Occupational stress and gender: a cross-cultural study

Occupat ional stress and gender:a cross-cul tural study

Karen Miller*,{,1, Mike Greyling1, Cary Cooper2, Luo Lu3, Kate Sparks2

and Paul E. Spector4

1University of the Witwatersand, South Africa2University of Manchester Institute of Science and Technology, U.K.3Graduate Institute of Behavioural Sciences, Taiwan4University of South Florida, U.S.

SummaryThe aim of this study was to examine the interaction of gender and culture in managers' experiences ofwork stress. Data were collected on sources of occupational stress (stressors), coping and consequencesof occupational stress (strains) from male and female managers from four countries Ð South Africa,the United Kingdom, United States of America and Taiwan. Few signi®cant results were found for theinteraction between country and gender on any of the measures. When the sample as a whole wasexamined, however, there were also virtually no differences in sources of work stress, but there weredifferences in the consequences of work stress for male and female managers. The implications of®nding a lack of differences in sources of work stress for males and females combined with ®ndingdifferences in strains for male and female managers are discussed. Copyright # 2000 John Wiley &Sons, Ltd.

Key Words

stress; work stress; gender and stress

The entry of women into managerial and profes-sional organisational roles over the last three tofour decades has been accompanied by a sub-stantial amount of research which has examinedthe impact of this change on the organisation, thewoman and her family.1 In particular, the issue ofwomen's work-related stress has received atten-tion, beginning in the 1980s when the effects ofthe increase in the numbers of women in non-traditional/senior organisational levels in theUnited Kingdom (UK) and the United States ofAmerica (USA) began to be felt. However, the

results of the research in this area tended tobe inconclusive. For example, Jick and Mitz2

reviewed the empirical evidence for sex differ-ences in work stress from 19 studies and foundthat women tend to report higher rates ofpsychological distress than men, but that menare more prone to more severe physical illness.Further, a review conducted by Nelson andQuick3 indicated that employed women experi-ence greater stress than both non-employedwomen and men because of several uniquestressors faced by employed women. Baruch,Biener and Barnett4 on the other hand, foundthat non-employed women experience greaterstress than employed women, while Martocchioand O'Leary5 who undertook a meta-analysis of15 studies that examined sex differences inoccupational stress, found no differences inexperienced and perceived work stress. On thebasis of their ®ndings, Martocchio and O'Leary5

S t r e s s M e d i c i n eStress Med. 16: 271±278 (2000)

Copyright # 2000 John Wiley & Sons, Ltd. Received 15 June 1999Accepted 24 November 1999

*Correspondence to: Karen Miller, Department ofPsychology, University of the Witwatersrand, PrivateBag 3, WITS 2050, South Africa. Fax: 27 11 7162476.{E-mail: [email protected]

argued that `this analysis does not representirrefutable evidence that there are no differencesin occupational stress. The burden of proof does,however, now lie with those researchers whosuggest that sex differences exist. Future research-ers should not spend time debating the nature ofthe sex occupational stress relationship when theexistence of the relationship is now the salientissue' (p. 500). Subsequent to this assertion, thereseem to have been fewer research studies aimedspeci®cally at comparing men and women'swork-related stress.

The results of the studies that have beendone more recently (e.g. Guppy and Rick6 andNarayanan, Menon and Spector7) have been ascontradictory as the studies of the 1980s. More-over, the emphasis in the literature on gender andstress seems to have shifted towards the develop-ment of theoretical perspectives relating towomen and stress (e.g. Greenglass8; Matuszek,Nelson and Quick9). This may be because theentry of women into `non traditional'/seniororganisational positions in developed countriessuch as the USA and the UK (from where much ofthis type of research has traditionally originated)became less topical. Women's entry into theabove mentioned positions in these countries,while still not equal to that of men, has become afait accompli, with non-discrimination againstwomen in a country like America being enforcedthrough equal opportunity legislation.

In developing countries, however, the entry ofwomen into managerial or senior managerialpositions has been slower, with women only nowbeginning to take up roles in these positions insigni®cant numbers. In South Africa, for example,gender discrimination in the workplace has beenoutlawed only recently with the passing of theLabour Relations Act in 1995 and the Employ-ment Equity Act last year. Moreover, even withthis legislation in place, issues related to racialdiscrimination in the workplace far outweighgender discrimination in South Africa ( forobvious reasons, given South Africa's recent,apartheid history). Similarly, in Taiwan there isanti-discrimination legislation, but it is of fairlyrecent origin; and while labor laws exist whichprovide for such things as maternity leave,employers do not always grant it. Women havealso complained of being forced to quit jobsbecause of age or child-bearing restrictions, and

restrictive quotas reputedly exist within certainoccupations.10 Women often receive less frequentpromotions and lower salaries than their malecounterparts.10

Given this context, it seems likely that genderdifferences in the work environment in develop-ing countries may still be an issue. Thus, thepresent study, which includes a sample of menand women managers from four differentcountries of differing levels of development,namely the UK, USA, Taiwan and South Africaprovides a unique opportunity to revisit some ofthe issues in gender and occupational stress froma cross cultural perspective.

The research questions

Based on the above discussion, a number ofresearch questions for this paper were developed.The ®rst set of questions are the same as many ofthose posed, but not satisfactorily answered inthe research of the 1980s. These are:

1. Do men and women differ in the sources ofwork stress (stressors) they experience?

2. Do men and women differ in the mechanismsthey use to cope with work stress?

3. Do men and women differ in their experiencesof the consequences of work stress (strains)?

Further questions, taking the cross-culturalnature of the study into account were thendeveloped as follows:

1. Is there an interaction between country andgender in managers' experiences of work stress(stressors)?

2. Is there an interaction between country andgender in the mechanisms managers use tocope with stress?

3. Is there an interaction between country andgender in managers' experiences of the con-sequences of work stress (strains)?

Method

In order to answer the above questions, data fromthe Collaborative International Study of Manage-rial Stress (CISMS) from four countries wasobtained. The CISMS project was established in

K. Miller et al.

272 Copyright # 2000 John Wiley & Sons, Ltd. Stress Med. 16: 271±278 (2000)

1996 as an effort among researchers in differentcountries to conduct cross-national research onstress. As mentioned above, data from fourcountries in the CISMS study were used for thepresent research, namely, Taiwan, the UK, SouthAfrica and the USA. The CISMS study includes24 countries, but not all the countries' data wereavailable at the time of doing this study. Thereasons for choosing Taiwan, South Africa, theUSA and the UK was to get data from developedcountries (the USA and UK), a developingcountry (Taiwan) and an underdeveloped country(South Africa). More details of the sample arepresented in Table I.

Sample

The sample comprised 822 managers from fourcountries. Subjects for the study were recruited bythe individual researchers in each country. Theresearchers were asked to try and obtain repre-sentative samples of managers for each countryand not to focus only on one organisation.However, other than this request, the recruitmentof the subjects was left entirely to the discretion ofeach researcher. As a result, the sample sizesvaried quite widely. The biographical details ofthe sample for each country are presented inTable I. What is important here is that, althoughthe numbers of male and female managers fromeach country are not equal, there are suf®cientnumbers of males and females in each sub-sampleto allow for meaningful statistical analyses. Themean age ranged from 36.2 (South Africa) to 43.9(the US) and the mean tenure ranged from 8.1(South Africa) to 15.5 (the UK). The majority ofparticipants from all the countries had attendedsome college, with the lowest proportion (64per cent) from the UK and the highest fromTaiwan (100 per cent). More than two thirds ofthe sample in each country were married.

Measures

The samples from all the 22 countries inthe original CISMS study were administeredthe Occupational Stress Indicator-2 (OSI-2), a90 item scale that assesses job strains, sources ofstress, personality and coping;11 the Work Locusof Control Scale;12 the Hofstede13 Values SurveyModule 1994 Questionnaire and 22 additionalquestions, including demographics. Only theOSI-2 was used for the purposes of the presentpaper and therefore it is the only scale that will bediscussed further.

The OSI-2 is a 90 item short form of the OSI.14

The Sources of Work Pressure (work stressors),Coping, Physical Strain and Mental Strainmeasures were relevant for the present study.Details of the sub-scales are given in Table II,including the estimated reliabilities for the currentsample. Using the cut-off of 0.7015 as an indica-tion of minimal acceptable reliability, three sub-scales did not meet this criterion. Therefore, onlythe composite scales for these variables were usedin the current study.

Results

The data was subjected to the statistical analysistechniques of ANOVA and the two independentsample t-test. First, t-tests were conducted toassess the differences between males and femaleson the above variables for the sample as a whole.Then a two way ANOVA was conducted to assessthe interaction between country and gender onthe variables under consideration.

The results of the t-tests for the sample as awhole are presented in Table III. As can be seenfrom this table, signi®cant differences emerged onfour of the variables. The ®rst is stress fromorganisational climate, a sub-scale of the sources

Table I. Biographical details of the sample. N � 822.

Country Males Females Mean age Meantenure

% college % married

N % N %

Taiwan 191 40.6 151 43.0 37.9 10.2 100 74UK 117 24.8 106 30.2 43.7 15.5 64 83SA 98 20.8 42 12.0 36.2 8.1 98 71USA 65 13.8 52 14.8 43.9 9.5 89 71

Total 471 100 351 100

Occupational stress and gender

Copyright # 2000 John Wiley & Sons, Ltd. Stress Med. 16: 271±278 (2000) 273

of job pressure measure. For this sub-scale, malesscored higher than females. There were alsosigni®cant differences on both the total copingscore and the coping using social support scale,with women being higher on coping using socialsupport and men higher on total coping. Finally,males scored signi®cantly higher on both physicaland mental well-being than women, indicatingbetter well-being among men.

As far as the interaction of gender and countryon differences in work stress are concerned,

signi®cant results were found for four variables,namely, stress from managerial role, copingstrategies, mental well-being and physical well-being. A summary of these results is given inTable IV.

In order to examine these interactions morecarefully, post hoc analyses were conductedon those variables that showed signi®cantinteractions. T-tests by country were conductedfor each of these variables, the results of which arepresented in Table V. Straight t-tests were used

Table II. A description of the OSI-2, including reliabilities estimated from the current sample. N � 822.

Variable Number of items Alpha Higher score indicates:

Sources of work stress 40 0.946 Higher stressWorkload 6 0.820Relationships 8 0.887Home/work balance 6 0.819Recognition 4 0.736Organisational climate 4 0.843Managerial role 4 0.586Personal responsibility 4 0.802Hassles at work 4 0.647

Coping strategies 10 0.783 Greater use of coping strategiesControl 6 0.776Support 4 0.514

Mental well-being 12 0.652 Better mental well beingContentment 5 0.800Peace of mind 3 0.555

Physical well-being 6 0.842 Better physical well-beingCalmness 3 0.903Energy 3 0.696

Table III. Results of the t-tests for the whole sample.

Variable Means df t

Males Females

Sources of pressureWorkload 21.822 22.322 805 ÿ1.1431Relationships 28.869 29.724 803 ÿ1.5316Home/work balance 18.558 18.539 790 0.0381Managerial role 13.167 13.158 807 0.0363Personal responsibility 14.575 15.117 802 ÿ1.8293Hassles 13.497 13.886 807 ÿ1.5416Recognition 14.231 13.939 804 0.9370Organisational climate 14.755 15.405 807 2.6072**

Coping strategies 40.981 42.170 810 ÿ2.8551**Mental well-being 31.144 29.578 807 3.1119**Physical well-being 26.201 25.104 812 2.6546**

*p5 0.05.**p50.01.***p50.001.

K. Miller et al.


rather than the usual least signi®cant differenceapproach to post hoc analyses for two reasons.First, the research question implies that the genderdifferences across the countries were of speci®cinterest. Thus, only four comparisons wereexamined, viz. males and females were comparedfor each country. Second, the t-tests wereperformed using a variance estimate based onlyon the country being compared rather than themean square error as is usually the case for posthoc comparisons. This allowed the procedure to

be more robust in the face of possible departuresfrom the variance homogeneity assumption.

From Table V, it is clear that as far as mentalwell-being is concerned, American males scoredhighest. South African males, British males andAmerican females were all fairly similar, followedfairly closely by Taiwanese females and thenmales, with British and South African femalesobtaining the lowest scores on this variable. Withregard to physical well-being, American malesagain scored highest followed by British males andAmerican females. British females came next,followed by Taiwanese males, South Africanmales and Taiwanese females, with South Africanfemales again scoring lowest on this scale. Withregard to coping, Taiwanese females scoredhighest, followed by Taiwanese males andAmerican and British females. In this case, SouthAfrican females, males and American malesrespectively were all fairly similar, while UK malesobtained the lowest scores. Finally, results relatingto the one stressor that emerged as signi®cant inthe ANOVA analyses Ð stress from managerialrole Ð did not show signi®cant differences for theindividual country t-tests. The large sample sizecoupled with a relatively small f statistic indicatesa small and probably unimportant effect.

Table IV. Summary of the results for the interactionbetween gender and country.

Variable df F

Sources of pressureWorkload 3,799 0.38Relationships 3,797 1.27Home/work balance 3,784 0.71Managerial role 3,801 2.88*Personal responsibility 3,796 0.20Hassles 3,801 0.66Recognition 3,798 1.03Organisational climate 3,801 1.26

Coping strategies 3,804 2.62*Mental well-being 3,801 3.43*Physical well-being 3,806 2.74*

Table V. Results of the t-tests by country.

Country Means df t

Males Females

Mental well-beingTaiwan 10.367 10.757 334 ÿ1.203UK 10.800 9.962 219 2.0712*SA 10.938 9.167 137 3.0926*USA 11.585 10.706 114 1.5388

Physical well-beingTaiwan 25.300 24.851 336 0.7458UK 27.374 25.972 219 1.9258SA 25.196 21.262 137 3.3856***USA 28.262 27.196 114 0.9527

Pressure from managerial roleTaiwan 14.368 15.203 331 ÿ2.4897UK 11.843 11.245 219 1.3572SA 12.711 12.143 137 0.8001USA 12.677 12.235 114 0.7730

Coping strategiesTaiwan 43.356 43.818 334 ÿ0.8103UK 38.287 40.783 219 ÿ4.1916***SA 39.948 39.452 137 0.3557USA 10.415 42.510 114 ÿ1.9049



Discussion

As far as the ®rst set of research questions areconcerned, the ®ndings indicate that for thewhole sample (i.e. all four countries) there werevirtually no differences in the stressors experi-enced by men and women (with the soleexception of stress from organisational climate).There were differences, however, in the strainsexperienced by men and women, speci®callymental well-being and physical well-being, withmen exhibiting better mental and physical well-being than women. This supports the conclusionof Matuszek, Nelson and Quick9 that men andwomen operating at roughly the same level withinorganisations generally experience the samestressors, but experience different levels ofdistress as a result. These ®ndings must, however,be viewed with caution as they come from thecombining of four data sets, each of which differsigni®cantly on the variables under consideration.

This, then, leads to the second set of researchquestions, which dealt speci®cally with theinteraction between country and gender ondifference in stressors, coping and strain. Forthese analyses, the only signi®cant difference instressors that emerged was managerial role as asource of pressure. Signi®cant differences oncoping emerged again in this set of analyses, aswell as on mental and physical well-being.Unfortunately, the ANOVA procedure used toestablished these differences does not indicatewhere the actual differences between the coun-tries and genders lie, and therefore t-tests bycountry were conducted for each of the variablesmentioned above. It is recognized that thisis neither a de®nitive nor ideal method ofestablishing the precise nature of the difference,but it does provide an indication of whetherpatterns are emerging which can give some insightinto the ®ndings.

The t-test results revealed the following:

. For both the UK and South Africa differencesemerged between men and women on mentalwell-being.

. South Africa was the only country where a signi-®cant difference between males and femalesemerged with regard to physical well-being.

. The UK was the only country where a signi-®cant difference between males and femalesemerged with regard to coping strategies.

There are two issues that need to be discussed asfar as the ®ndings are concerned. The ®rst issue iswhether any pattern emerged with regard tocountries and stress and how such a pattern, ifany, relates to the assumptions underlying thisresearch. The second issue is the nature of thevariables that emerged as signi®cantly different,regardless of country. As far as the ®rst issue isconcerned, it appears as if not much of a patternseems to be emerging at all. This goes against theunderlying assumption of this research Ð thatdeveloping countries are more likely than devel-oped countries to exhibit gender differences due tothe relatively recent entry of women into manage-rial positions in these countries and due to arelatively lower emphasis on women's rights indeveloping countries. The lack of signi®cantdifferences in the USA between men and womenon all the variables does, however, lend partialsupport to the initial argument that prompted thisresearch. This must, however, be viewed withcaution, given the particularly small Americansample size. It is also possible that the relativelyfew countries included in the study could accountfor the lack of signi®cant ®ndings with regard tothe interaction between country and gender on thestressors. Future research could perhaps include alarger number of developed, developing andunderdeveloped countries to establish any differ-ences more clearly.

It appears that what is perhaps more relevant iswhat was not found in the current study. Discuss-ing non-signi®cant ®ndings is somewhat unusual,if not generally frowned upon in research papers;however, in the context of the debates in theliterature on occupational stress and gender itseems appropriate to explore this issue brie¯y. Therelatively large sample size also gives more weightto the meaning of non-signi®cant ®ndings.* Asmentioned at the start of this paper, Martocchioand O'Leary5 stated that no differences inexperienced stress between men and women in a

*Large samples imply increased statistical power andhence lower the likelihood of a type II error. In essence,the higher power implies that if a mean difference waslarge enough to be of import, then the statisticalanalysis should be able to ®nd it.

K. Miller et al.


work setting had been established, while Jick andMitz2 did ®nd gender differences in this area.Subsequent research6,7 has tended to be equallycontradictory. The aim of this study was to re-visitthe issue from a cross-cultural perspective in anattempt to provide greater clari®cation. The®ndings of the current study support the view ofMartocchio and O'Leary.5 The only differences tohave emerged consistently from the analyses in thecurrent study were on the measures of strain(distress) rather than the work stressors. At thispoint, it needs to be acknowledged that this couldbe a function of women being more likely toreport feelings of physical and mental ill-healththan men8 or perhaps being more aware ofsymptoms of mental or physical health thanmen. This is a dif®cult issue to overcome, sinceeven seemingly more objective measures than selfreporting of symptoms, such as visits to doctorsand sick leave, may be affected by this funda-mental bias, if it exists.

An alternative explanation for the lack ofsigni®cant results regarding work stress andcoping, together with the ®nding of signi®cantdifferences in some of the strains measures, may bethat the greater distress experienced by women hasnothing to do with their work role/s. It is oftenassumed that the strains experienced by workingwomen are a consequence of their work-relatedstress or at the very least work/home overload andcon¯ict. This may not be the case. Indeed,Greenglass8 argues that because stress researchhas focused primarily on men, the job has beenidenti®ed as the primary stressor, having negativepsychological effects. The home, on the otherhand, has been viewed as a sanctuary from work-related stress. She goes on to say that `such aperspective re¯ects . . . the assumption that, forwomen, the roles associated with home, wife,mother, homemaker are somewhat free fromundue stress' (p. 123). The greater psychologicaland physical ill-health reported by women in thisstudy could thus be attributed to greater stressfrom the external environment, rather than fromthe workplace. Even if this argument is acceptedas plausible, however, this does not mean that thegreater distress experienced by women as opposedto male managers should not be a workplaceconcern. Taking an open systems perspective,whereby organisations in¯uence and are in¯u-enced by their external environments, ignoring the

impact of non-work related stress on employeescould jeopardize workplace stress reductionefforts. To address workplace stress from a holis-tic perspective therefore, it may be necessary toaddress both the work and non-work stress experi-enced by women managers in particular, andfuture research efforts could perhaps be directedtowards examining the interaction of work andnon-work stressors in determining strain.

Conclusion

The aim of this study was to exploit the oppor-tunities provided by the CISMS project toexamine gender differences in work stress froma cross-cultural perspective. Like many previousstudies in this area, the present research failed tosupport the notion that there are gender differ-ences in work stress and only limited support wasfound for the notion of an interaction betweengender and country on work stress. Those differ-ences that did emerge in the various analysesconducted, related primarily to differences instrains (distress). A number of possible explana-tions for the ®ndings were provided. Theseranged from attributing the lack of signi®cant®ndings to the limited number of countries in thestudy, to accepting that the lack of signi®cant®ndings may indeed be re¯ecting that genderissues in work stress is a non-issue, at least in theoccupational stress variables measured here.Finally, the implications of the ®ndings regardingthe differences in the strain measures wereexamined, with the main one being the import-ance of examining the role of extra-organisa-tional stressors on female managers' distress.

References

1. Barling J, Sorensen D. Work and family: in search of arelevant research agenda. In Creating Tomorrow'sOrganizations. A Handbook for Future Research inOrganizational Behaviour, Cooper CL, Jackson SE (eds).John Wiley and Sons: Chichester, 1997; 157±170.

2. Jick TD, Mitz LF. Sex differences in work stress. Acad.Management Rev. 1985; 10(3): 408±420.

3. Nelson DL, Quick JC. Professional women: are distressand disease inevitable? Acad. Management Rev. 1985;10(2): 206±218.

4. Baruch GK, Biener L, Barnett RC. Women and gender inresearch on work and family stress. Am. Psychol. 1987;42: 130±136.



5. Martocchio JJ, O'Leary AM. Sex differences in occupa-tional stress: a meta-analytic review. J. Appl. Psychol.1989; 74(3): 495±501.

6. Guppy A, Rick J. The in¯uence of gender and grade onperceived work stress and job satisfaction in white-collaremployees. Work and Stress 1996; 10(2): 154±164.

7. Narayanan L, Menon S, Spector PE. Stress in the work-place: a comparison of gender and occupations. J. Org.Behav. 1999; 20: 63±73.

8. Greenglass ER. Gender, work stress and coping:theoretical implications. J. Soc. Behav. Personal. 1995;10(6): 121±134.

9. Matuszek PAC, Nelson DL, Quick JC. Gender differencesin distress: are we asking all the right questions? J. Soc.Behav. Personal. 1995; 10(6): 99±120.

10. U.S. Department of State. Taiwan Human RightsPractices. 1994.

11. Cooper CL, Williams S. Occupational Stress IndicatorVersion 2.0. NFER-Nelson: Windsor, 1996.

12. Spector PE. Development of the work locus of controlscale. J. Occupat. Psychol. 1988; 61: 335±340.

13. Hofstede G. Values Survey Module 1994 Manual. Institutefor Research on Intercultural Co-operation: Maastrich,1994.

14. Cooper CL, Sloan SJ, Williams S. Occupational StressIndicator Management Guide. NFER-Nelson: Windsor,1988.

15. Nunnally JC. Psychometric Theory (2nd ed). McGrawHill: New York, 1978.

K. Miller et al.


Occupational stress and gender: a cross-cultural study

Documents

Transcript of Occupational stress and gender: a cross-cultural study