December 7, 2016 | Author: senoritapicha | Category: N/A
Gender, Time Use, and Health Author(s): Chloe E. Bird and Allen M. Fremont Source: Journal of Health and Social Behavior, Vol. 32, No. 2 (Jun., 1991), pp. 114-129 Published by: American Sociological Association Stable URL: http://www.jstor.org/stable/2137147 . Accessed: 12/09/2013 15:53 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp
. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact
[email protected].
.
American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access to Journal of Health and Social Behavior.
http://www.jstor.org
This content downloaded from 201.213.38.62 on Thu, 12 Sep 2013 15:53:59 PM All use subject to JSTOR Terms and Conditions
Gender,TimeUse, and Health* CHLOE E. BIRD ALLEN M. FREMONT University ofIllinois
Journal of Healthand Social Behavior1991, Vol. 32 (June):114-129
One of the continuing paradoxesfacing social epidemiologists concernssex in morbidity differences and mortality. womenlivelongerthanmen,they Although apparently get sick more. We hypothesize thatwomen'shighermorbidity levels resultfromless paid workand lowerwages combinedwithmorehoursspentin householdlabor, childcare, and helpingothers,and fewerhoursof leisureand sleep. Men and womenhold different social roles; menhold mostof thehighly rewardingroles. We operationalizesocial roles as timecommitments to various role-related activities.Thisapproachprovidesinterval-level measuressuchas time spent in caringfor childreninstead of simple dichotomiessuch as parent! nonparent.Wefindthatwhengenderdifferences in social roles are controlled, beingmale is associatedwithpoorerhealththanbeingfemale.Weconcludethatif genderroles weremoreequal, womenwouldexperiencebetterhealththanmen, moreconsistent withtheirgreaterlongevity. Althoughmuchtimeandenergyhavebeen devoted to measuringsex differencesin morbidity andmortality, farless untilrecently efforthas been applied to explainingwhy thesedifferences exist.Healthstatistics show thatwomenhave moreillnessand disabilities thanmen,includingacuteconditions suchas respiratory infections and chronicconditions such as arthritis (NationalCenterforHealth menhavehigher Statistics1983). In contrast, * We are indebtedto John Mirowskyand CatherineRoss for theiradvice and encouragement. We thank Lowell Hargens, Beth Anne Shelton,and Gray Swicegood for assistancein variousphasesof thework.We wouldalso liketo thank Alan Peshkin, Barbara Reskin, Gillian Stevens, and anonymousreviewersfor helpful comments on earlierdrafts. to Chloe E. Bird, Direct all correspondence of Sociology,University of Illinois, Department 326 LincolnHall, 702 S. WrightSt., Urbana,IL 61801. The dataused in thisarticlewerecollected by ThomasJusteret al. (1983), and were made available by the Inter-University Consortiumfor Politicaland Social Research. An earlierversionof thispaperwas presented at the 1989 annual meetings of the American SociologicalAssociation.The authorsare listedin alphabeticalorder.
ratesof life-threatening diseasessuchas heart disease,whichcause morepermanent disability and earlier death (Verbrugge 1985; Wingard 1982). These differencespersist even after illness related to reproductive physiologyis excluded;moreover,theyare notexplainedby differences in tendenciesto visitphysiciansor by reporting bias (Cleary, Mechanic, and Greenly1982; Gove 1984; Verbrugge1985). Sex differences in morbidity are consistent with a stress-illness model. Women have higherratesof psychological distressincluding anxiety,depression,worry,and demoralization(Gove and Tudor 1973; Kesslerand McRae 1981; Mirowskyand Ross 1989). Social scientistsfind consistently that sex differencesin psychologicaldistress are caused by role stress,role conflict,and the degreeof commitment to genderroles(Gove 1984). For example,womentypicallybear majorresponsibility forhouseworkand child care even when theyare employed(Ross, Mirowsky,and Huber 1983). Though the biochemicalmechanisms are notwell understood,a largebodyof evidencesuggeststhat distressanddepression psychological can lead to physicalillness.For instance,researchhas shownthatpsychological distressand depres-
114
This content downloaded from 201.213.38.62 on Thu, 12 Sep 2013 15:53:59 PM All use subject to JSTOR Terms and Conditions
115
GENDER, TIME USE, AND HEALTH
Model of Gender, sion suppresstheimmunesystem,makingan FIGURE 1. Hypothesized Time Use in VariousActivities, individualmoresusceptibleto disease (JemWages,and Health mottand Locke 1984; Lazarus and Folkman Wages 1984; Solomon1985). dison theliterature psychological Although Paid Work + about theories tresshasdevelopedmorerefined Housework thecausal impactof social rolesthanhas the meacrudeandindirect literature, health physical ChildCare _ /// suresof social roleshamperresearchin both Male Health areas.Forexample,theroleofparenttypically HelpingOthers an status(i.e., whether is measuredas parental ActiveLeisure of measures Similarly, haschildren). individual Passive Leisure + arebasedon questionssuch childcaretypically as "Who does mostof thechildcare,you or Sleep yourspouse?" Althoughsuch questionsmay responthe whether yieldreliablemeasuresof dentis a parent,or whichspouseis primarily being?Gove and Hughes(1979) arguedthat andempiricalgroundsfor givelittleinfor- therearetheoretical forchildcare,they, responsible is actually assumingthatcertainsocial roles are related mationon howmuchtimeandeffort In addition, to poormentalhealth,whichin turnis linked spentintakingcareofthechildren. typeof acrossindividuals to mildphysicalillness-the primary suchdataarenotcomparable experiencedby women. The auindividualsor couplesmay morbidity because different amountsof timeto childcare thorsreasoned that women typicallyhave devotedifferent dependingin parton theage more role obligationswhich require conand housework, stantlycaringforothers,such as childrenor and numberof children. roleobligationscan We employ time-use measures as an spouses.These nurturant with women's abilityto care for of social roles. interfere operationalization alternative For example,theextentto whichan individ- themselvesproperlyand can affecttheir ual fulfills the role of "housewife" is healthnegatively.When theycontrolledfor role activitiesas well as formarital operationalizedas the amountof time that nurturant and psychiatric arrangements, living status, in activities week each individualspends related to the role (e.g., cleaning, doing symptoms,Gove and Hughes found that betweenmen and women laundry,cooking). In contrastto data typi- healthdifferences cally used, data fromtime-usestudiesoffer disappeared.In a relatedstudy,Kesslerand more precise measures and thus provide McLeod (1984) foundthatwomen'stendency involvedin thelives on how roles influence to be moreemotionally clearerinformation of social of those around them made them more healthand allow greatercomparison thanmento negativelifeeventsin vulnerable rolesacrossindividuals.have is thattimespentin theirsocial network;otherresearchers Our generalhypothesis social rolesexplainstheeffectsof genderon linkednegativelifeeventsto physicalillness health(see Figure1). We expectthatwomen (Holmesand Rahe 1967). (1989) analyzedsex differences Verbrugge spend moretimein housework,child care, controllingfor an unusually and helpingothers,and less time in paid in morbidity, work,leisure,and sleep. In turn,we expect wide varietyof variablesin additionto social and thatspendingmoretimein housework,child roles.She foundthatstress,unhappiness, associated were of in levels employment time low less and others, and helping care, in sleep, worsens health, whereas spending withpoorerhealth,whereasparticipation roleswas and personallyfulfilling moretimein paid workand leisureimproves productive for or maintainshealth.In addition,we expect associatedwithbetterhealth.Controlling thatmenwill have higherwagesthanwomen these and other social factorscaused sex in healthto narrowand oftento and thathigherwages will be associatedwith differences on In fact,sex differences vanishstatistically. betterhealth. a numberof healthmeasures(six out of 67) Verwere reversed,albeit nonsignificantly. PREVIOUS RESEARCH bruggeconcludedthatthesereversalsindicate formen. healthdisadvantage How do social roles affectphysicalwell- an underlying +
-
This content downloaded from 201.213.38.62 on Thu, 12 Sep 2013 15:53:59 PM All use subject to JSTOR Terms and Conditions
116
JOURNALOF HEALTH AND SOCIAL BEHAVIOR
researchon sex differences and containsdetailed information Moregenerally, on how spenttheirtimein thehomeand in physicalhealthhas focusedon theimpact respondents employ- theworkplace,demographic ofthreemajorsocialroles:marriage, information, and reportsof theirhealth. ment, and parenthood.Most researchhas respondents' The 1981 studyis a follow-upof a 1976 foundthat marriedmen and women have lower morbidityand mortalityrates than study. In the original study, data were sample unmarriedpersons, though men tend to collectedfroma nationalprobability benefitmore from marriagethan women of adults living in the contiguousUnited (Sorensenand Verbrugge1987; Verbrugge States(N= 1,519). Bothstudiesused a panel and Madans 1985). Some research,however, designwithfourwaves of data, collectedin has foundthatdivorcedwomenare healthier 1976 and again in 1981. The 1981 sample who were at least than unhappilymarriedwomen (Wingard includesonlyrespondents tendsto 18 yearsold at thetimeof theoriginalstudy. 1982). Like marriage,employment forbothmenandwomen In addition,only those respondentsfrom yieldhealthbenefits (Verbrugge1985). Researchon theimpactof whomat least threewaves of data had been on physicalhealthhas produced collectedin 1976 wereeligibleforinclusion parenthood (Wingard1982), though inthe1981follow-up. Fromthisgroupof920 inconsistent findings 620 personswerecontactedand in a reviewMcLanahanand Adams (1987) respondents, concludethatchildrenhave a smallnegative interviewed. Attritionbetween the original and the well-being. impacton psychological Researchbasedon time-usedatashowsthat follow-upstudyresultedin some disproporof household tionateloss of low-incomeand less-educated womenstilldo thevastmajority Even so, comparisonsbetween tasks and child care, regardlessof their respondents. status(Shel- theoriginalandthe1981sampleon important educationallevelor employment ton 1989; Sheltonand Coverman1988). Hill demographic characteristics (e.g., meaneduand Stafford(1980) found that college- cationallevel) revealedno large deviations educatedmotherswho workedmorethan20 (Sheltonand Firestone1988). Althoughthe hoursa week spentless timein child-related 1981datamayhavelostsomegeneralizability activities;the reductionswere surprisinglyin comparisonto the 1976 data, thisloss is small, however. Furtheranalysis by the offsetby a large and highly significant in thequalityof the 1981 data. authorsrevealed that a workingmother's improvement and Stafford (1985) reportthatchanges abilityto workand to sustaintimespentin Juster child care was "financed"by reductionsin in data collectiontechniquesimprovedthe sleep)andin qualityof the 1981 data 20 percentoverthat herpersonalcaretime(including passive leisure,such as watchingtelevision. of the 1976 data.' withBerkand is consistent As mentioned above,datawerecollectedin This observation thatwomentakingcare fourwavesinthe1981study.Each ofthefour Berk's(1979) finding was conductedduringa have significantly fewer"pleasant waves of interviews of children were minutes" than either their husbands or different season; two of the interviews marriedwomen withoutchildren.Although conductedon weekdays,the othertwo on Hill and Stafforddid not examine the weekends.This approachgreatlyincreased databydecreasing relationshipbetween time use and health thevalidityofthetime-use outcomes, they speculated that working the possibilitythat the data reflectedan mothers' reallocationsof their time are atypicalday. Althoughtheuse of fourwaves Justerand Stafford(1985) achievedat the expenseof theirhealthand caused attrition, who remainedin the foundthatrespondents well-being. panel for all fourwaves producedhigherquality diaries on the initial wave than who appeared in Wave 1 but respondents METHOD subsequently droppedout. In addition,the qualityof diariesforlaterwaves was higher Data than for earlier waves when the authors forattrition. We use data fromthe 1981 Studyof Time standardized Timeuse was assessedbytimediaries.ReUse, collected by the Institutefor Social Research(Justeret al. 1983). The time-use spondentswereasked to recountin as much datasetconsistsofdatafrom620 respondents detail as possible how theyhad spenttheir
This content downloaded from 201.213.38.62 on Thu, 12 Sep 2013 15:53:59 PM All use subject to JSTOR Terms and Conditions
GENDER, TIME USE, AND HEALTH
117
timeduringtheprevious24-hourperiod.They more,usingsuchspecificindicesto examine in morbidity wereaskedwheretheywere,whomtheywere causes of overallsex differences they may be confoundedby differences between with,whattheyweredoing,andwhether processesin men and in were doing morethanone thingat a time. pathophysiological Robinson(1985) concludesthat24-hourrecall women(Waldron1983).5 meth- Althoughthereare disadvantagesinherent diarieswereas reliableas otherdiary-type of health, they are ods such as electronic"beeper" studies,and in using self-reports general probablyno greater thanthoseassociatedwith wereclearlysuperiorto respondents' ofhowmuchtimetheyspentin var- using"objective"measuresof healthsuchas estimates discussionof the physicians'assessments.Indeed,forgeneral ious activities.(For further oftimediariessee Juster measuresof health,self ratingsmay be as validity andreliability 1985,pp. 63-88 andRobinson1985,pp. 33- valid as ratingsby physicians.The assumption thatphysician-based measuresare the 59.) shouldbe Juster et al. (1983) combinedand weighted standardagainstwhichself-reports time-usedatacollectedduringthefourwaves evaluatedis notwell supported.Not onlydo -in theirspecito computea weeklyaverage or synthetic clinicalmeasuresvarywidely sensitivity, and abilityto predictfuture week offersa strong ficity, week.2 The synthetic measureof how timeis actuallyused over a healthstatus,but researchhas demonstrated whole week ratherthan on a given day manybiases in theway in whichphysicians ages, (Robinson1980). We use only respondents assess and treatpatientsof different week could be con- sexes, incomes, physicalappearances,and for whom a synthetic structed(i.e., theycompletedat least three ethnicbackgrounds(Eisenberg1979, 1986; and forwhomthereare Kaplan and Camacho 1983). Moreover, waves of interviews) healthis a The researchhas shownthatself-rated no missingdata on variablesof interest. 1981 data are limitedto Whitesonlybecause strongerpredictorof mortalitythan are of thesmallnumberof Blacks who remained physicians'assessments;only age predicts morestrongly (Mossey and Shapiro in the study.Our finalsample size is 469, morality with186 menand 283 women(see appendix 1982). Self-ratedhealth is a significant evenwhenhealthstatus of mortality for a table comparingcharacteristicsof predictor used withthose of respondents measuresare controlled.For instance,Idler, respondents Kasl, and Lemke (1990) found that poor excludedfromtheanalysis).3 self-ratedhealthwas a strongpredictorof mortalityover a four-yearperiod despite extensivecontrolsforbaselinehealthstatus. Measures is made, we unless a distinction (Hereafter, Dependentvariable. The dependentvari- use the terms "self-rated health" and healthas assessedbytheques- "health"interchangeably.) ableis self-rated Independentvariables. Independentvarition, "Comparedto otherpeople yourage, wouldyousaythatyourhealthis (1) poor,(2) ables used in theanalysisincludesociodemofair,(3) good,or(4) excellent?"Thismeasure graphicvariablesand indicesof timespentin of generalhealthis bothreliableand repro- activitiesrelatedto variousroles. Sex and withmore maritalstatusare dummyvariables.Sex is strongly ducible,and is correlated "objective"measuressuchas physicians'as- coded 1 formales; maritalstatusis coded 1 personswithspousepresent.Age sessments (MaddoxandDouglass 1973;Mos- formarried seyandShapiro1982;OkunandGeorge1984). and educationare measuredin years. "ChilWe use a generalhealthmeasurebecausewe dren" is the numberof childrenaged 17 or overallhealthrather than under in the household.Wages, including focusonrespondents' on specificdiseases.4Researchon stressand salaryand bonuses, are measuredin thouillnesssuggeststhatstresscan increasean in- sandsof dollarsperyear. All time-usevariables are measuredin to disease. This individual'ssusceptibility creasedsusceptibility, however,does notnec- hoursperweek; theyincludeindicesof time essarilyproducespecificpatternsof disease spent in paid work (i.e., employment), childcare, and caringforothers, and Locke 1984; housework, acrossindividuals(Jemmot specificindicesof as well as timespentin active and passive Selye 1985). Consequently, diseasesmay leisureandin sleep.Table 1 liststheactivities healthor measuresof particular be less valid thangeneralmeasures.Further- includedin each time-usevariable.We adopt
This content downloaded from 201.213.38.62 on Thu, 12 Sep 2013 15:53:59 PM All use subject to JSTOR Terms and Conditions
118
JOURNALOF HEALTH AND SOCIAL BEHAVIOR
TABLE 1. ActivitiesIncludedin Time-UseVariablesa Paid workb Housework Childcare
Helpingothers
Passiveleisure Activeleisure
Sleep
Employment Meal preparation; meal cleanup;miscellaneouschores;indoorcleaning;laundry;indoorrepair and maintenance (e.g., paintinga room,plumbing);applianceand furniture repair Babycare;childcare;helping/teaching children (e.g., makingcookies);helpingwithhomework; givingchildorders/instructions; disciplining, readingto children; conversations with;indoorand outdoor playing with; giving or obtainingmedical care for; babysitting(unpaid) for nonhouseholdchildren;coordinating/facilitating child's social or instructional activities; miscellaneous activities relatedto childcare(e.g., makingphonecalls aboutchild);child-related travel Medicalcareto adultsin household(HH); nonmedical careto adultsin HH (e.g., rana bathfor husband);helpcare to relativesnotlivingin HH; help and care to neighbors and friends;help and care to others(unrelated);watchingpersonalcare activitiesof others;travelrelatedto helpingothers Radio; television;listeningto music;reading;conversations (includingphoneconversations); conversation withhouseholdmemberswritingletters;readingmail; relaxing;thinkingand planning;sitting;miscellaneouspassiveleisure Team sports;racquetsports;golfing;swimming;skating;skiing;bowling;pool; ping-pong; pinball;frisbee;catch;exercises;yoga; hunting; fishing;boating;sailing;walkingforpleasure; hiking;jogging; running;bicycling;horsebackriding;dance; gymnastics;lessons in sports, gymnastics, ordance;pleasuredrives;rideswithfamily;picnicking; photography; working on or leisureequipment;collectionsor scrapbooks;carpentry repairing and woodworking as hobby; preserving foodstuffs; knitting; needlework;sewing;animalcare (if not farmer);art; writing literature or diary;playinginstrument; singing;acting;playingcards;boardgames;social games (e.g., scavengerhunt);going campingor to the beach; puzzles; lessons in music; crafts; miscellaneous travelrelatedto activeleisure Nightsleep
All time-usevariablesare measuredas thenumberof hoursperweekspentby respondents in a givenactivity or set of activities.Withtheexceptionof paid work,values forall variableslistedhereare synthetic week estimates derivedfromtimediaries. b Hoursofpaid workare assessedby thequestion,"Abouthowmanyhoursdo youworkon yourjob in an average week,including bothpaid and unpaidovertime?"Because somerespondents workpart-time, we expectthismeasure of employment to be morereliablethana synthetic-week estimate. a
Juster et al.'s (1985) measuresof activeand passiveleisure.Activeleisureincludesa wide varietyof recreational activities,all of which requirephysicalor mentalexertion,such as team sports,swimming,horsebackriding, picnicking, and boardgames. Passive leisure includes recreationalactivitiesthat do not requirephysicalor mentalexertion,such as watchingtelevision,listeningto recordsor tapes,readingnewspapers,and talkingwith others.
negativelywithpassive leisure(r = -. 15) and timespentin sleeping(r = - .2 1). We suspectthatthese negativeassociationsare due in partto the cumulativenatureof time use. For example, the alternativeto time spent in passive leisure or sleepingis not merelybeingawake,butpursuingsomeother activity thatmayhavegreaterhealthbenefits. The benefits of passiveleisureand sleep may diminish at highlevels,indicating a nonlinear relationship. To whatextentdo conventional categorical measuresof social roles explain sex differences in health?Figures2 and 3 illustrate RESULTS the effects of marriage andemployment on men's for Table 2 shows the correlationsof all and women's healthwithoutcontrolling measuredvariables.Self-rated healthis asso- timeuse.6Figure2 showsthedeviationfrom ciatedpositivelywithbeingmale (r = .07), theoverallmeanhealthlevel formenand for marriage(r = .10), education(r = .31), womenby maritalstatus.7Thoughmarriage number ofchildren(r = .07), paid work(r = has positive effectsfor both sexes, men .26), and wages (r = .26), and negatively benefitmorethanwomen. Figure 3 shows the deviationfromthe with age (r = - .19) and time spent in household labor (r = - .15). Self-rated overall mean healthlevel of men and for healthis not relatedsignificantly to hours womenwho wereemployedfull-time and for spentin childcare, activeleisure,or helping thosewhowerenotemployed.Like marriage, others. Surprisingly,health is associated employment has a positiveeffect on healthfor
This content downloaded from 201.213.38.62 on Thu, 12 Sep 2013 15:53:59 PM All use subject to JSTOR Terms and Conditions
TABLE 2. Correlations,All Measured Variables Health Health Married Education Age Sex (Male = 1) Children Sleep Paid work Housework Childcare Passive leisure Wages Activeleisure Helpingothers tp