. Palo Alto, CA:Consulting Psychologists Press.

Mulaik, S. A., James, L. R., Van Alstine, J., Bennett, N., Lind, S., &Stilwell, C. D. (1989). Evaluation of goodness-of-fit indices forstructural equation models. Psychological Bulletin, 105, 430-445.

Nezu, A. M., & Nezu, C. M. (1987). Psychological distress, problemsolving, and coping reactions: Sex role differences. Sex Roles, 16,205-214.

Parasuraman, S., & Cleek, M. A. (1984). Coping behaviors andmanager's affective reactions to role stressors. Journal of VocationalBehavior, 24, 179-193.

Parkes, K. R. (1986). Coping in stressful episodes: The role ofindividual differences, environmental factors, and situational char-acteristics. Journal of Personality and Social Psychology, 51, 1277—1292.

Pearlin, L. I., & Schooler, C. (1978). The structure of coping. Journalof Health and Social Behavior, 19, 2-21.

Quinn, R. P., & Staines, G. L. (1979). The 1977 Quality of Employ-ment Survey: Descriptive statistics with comparison data from the1969-1970 and 1972-1973 survey. Ann Arbor, MI: Institute forSocial Research.

Repetti, R. L., Matthews, K. A., & Waldron, I. (1989). Employmentand women's health: Effects of paid employment on women'smental and physical health. American Psychologist, 44, 1394-1401.

Rook, K. S. (1984). The negative side of social interaction: Impacton psychological well being. Journal of Personality and SocialPsychology, 46, 1097-1108.

Scheier, M. F., & Carver, C. S. (1985). Optimism, coping, and health:Assessment and implications of generalized outcome expectancies.Health Psychology, 4, 219-247.

Scheier, M. F., Weintraub, J. K., & Carver, C. S. (1986). Coping withstress: Divergent strategies of optimists and pessimists. Journal ofPersonality and Social Psychology, 51, 1257-1264.

Sherer, M., Maddux, J. E., Mercandante, G., Prentice-Dunn, S.,Jacobs, B., & Rogers, R. W. (1982). The Self-Efficacy Scale: Con-struction and validation. Psychological Reports, 51, 663-671.

Smith, M. D., & Self, G. D. (1981). Feminists and traditionalists: Anattitudinal comparison. Sex Roles, 7, 183-189.

Spence, J., & Helmreich, R. (1978). Masculinity and femininity: Theirpsychological dimensions, correlates and antecedents. Austin, TX:University of Texas Press.

Terborg, J. R. (1985). Working women and stress. In T. A. Beehr &R. S. Bhagat (Eds.), Human stress and cognition in organizations(pp. 245-286). New York: Wiley.

Thoits, P. A. (1983). Multiple identities and psychological well-being.American Sociological Review, 48, 174-187.

Verbrugge, L. M., & Madans, J. H. (1985). Social roles and healthtrends of American women. Milbank Memorial Fund Quarterly/Health and Society, 63, 691-735.

Vitaliano, P. P., Maiuro, R. D., Russo, J., & Becker, J. (1987). Rawversus relative scores in the assessment of coping strategies. Journalof Behavioral Medicine, 10, 1-18.

Warr, P., Cook, J., & Wall, T. (1979). Scales for the measurement ofsome work attitudes and aspects of psychological well-being. Jour-nal of Occupational Psychology, 52, 129-148.

Wong, P. R., & Reker, G. T. (1983). Validity of the Coping Inventory.Paper presented at the Canadian Association on Gerontology,Moncton, New Brunswick, Canada.

Received March 11, 1991Revision received August 16, 1991

Accepted August 26, 1991