.3... cm ... anmy w mi .. .3. . *. .. . .Mflvmmwmfi 4.“ .. ummmnykww v .. . . I. , , . 2. . . . . . ,. . . Iv...”- . . .. . am” . vlmua . . . . | Ir! .5 . . _ . , ...... m... a.” Fab». .3 . . .7... :. wwgmmmwww ., .. wwummwfimmfi . u . i .. . . {I , ,A u):- . v (V . . . . .. .053... .w......;§ .z. m..mw...m,...3..,fi.u inausfiaait .. . H...» flu a. . in”. aw... 1.4+? 1......“ 5. . v.2: 4.. 2.: yl: nu airway“! ~ A“ I 534...: vi... u. ~\c¢; I I. I .tu’a-btr K v .» {Runnw 53%.... . .19.“: .6. .Hlavfiyat... . . ,fiMAmmrm fawmhb? tin-31$.fimwd I . a-.. bf. n .- . . - {Jamaww r.-. WM... 9.. .... «.3 .Mpfivfifi . ,. D..M-‘I.5 l -. y .‘t CY“. xtiONMJMVufll c .l’l‘..."!v|bnfu-§~Vu Jun")? .a? I: zhwwmmvuundnufioe .2..u£.....sw ”(duo-v. O .-v¢o ‘. 1 . 'LV c. 3 2.... .. iv -3? _ .|.. . u a, I! ‘ .Mudnxn... 0. 0!. ,i o. 00 1!)?! .nrtoziru “fin. E..- . 5.3795»... . «4.5... .690. thxni 53.9mm»? .. ‘1’}.lll no.8... .75....3: i. .v.t.h.t¥| ' I. . 'ul . . 1 Ilblllvo 't’ 15:111.. 11...... gamma . “Nah: .. to": ODIIIC on ‘vuv’ 13.71? Ethyl). t!!!) .I. ‘4 I!V\Il...av’ . . .1. I .L lup‘rt «KRONE. . Vital..- . v“ nus. v .li v rluu. .9 u... .. .3393.) c... . . "Mil”..v. “0.4.9". :1... :ul .I, .Ic..u .«Juubfiozwlet .. 5...!” .uuvvs'nt 2. . .. . hfihbvvr .v. . WM. .3 .8. . . .4 v :‘lf .1 NICK-s! ll! hilt. I. ll. 1.11.31. Zn. 9 9+4? 1 .1... .. v :4. , flaws... 2---, A. , yum... no.9.» , . *9. .. .. . . . v. t I... v a...» 5. him. . : .nufiaiwmrlmwfiiwwmm Am“... i... .. - 3W”. 5... U... $3..wa . «5%.. .6 fwwwfip .Mme . V1.1. m a... w. 2.-.... t . w”. r c I .5 I.|II .5115... th z . ~ {Dy-0ft v vlvl < .1» I . n Tm...“ . .q fieryamflluflu MAT... wEIw 2. .. o v .Li .I.. { it. (a .fi :3 .3... . . .: .Wflfi . ... r. ”up ...I is . 4% _ in a. . tn; 9 . . .mk .«Mflhm. .fimm. u... .u... . (m... :m. a. 1.4.4. . 1. .. . a 9 . b... it... 1.95.1.3. . .fiMrMfi- ”£4“..ng . h. ... V 2. , .1... haffiaw ‘ - 34.. .2: 1 3.3.0.95. . .33.)»: . l: 2 “33.9.... ......Wmma..n ; L :r. .1 1.. V tr . z- .- 1 . a? .. ..9.... ., _ V . ...........m.m_u m1... ., Um... .IIVAL’T ”zuk...,$w afinfik. WV. . . .1 m. . . um. . z . . ‘4‘ e ,. .. 5mm. xv. ' ‘c‘; . . . 3"“? - (1'; . ...1..3..F..v.....!. \é.‘.fi< u. - n. V in”: . n, L . . .‘kr q . ,. .. ...... .. .Equn. .3. , r. .2. fi-I$\ . . . by to Il-ZT . .5. . . I. 2.22.4-3}. . 9...: . . mmfiur 1x5...- 1 ... w :0 l‘ 1.5.1: t . .. . ah”...- !:S .W. ... v.12... .. . . I St 1,. fill? ‘ .. '5.vv..fii« 5:11.9h. 3 war .53.}... «.3 . . fix...&$.3.€.. thw‘ficmtm"" . n Y... . I 72.12.! 2... 3.....o.l.l..1...... V.. .1. 2:. . ‘lsfldli . vcl yvnv?b.t.w‘ Dhn~.!o.luvi|.l.'.0. § lit! val Ina E11 . c {.1 v ! a.vx.'6u..-v.tn....v-.. | ... 3.. 2...- 1! mu. 1 4k» I~.vy‘.f. .1...v|.. 10. 1.1!-“ ........ .I» 7.: l - rI-r‘\u.. l 110. ‘ n 0- nifl. I Illlllllllllllllllllllllllllllll 3 1293 01570 638 This is to certify that the dissertation entitled The Gender Division of Labor in Family Elder Care presented by Sping Wang has been accepted towards fulfillment of the requirements for PhD degree in Sociology Rae/{M Major professor Date 31 W/La‘fl /‘7?¥— U MS U is an Affirmative Action/Equal Opportunity Institution 0-12771 LIBRARY University Mlchlgan State PLACE IN RETURN 30X to remove this check To AVOID FINES Mum on or baton dot. duo. DATE DUE DATE DUE outmeyunrooord. l DATE DUE THE GENDER DIVISION OF LABOR m FAMILY ELDER CARE BY Sping Wang A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Sociology 1997 Copyright by Sping Wang 1997 ABSTRACT THE GENDER DIVISION OF LABOR IN FAMILY ELDER CARE BY Sping Wang This study took a gender-difference approach to examine how men and women differ in the care they provide to their frail relatives in a home care setting. Specifically, the study aims to explore how (1) the primary activities of caregiving in terms of task involvement and the division of labor differ on the basis of gender; and (2) gender intersects with family relationship to influence caregivers’ behaviors. With a multi-faceted quantification of the division of labor, this study examines primary caregivers’ level of task involvement, task exclusivity, and relative contribution in a task domain analysis of multivariate logistic regression. The data showed that the intersection of gender and family relationship influences the way caregiving labor is carried out, especially in the labor-intensive and time-demanding task domains. Governed by the normative expectations of marriage, spouses, both husbands and wives, show great commitment to caregiving even after a long-period of providing care. After the deterioration of care recipients’ health, they seem to have less discretion about what tasks to take on than do adult children. When primary caregivers’ assistance is examined in the context of networks of support, spouses are not very different from adult children in the way they share tasks with other caregivers, although spouses remain the major care providers of their frail relatives. Aside from family relationship, gender is a significant predictor of the care provided by primary caregivers, particularly among child caregivers. In contrast to the overall lack of gender differences in task participation between spouse caregivers, sons and daughters differ in the level of participation in tasks that are an extension of women’s reproductive roles but not in tasks that are congruent with men’s gender roles. When caregivers’ efforts are viewed in terms of networks of support, gender is the organizing principle of the division of labor in terms of both degree of task sharing and proportionate volume of care for both spouses and adult children, but only in labor-intensive personal care and housework domains. Other factors of significance include duration of care, the functional dependencies of the elderly, living arrangements, and additional adults residing in care recipients’ household, but often according to the gender and family relationship of primary caregiver and to the type of tasks involved. The significance of the gender effect throws into doubt the hypotheses of power/resources, time, and ideology of role, that have been used to explain the gender differences found in the often bivariate analysis of caregiving research. A feminist theory of gender stratification is drawn upon to explain these persistent gender differences in caregiving and to propose a model for future research in which power/resources, time, and ideology of roles are placed in the sex/gender system to examine their impact on the way men and women provide care. ACKNOWLEDGEMENTS The completion of this dissertation was not possible without the encouragement, sponsorship, and guidance of a great many individuals throughout my years of study and the completion of this dissertation. First and foremost, I would like to express my gratitude to my dissertation adviser, Dr. Rita Gallin, for her constructive guidance, effort, and encouragement through the years. Her dedicated assistance and unfailing support was extremely beneficial throughout the writing of this dissertation. I am also indebted to Dr. Manfred Stommel for his support and encouragement of my intellectual development. His research guidance has been a crucial source of inspiration through most stages of the dissertation. My thanks also go to Dr. Stan Kaplowitz and Dr. Janet Bokemeier for serving on my dissertation committee and for their comments, guidance, and suggestions during the completion of this study. I am also grateful to Dr. Richard Hill, whose intellectual advising and guidance will remain unforgettable. Special thanks go to Dr. Bill Given, Dr. Barbara Given, and Dr. Clare Collins for their financial support and their giving me permission to use their data. A debt of gratitude, especially, is owed to Dr. Bill Given for his financial support and the research opportunities he offered me through the years. He taught me to look at life through eyes which see the glass in half full, not half empty. Most of all, I would like to thank my family for their love, and my friends for their encouragement and their pushing me forward. Without them, the completion of this dissertation would have been impossible. vi TABLE OF CONTENTS LIST OF TABLES ..................................... vi LIST OF FIGURES .................................... ix CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW .......... 1 Introduction ........................................ 1 Literature Review .................................... 4 Elder Care vs. Household Labor ........................ 4 Gender and Task Involvement .......................... 7 Gender, the Division of Labor, and the Family Helping Network . . . 11 Explanatory Frameworks Tested ....................... 16 Hypotheses ....................................... 29 Summary ........................................ 30 CHAPTER 2 DATA AND METHODS ......................... 31 The Six Panel Home Care Studies ......................... 31 Outcome: Task Participation and Division of Labor ............... 39 Construction of Caregiving Tasks ...................... 4O Operationalization of the Range of Task Involvement and the Division of Labor .................................. 46 Major Predictor Variables .............................. 58 Data Analysis ...................................... 66 Summary ........................................ 70 CHAPTER 3 DESCRIPTIVES OF THE SAMPLE AND CAREGIVING PATTERNS ....................................... 72 The Frail Elderly and Family Care Providers ................... 73 Characteristics of the Sample ......................... 73 The Role Responsibility of Caregiving ................... 84 Caregiving Involvement and Task Allocation ................... 87 Task Participation ................................ 87 Division of Labor: Relative Contribution and Task Exclusivity ..... 94 Summary ........................................ 111 CHAPTER 4 GENDER STEREOTYPIC TASK PARTICIPATION ........ 114 Range of Task Participation for All Caregivers .................. 115 Adult Children’s Range of Task Participation ................... 133 Cultural Taboo and Adult Children’s Involvement in Personal Care ..... 138 vii Concluding Remarks and Discussion ........................ 142 Summary ........................................ 156 CHAPTER 5 THE DIVISION OF CAREGIVING LABOR ......................................... 158 Task Exclusivity and the Incorporation of Helpers in Task Assistance . . . . 159 Full-Sample Analysis of Task Exclusivity ................................. 161 Adult-Children-Only Analysis of Task Exclusivity ................................. 172 Relative Contribution in the Network of Support ................. 175 Full-Sample Analysis of Relative Contribution ............... 176 Adult-Children-Only Analysis of Relative Contribution .......... 184 Concluding Remarks and Discussion ........................ 187 Summary ........................................ 203 CHAPTER 6 A FEMINIST INTERPRETATION ................... 205 From a Gender Differences Approach ....................... 205 A Feminist Theory of Gender Stratification .................... 207 A Proposed Model for Future Research ...................... 215 A Tentative Testing .................................. 224 Summary ........................................ 231 CHAPTER 7 CONCLUSION AND PUBLIC IMPLICATION ........... 233 Summary of Findings ................................. 234 Policy Implications .................................. 241 BIBLIOGRAPHY ...................................... 245 viii Table 2. l 2.2 2.3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 LIST OF TABLES Sample Description by Sub-studies ....................... Reliability of Task Participation, Relative Contribution and Task Exclusivity for Each Task Domain, by Relationship of Caregiver to Care Recipient ................................... Number of Cases Applicable in Each Task Domain ............. Major Characteristics of Care Recipients by Relationship of Caregiver to Care Recipient ................................. Major Characteristics of Primary caregivers by Relationship of Caregiver to Care Recipient ........................... Child Caregivers’ Household Income and Their Parents’ Living Arrangement by Marital Status ......................... Caregivers’ Attitude toward Caregiving Role by Relationship of Caregiver to Care Recipient ........................... Task Participation of Primary Caregivers by Relationship of Caregiver to Care Recipient ................................. Range of Task Participation on a Continuous Scale by Relationship of Caregiver to Care Recipient ........................... Range of Task Participation in Collapsed Categories by Relationship of Caregiver to Care Recipient ........................... Relative Task Contribution Scale by Relationship of Caregiver to Care Recipient ...................................... Two Indices of Division of Labor on a Continuous Scale by Relationship of Caregiver to Care Recipient ................. Relative Contribution in Collapsed Categories by Relationship of Caregiver to Care Recipient .......................... 3.11 Task Exclusivity in Collapsed Categories by Relationship of Caregiver 4.1 4.2 4.3 4.4 4.5 4.6 4.7 5.1 5.2 5.3 5.4 6.1 6.2 to Care Recipient ................................ Coefficients Used to Predict Task Participation for Relation-by-gender and Relation-by-duration-of-care Subgroups in the Model 103 (OINTIG,R,DC)=B0 + Bl*Gender + 32*Relation + B3*Duration- of-care + B4*G*R + BS*R*DC, Where Male, Child, and Short-term Caregivers Are Base-line Groups ....................... Odds Ratio of Caregivers Participating in All Tasks, by Task Domain . Odds Ratio of Child Caregivers Participating in All Tasks, by Task Domain ...................................... Odds Ratio of Child Caregivers Participating in All Tasks in the Personal Care Domain ............................. Coefficients Used to Predict Task Participation for Child Caregiver’s Gender (CGender) by the Gender of the Parent for Whom They Care (PGender) in the Model log (OINTICGender,PGender)=BO + B1*CGender + BZ*PGender + B3*CGender*PGender, Where Son and Father are Base—line Groups ....................... Sample Mean and Significance Test of the Index of Caregivers’ Felt Role Responsibility to Caregiving by Relationship of Caregiver to Care Recipient and Duration of Care ........................ Selective Sample Descriptives by Family Relation and Duration of Care Two Sets of Odds Ratio Related to Caregivers’ Task Exclusivity Pattern, by Task Domain ............................ Two Sets of Odds Ratio Related to Child Caregivers’ Task Exclusivity Pattern, by Task Domain ............................ Two Sets of Odds Ratio Related to Caregivers’ Relative Contribution Pattern, by Task Domain ............................ Two Sets of Odds Ratio Related to Child Caregivers’ Relative Contribution Pattern, by Task Domain .................... Odds Ratio of Caregivers Participating in All Tasks in Total Task Activities, Personal Care, and Housework, by Family Relation ..... Two Sets of Odds Ratio Related to Caregivers’ Task Exclusivity Pattern in Total Task Activities and Two Female-Oriented Task 162 173 178 185 226 6.3 Domains, by Family Relation ......................... Two Sets of Odds Ratio Related to Caregivers’ Relative Contribution Pattern in Total Task Activities and Two Female-Oriented Task Domains, by Family Relation ......................... xi Figure 2.1 2.2 LIST OF FIGURES Recruitment Criteria for Caregiver-patient Dyads .............. List of Caregiving Tasks ............................. xii CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW Introduction Interest in gender and caregiving has escalated in family caregiving research in recent years (Dwyer and Coward 1992; Miller and Cafasso 1992; Fisher and Tronto 1990; Finley 1989; Graham 1983). The gendered nature of family caregiving labor is evident in that, first, women constitute the predominant workforce in primary caregiving labor, and second, work performed by primary caregivers is carried out along gender-defined lines. According to a national survey (Stone, Cafferata, and Sangl 1987), women account for more than two-thirds of primary caregivers who provide direct care to a family member at home, whereas men are more likely to assume a secondary caregiving role, providing sporadic care and giving intermittent help. Second, men and women who assume primary caregiving responsibilities diverge in the tasks they perform (Horowitz 1985a; Montgomery and Kamo 1989; Stoller 1990). Men tend to participate in tasks that are extensions of their masculine roles while women tend to work in areas congruent with their reproductive roles and at times often extend their scope of care to masculine tasks (Iutras and Veilleux 1991). In addition, the division of labor between primary and secondary caregivers is dependent on the gender of primary caregivers. Women, particularly wives, receive much less assistance from secondary caregivers than do men (T ennstedt, McKinlay and Sullivan 1989). This gender inequality in the task involvement of primary 2 caregivers and in the division of labor between primary caregivers and other care providers is the focus of this study. Various sociological perspectives, including time availability, role ideology, and power/resources have been used to explain the gender division of household labor and have been tested in research on gender and caregiving labor (Finley 1989; Dwyer and Coward 1991; Stoller 1983; Matthews and Rosner 1988; Brody and Schoonover 1986). Nevertheless, studies that adopt or test these frameworks have been criticized for their methodological shortcomings and theoretical frameworks (Dwyer and Seccombe 1991; Finley 1989). At the methodological level, extant studies lack sound designs to capture the complexity of the gendered division of labor in caregiving. First, the arbitrary categorization of tasks into the same task domain of work that is differently gendered is dubious. Second, the simplicity of research designs (often bivariate analyses) fails to consider a wide range of characteristics that are associated with the provision of care. Third, primary caregivers’ level of task involvement or hours of care is extensively examined but how labor is divided between primary and secondary caregivers is mostly unknown. At the theoretical level, caregiving research often ignores the aspect of the obligatory component and norm of reciprocity in a marital relationship, factors that may outweigh gender relations in constraining spouse caregivers’ care provision behaviors. How gender intersects with family relationships has not been systematically addressed in research on gender and caregiving labor (Dwyer and Seccombe 1991). In addition, perspectives that resort to power/resources, time, and role ideology to explain gender differences in caregiving fail to see the structural nature of gender 3 differences, that is, to consider that gender differences in caregiving have been institutionalized and thus taken on a life of their own (Finley 1989; England and Farkas 1986). Most of all, these factors have to be situated in the context of the sex/gender system to understand their influence on how men and women provide care. The ”sex/gender system," a concept introduced by Gayle Rubin, is "the set of arrangements by which a society transforms biological sexuality into products of human activity" (Rubin 1975, p.159). It ensures the social creation of two dichotomous genders from biological sex and a particular sexual division of labor organized in ways oppressive to women (Thorne 1982). And, it is within the sex/gender system that gender differences and, thus inequality, are produced and reproduced and that the conception of time, the benefit attached to power, and the application of gender role beliefs become gendered so as to make the caregiving experience a woman’s disadvantage. The purpose of this study is to take a gender-difference approach to explore gender differences in caregiving and to specify where differences between men and women occur. More specifically, research questions are: (1) How do the primary activities of caregiving men and women differ on the basis of gender? (2) How does the intersection of gender and family relationship influence the way caregiving labor is carried out, and for what type of tasks does such interaction surface? Limited by data, this study is not able to test the gendered mechanisms of time, power/resources, and gender-role attitudes or to propose and test a conceptual framework to answer questions about why men and women provide care differently. A feminist stance, however, is taken to theorize about the gender differences in caregiving for elderly 4 relatives and to address further the issue of the gendered mechanisms involved in producing the way men and women provide care to their frail family members. Some data analyses are presented as a preliminary guide for future research addressed to the question of why such gender differences occur and how power/resources, time, and gender ideology affect men’s and women’s caregiving behaviors differently. Literature Review The literature review is organized into three parts. In the first part, caregiving labor as a special family labor is highlighted. The second part presents empirical evidence of gender differences in caregiving and how this association is confounded by the family relationship of caregivers to care recipients. In the third part, several sociological frameworks that have been offered to explain gender differences in caregiving are reviewed and their empirical validity and theoretical grounding are critically examined. Elder Care vs. Household Labor Caring has long been identified with feeling and affectionate love. The overriding image of "caring-as—feeling" that emphasizes the emotional and symbolic bonds between the carer and cared has obscured the material aspects of caring. Not until recently has elder care been conceptualized as one type of unpaid domestic labor, which is hidden under the cultural rhetoric that sees caring as emotional bonds based on compassion, feeling, and love (Graham 1983; Finley 1989; Stoller 1990; Abel 1986; Finch and Groves 1983). Although research on elder care now agrees that there needs to be a shift away from the romanticizing of caring to conceptualizing elder care as a form of family labor, elder care shares many similar and yet 5 fundamentally different features with other domestic work (Stoller 1990). Both elder care and other domestic labor are "monotonous and repetitive, with no boundaries in time and no sense of completion" (Stoller 1990, p.228). Like domestic labor, caregiving is part of the social division of labor in which social production is accomplished under the guise of family responsibility (Humphries and Rubery 1984). Like domestic labor, caregiving labor is often undertaken by female family members. The significance of gender and gender inequality can be seen in the predominance of women in the caregiving work force (Stone, Cafferata, and Sang] 1987), differences in men’s and women’s levels of involvement and task participation, and the division of work based on the gender of caregivers (Horowitz 1985a). Nevertheless, caregiving labor is more than one type of family labor. First, there is no fixed set of tasks applicable to all households. Family caregiving begins when a family member becomes impaired physically or mentally (Dwyer and Coward 1992). The demands of care depend on the functional capacity of the impaired relative to perform daily living activities and routine household responsibilities. Although both caregiving labor and domestic labor are obligatory, caregiving labor is more time demanding and restrictive, requires immediate physical presence, and includes a wider range of tasks than does domestic labor. Second, unlike domestic labor, "there is a strong emotional component to the mode of thinking caregivers employ” (Abel and Nelson 1990, p.5). The experience of caregiving, as explained by Graham (1983, p. 15), is "a labour of love." Other than the labor that is involved in caregiving, nurturance, emotion, connection, and obligations are also an integral component of caregiving as services are provided. 6 Women, who display the qualities of sensitivity, empathy and compassion, however, are especially vulnerable to this mode of thinking as it interacts With the patriarchal definition of caring as ”natural" for women and thus as their responsibility (Graham 1983). Third, in routine domestic work, spouses are usually the major contributors.‘ Yet in elder care, tasks are undertaken by a variety of family members; spouses and children are usually the main providers. Given the qualitatively dissimilar relationship between husband-wife dyads and children-parent dyads, the gendered division of caregiving labor may be complicated by the family relationship of primary caregivers to care recipients. Lastly, the focus of the gender division of work in caregiving and domestic labor is different. Research on the domestic labor of husband-wife dyads explores the amount of time husbands or wives spend on housework and/or child-care tasks (South and Spitze 1994; Coverman and Sheley 1986; Berardo, et a1. 1987; Vanek 1974) or the division of domestic labor between husband-wife dyads (Nickols and Metzen 1982; Coverman 1983; Model 1981; Berk 1985). In caregiving research, researchers inquire about the factors that contribute to the variation in the time and tasks male and female primary caregivers perform in caring for their relatives (Enright 1991; Dwyer and Seccombe 1991). Foremost but not most often, gender inequality is placed in the context of the allocation of work to primary and secondary caregivers and researchers ask if this allocation of work is based on the gender of primary caregivers. 1Berg (1985) and Berk and Berk (1978) are the few exceptions who discuss children’s roles in task sharing of domestic labor. 7 In sum, caregiving labor is often compared with domestic labor in caregiving research both theoretically and empirically. Similarities and differences between the two are thus first highlighted. Like domestic labor, the process of caregiving is part of the social division of labor in which gender is often the organizing principle of who does what. The focus of the gender division of labor in caregiving, however, is different from that in domestic labor. Caregiving research is concerned with whether the division of labor between primary and secondary caregivers is based on gender and if, given the qualitatively dissimilar relationship of patient-caregiver dyads, this relationship is complicated by the family relationship of caregivers to the elderly for whom they care. Gender and Task Involvement In many ways, the gendered disparity in the level of involvement found among primary caregivers in elder care is similar to the gender-segregated division of labor found among married couples. Male caregivers, as compared to their female counterparts, usually limit their role as primary caregivers in both time and level of task participation, and the tasks in which they participate are structured along a traditional gendered division of household labor (Jutras and Veilleux 1991). The gendered nature of work in caregiving is, however, more complicated than that in other domestic labor, primarily because the family relationship of primary caregivers to care recipients is likely to confound the association between gender and family caregiving (Dwyer and Seccombe 1991). In caregiving research that focused on adult children, sons were found to spend fewer hours in caring for parents and to engage in fewer tasks than were 8 daughters (Stoller 1983). These gender differences in the level of involvement become more obvious when parents’ need level intensifies. When the demands of care are limited to occasional household tasks, sons are as likely as daughters to help their parents. As parents’ needs escalate and they require intensive help with routine domestic tasks and personal care, sons’ involvement falls considerably below that of daughters (Stoller 1990; Montgomery and Kamo 1989). Further, the types of tasks carried out by sons and daughters are subject to stereotypic gender roles (Horowitz 1985a). In the analysis of modalities of caregiving, Archbold (1983) noted that daughters are more likely to be care providers involved in hands-on care while sons tend to be care managers, giving commands, organizing help, and offering intermittent assistance. Matthews and Rosner (1988), in an examination of style of caregiving, also noted that daughters were more likely to assume a routine caregiving style while sons were more likely to assume a "backup," ”circumscribed” or ”sporadic” style of care. The gendered style and modalities of caregiving manifest themselves not only in the division of work between primary caregivers (usually women as care providers) and secondary caregivers (usually men as care managers) but also in the work performed by primary caregivers (Horowitz 1985a; Montgomery and Kamo 1989; Stoller 1990). For example, Horowitz (1985a) found that when daughters assumed the primary caregiving role, they tended to be committed to direct caregiving in areas such as personal care (bathing, toileting, dressing) and household chores, which are extensions of their nurturing role. Sons, in contrast, primarily assumed roles in areas that are extensions of the masculine role, such as transportation, household repairs, and financial assistance. 9 Although task participation in caregiving tends to be gender-specific, that is, men engaging in male task domains and women in female task domains, the gendered specialization of care is not entirely supported. While daughters have consistently been found to be more likely than sons to provide personal care and to do household tasks, sons’ involvement in tasks that tend to be male-oriented, such as financial assistance, is not significantly greater than that of daughters (Horowitz 1985a; Montgomery and Kamo 1989; Stoller 1983). Further, Jutras and Veilleux (1991) have noted that when sons become primary caregivers, they limit their caregiving role to traditional male domains whereas daughters tend to extend their nurturing roles, becoming care managers as well as care providers. The convergence of gender roles, in short, is seen mostly on the part of daughters. Sons’ involvement in the "women’s sphere," if it occurs at all, lags considerably behind daughter’s entry into the "men’s sphere" as caregiving responsibility is assumed. In contrast to the research attention given to adult children’s involvement, spouses’ experience in caregiving is often neglected (e.g., Dwyer and Coward 1991; Finley 1989; Stoller 1983) or is not distinguished from that of adult children (e. g., Jutras and Veilleux 1991; Miller and Cafasso 1992). It is suggested that spouses’ caregiving behaviors are likely to be constrained by the obligatory component and norm of reciprocity underlying marital relationships and, thus, is less likely to be constrained by traditional gender roles (Stoller 1992; Chang and White-Means 1991). For example, the roles of women as care providers and men as care managers are not always as distinctive among elderly spouses as they are among children who provide parental care. Husband caregivers are more willing to transcend gender roles 10 than are other male caregivers (Chang and White-Means 1991), reporting more, or at least equivalent, hours of care as do wife caregivers (Stone, Cafferata, and Sangl 1987; Enright 1991; Dwyer and Seccombe 1991). There are also no gender differences in the scope of care or the type of tasks performed by husbands and wives, especially when their spouses’ health condition deteriorates (Brubaker 1985; McAuley, Jacobs, and Carr 1984). Indeed, Dwyer and Seccombe (1991), found that husbands even reported performing a greater proportion of personal care tasks and household tasks than did wife caregivers. In a task-by—task analysis, Stone, Cafferata, and Sangl (1987) also noted that a greater proportion of husbands than wives performed housework chores for which their disabled spouses needed assistance, although the gender-specific participation in other tasks remained remarkable: more wives than husbands became involved in personal care tasks such as eating and dressing while more husbands than wives helped with transportation and moving around the house.2 Two explanations have been proposed for the departure of elderly spousal care from parental caregiving and other domestic labor. First, as suggested by Johnson (1983), the normative expectation of marriage in most societies emphasizes the obligation of husbands and wives to fulfill each other’s instrumental needs while providing emotional care, especially when spouses become ill. Any gendered norms 2The results from Stone, Cafferata, and Sangl (1987) should be read with caution because their analysis was based on a sample of primary and secondary caregivers. The importance of distinguishing primary caregivers from secondary caregivers is documented in Tennstedt, McKinlay, and Sullivan (1989). The caregiving responsibilities primary caregivers assume are very different from those of secondary caregivers in terms of level of involvement and type of tasks. 11 and role expectations involved in the couple’s division of labor prior to the onset of disability are likely to be superseded as spouses become disabled (Brubaker 1985; Szinovacz 1989). Second, it is speculated that husbands tend to overestimate their involvement in activities that are not part of their normative role behaviors while wives tend to underestimate their involvement in activities that are congruent with their previous role behaviors, especially if respondents are asked to report extra hours spent on household chores (Miller 1987; Stone, Cafferata, and Sangl 1987). Thus far, the findings suggest that when assuming primary caregiving responsibilities, sons are likely to participate in tasks along gender-defined lines whereas daughters tend to extend their caregiving to tasks associated with both traditional male and female gender roles. In contrast, spouses’ task performance is likely to be constrained by the norms of reciprocity in marital relationships more than by stereotypic gender roles. The constraints caregivers face in engaging in tasks seem to vary according to their gender, relationship to the person for whom they care, and the type of tasks involved. Systematic research, however, is needed to determine where the differences emerge. Gender, the Division of Labor, and the Family Helping Network To date, research on gender and caregiving labor has primarily focused on identifying gender differences in caregivers’ level of involvement. Rarely has it centered around the inequitable gender division of caregiving labor between primary and secondary caregivers. Primary caregivers may very well take on the same number and types of tasks, but how tasks are shared between them and other caregivers may be different from one caregiver to the other. The closest analysis that examines the 12 division of caregiving labor among primary and secondary caregivers is research on the size and stability of support networks (Miller and McFall 1991; Stoller and Pugliesi 1991), the configuration of the helping networks of care recipients (Coward 1987; Stoller and Earl 1983), and the support primary caregivers receive from other care providers (Tennstedt, McKinlay and Sullivan 1989; Stoller and Cutler 1992). According to the National Long Term Care Survey, of all the primary caregivers identified by elderly (i.e., 74% of total informal helpers), 46 percent were sole providers, 40 percent received only informal help, and 14 percent received both paid and unpaid help (Stone, Cafferata, and Sangl 1987). Although the majority of primary caregivers receive outside help, the amount of assistance received is often minor. Tennstedt, McKinlay, and Sullivan (1989) reported that secondary caregivers, on average, provided only one-third as much time as did primary caregivers. In addition, the division of labor between primary and secondary caregivers is likely to vary depending on the nature of the tasks to be performed. Primary caregivers tend to be the sole providers in time demanding tasks such as personal care, housekeeping, and meal preparation whereas secondary caregivers play a relatively important role only in intermittent tasks such as shopping, transportation, and home repairs. Moreover, the configuration of support networks is often dependent on who primary caregivers are in terms of family relation and gender. Spouse caregivers are less likely than child caregivers to receive help from others (Horowitz 1985b; Tennstedt, McKinlay, and Sullivan 1989; Miller and McFall 1991; Given et al. 1994), especially when intimate tasks such as dressing and bathing are involved (Finch 1989; Stone, Cafferata, and Sangl 1987; Hess and Soldo 1985). When a 13 spouse was available, as reported by Penning (1990) and Stoller and Earl (1983), a disabled elder received assistance mostly from a living spouse, and help from children and other sources was relatively limited. When a spouse was not available, adult children, particularly daughters, were usually the major providers, but other sources such as relatives, friends, and formal organizations played a significant, supplementary role in all areas of tasks for which their frail parents needed assistance. In addition to family relationship, the gender of primary caregivers plays a significant role in the help they receive from other care providers (Stoller and Cutler 1992; Enright 1991). Although husbands and wives may not differ in caregiving tasks or efforts (Enright 1991; Dwyer and Seccombe 1991) or in their use of other helpers (Miller 1990), the amount of assistance wives receive, including help in personal care and housework tasks, from both formal and informal sources is much less than what husbands or any other caregivers received (Allen 1994; Stoller and Cutler 1992; Enright 1991; Zarit, Todd and Zarit 1986; Noelker and Wallace 1985; Antonucci and Akiyama 1987; Johnson 1983; Pruchno 1990). The notion of "male helplessness" is particularly applicable to husbands who assume the caregiving role (Stoller 1992; Hooyman 1989). Compared to wives, husbands earn more attention, sympathy, and compassion for their stressful role as primary caregivers, and are thus more likely to receive help or legitimate the use of outside help (Hooyman 1989). In a society that considers caring as "natural" for women, Stoller (1992) concludes, wives’ caregiving efforts are likely to go unnoticed and be expected as compared to men’s efforts in caregiving that are acknowledged and praised. In contrast to the relatively large research literature on spouses’ helping networks, no research has explored the 14 patterns of gender differences, if any, in adult children’s helping networks and if the association is confounded by the family relationship of caregivers to care recipients, that is, if gender differences in the support received by adult children is greater, or less, than that of spouses. Despite the growing interest in the support networks of the elderly or of primary caregivers, only a few researchers (T ennstedt, McKinlay, and Sullivan 1989; Stommel et al. 1995) have systematically studied the division of labor between primary and secondary caregivers. Little is known about how the division of labor between primary and secondary caregivers varies according to the gender and family relation of primary caregivers and to the type of tasks performed, and if such caregiving patterns are different from the patterns found in the analysis of caregivers’ task involvement. The above review suggests that adult children’s caregiving behaviors in terms of both task performance and configuration of support networks are likely to be prescribed by stereotypic gender roles whereas spouses’ task involvement is more likely to be constrained by the normative expectation of the marriage relationship, although the division of caregiving labor remains very much gendered for spouse caregivers. In sum, although gender has been much in focus in family caregiving studies of parent care and spouse care, with only a few exceptions (e.g., Dwyer and Seccombe 1991; Enright 1991; Chang and White-Means 1991), research either fails to distinguish between spouses’ caregiving experience and that of adult children or inadequately examines how gender intersects with family relation to influence the way caregiving responsibilities are carried out. As suggested by Enright (1991) and Dwyer 15 and Seccombe (1991), because of the norms and expectations involved in the conjugal contract, spouses may not reveal the same patterns of gender differences as do children. According to the review above, spouses’ task involvement seems to be influenced by the normative expectation of marriage relationship although the division of caregiving labor remains gendered for spouses just as it does for adult children. More research is needed to unfold the complex relationship among caregiving labor, gender, and family relationship. The issue of time is another aspect that is ignored in caregiving research. Although duration of care is not the focus of this study, it should not be left out of model specification. As noted by Mancini and Blieszner (1989), caregiving research often lacks a longitudinal approach to understand how a division of labor changes and role adaptations develop over time. In the handful of studies that examine the size of informal helping networks, it has been found that the scope of assistance increases over time, but the size of helping networks neither increases nor decreases with it, suggesting the stability of support networks over time and a possible shift of work between primary and secondary caregivers (Stoller and Pugliesi 1991; Miller and McFall 1991). Moreover, it has been suggested that daughters, not sons, tend to progressively assume more intensive and time-consuming tasks as their caregiving career is extended (Montgomery and Kamo 1989). Nevertheless, little is known about how work is divided between primary and secondary caregivers over time and how the stability of (or change in) this division is associated with gender and family relation and if it is subject to the type of tasks for which assistance is required. Although this study does not rely on longitudinal data, retrospective measures of the 16 years spent caregiving allows for a fuller examination of the role of length of time in caregiving and to improve our empirical understanding of its association with gender, family relation, and type of tasks involved. Explanatory Frameworks Tested According to Finley ( 1989) and Montgomery (1992), to explain gender differences in caregiving, mostly found in the bivariate analysis of caregiving research (e. g., Horowitz 1985a; Montgomery and Kamo 1989), researchers have resorted to three explanatory frameworks that are commonly used in domestic labor studies: time availability, power/resources, and role ideology (Coverman 1985, 1989; Ross 1987; Geerken and Gove 1983; Dwyer and Seccombe 1991). Two additional explanatory frameworks --specialization-of-tasks and cultural taboos-- have also been suggested as explanations for gendered caregiving labor (Montgomery and Kamo 1989). In this section, both the empirical evidence and the conceptual flaws of these explanatory frameworks are reviewed. The Time—availability Model The time-availability model suggests that the division of household labor is based on rational choice: those who have the most time do the household tasks. Availability of time, as explained by Coverman (1985, 1989), is constrained in two ways: (1) competing time and role demands placed on individuals, such as employment status and number of young children, and (2) individuals’ capacity, such as the flexibility of work schedules, to respond to these demands. The predominance of women, particularly daughters, in the primary caregiving role and the gender- specific caregiving in which women assume time-demanding tasks and men 17 intermittent tasks can be explained as a consequence of competing demands; women who assume the caregiving role usually are not employed outside of the home and thus have more available time and less competing demands and role conflicts than do men to engage in time-consuming and labor-intensive tasks. Employment status and presence of spouses or young children are two of the indicators that are frequently used to measure primary caregivers’ work schedule conflicts and competing family responsibilities. Propinquity, or joint residence or geographic proximity, another dimension of time availability unique in the context of caregiving for child caregivers, measures the accessibility of help and caregivers’ capacity to respond to an elderly’s immediate needs. Caregivers who live close by are less burdened by the time demands of travelling between the residences of care recipients and caregivers and are, thus, more accessible to help than are caregivers who live far away. It is found that children who are married provide less support for their elderly parents than do children who are single (Stoller 1983; Litvin et al. 1995) and caregivers’ capacity to assist the elderly is diminished as they participate in the labor force (Lang and Brody 1983). In addition, propinquity is a primary factor that determines adult children’s caregiving, especially when the tasks rendered require daily or hourly face-to-face contact (Litwak and Kulis 1987). Coresident children provide greater amounts of personal care than do children living apart from frail parents (Soldo and Myllyluoma 1983). Nevertheless, despite the relevance of the time factor to caregivers’ caregiving, Dwyer and Coward (1991) found that gender remains a significant factor in adult children’s task assistance after the effects of adult 18 children’s work schedule conflict, competing family responsibilities, and proximity are controlled. Finley (1989), too, found that daughters, who in fact experienced more role conflicts than did sons, were more involved in caregiving than were sons after the effect of time conflict is taken into account. The Power/Resourgs Model The power/resources perspective, another framework often adopted in caregiving research to account for the gender differences found in bivariate analyses of caregiving research, stems from a sociological power framework (Blood and Wolfe 1960) and microeconomics (England and Farkas 1986). It postulates that external resources, such as income and education, determine the power dynamics within a family, thereby affecting the decision-making process and the division of labor (Ross 1987; Coverman 1989). Caregivers who possess social resources exert more power in family negotiation and, consequently, participate in fewer labor-intensive and time- demanding tasks than those without such resources (Archbold 1983; Kinnear and Graycar 1984; Noelker and Poulshock 1982).3 Women’s greater participation in hands-on activities such as personal care or housework in home care thus is a consequence of their low educational level and earnings, which limit their exercise of power in family negotiations and their ability to purchase paid help. Sons, usually 3In contrast to domestic labor research that examines the power relationship of husband-wife dyads in a relative sense, no caregiving research has collected data on the resources/power of primary caregivers in relation to that of secondary caregivers, probably for the following two reasons. First, secondary caregivers’ power/resources, when quantified, usually have a wide margin of error if data are directly collected from primary caregivers. Second, sometimes the data are less relevant, especially when secondary caregivers are paid helper or workers from a formal service sector. For example, the availability of paid help may in itself be evidence of the economic resources a primary caregiver possesses to relieve his/her caregiving demands. 19 with more relative power within the family, as suggested by Montgomery and Datwyler (1990), are able to purchase services or to relegate direct care tasks to female relatives (e. g., their sisters and their wives). Empirical research that supports this hypothesis is mixed. Archbold (1983) concluded that caregivers’ income is the primary determinant of the caregiving modality. Caregivers with high socio-economic status are likely to be care managers while those of low socio-economic status tend to be care providers. Nevertheless, when gender is taken into account, the effect of power/resources is no longer conclusive. Finley (1989) found that gender differences remain significant after power/resources are controlled. Brody and Schoonover (1986) suggested that an increase in socioeconomic status is accompanied by a decrease in the time and scope of care caregivers provide, particularly in personal care, and that the difference is often offset by purchased help so that the total demand for care of the dependent elderly is met. The Gender-Role Ideology Model The gender-role ideology perspective holds that the attitudes caregivers learn through the socialization process about women’s and men’s place in the household and market place influence the way the meaning of family care is conceptualized, the degree of adherence to a stereotypic division of labor, and, accordingly, how caregiving labor is carried out (Coverman 1989; Finley 1989; Calasanti and Bailey 1991). Because women have been socialized to a nurturing role, they are more likely than men to show concern for family and take caregiving as their responsibility (Finch 1989). The differences between men and women in caregiving may be a result of the 20 gender role beliefs they learn through the socialization processes. Support for the gender-role ideology hypothesis is limited. Men and women conceptualize the meaning of caregiving in the same manner, considering care to be manifested in instrumental help and emotional contact (Dressel and Clark 1990). Additionally, daughters expect their male siblings to take care of elderly parents just as they expect such care of themselves (Brody et al., 1983). Nevertheless, as Finley (1989) showed, women are more likely than men to care for their elderly family even after attitudes of role obligation are controlled. Men feel responsible to take care of their elderly family, but they do not actually fulfill this responsibility to the extent that women do. SpecialigtiQn-of-Tasks Model The specialization-of-tasks hypothesis, mirroring the thoughts of Parson and Bales (1955) and Becker (1981), asserts that the specialization of tasks between the genders --the assignment of men and women to different tasks-- is a mechanism that stabilizes the social system and maximizes the well-being of the family as a whole. This model, reviewed by Finley (1989), can be seen as a variant model of gender-role ideology. It postulates that since men are socialized by society to be providers and women to be caretakers, it will be best if they apply the skills they learned by fulfilling their caregiving roles in domains that are extensions of their gender roles. A type of parity can then be reached in which specialized tasks complement each other (Finley 1989). Empirical evidence that supports the specialization-of-tasks hypothesis is very limited. As reviewed earlier, daughters indeed are more likely than sons to provide 21 emotional assistance and help with household chores and personal care tasks (Horowitz 19853; Montgomery and Kamo 1989; Stoller 1983). Sons, however, are unlikely to be the primary provider of tasks that are less gender-specific or male- oriented. If specialization does take place, as Jutras and Veilleux (1991) and Montgomery (1992) contend, it does so on the men’s side only; women take on tasks congruent with the reproductive role they have learned through socialization while extending their caregiving to tasks defined as traditionally male. This model not only receives little empirical evidence but has theoretical flaws as well. Not only does it appear that the assignment of tasks to female caregivers extends beyond gender beliefs learned through socialization, the division of labor is organized in a way that is likely to be disadvantageous for women because it is usually women, not men, who are responsible for assistance with tasks that demand time and labor. The social organization of caregiving in the sex/gender system has been characterized by gender inequality rather than by the equilibrium of the social system.‘ Cultural Taboo Model Researchers who propose the existence of a cultural taboo with respect to personal care tasks (Montgomery and Kamo 1989; Montgomery 1992) hold that differential patterns of caregiving arise because caregiving tasks that require intimate bodily contact are likely to be viewed as inappropriate for adult children, particularly 4Specialization does not necessarily imply inequality. In reality, however, they frequently go hand in hand. Why has gender become an organizing principle of the specialization? Why has the division of work, based on the notion of specialization, placed women in a disadvantageous position? All these questions will be discussed in Chapter 6. 22 sons, to perform. Child caregivers are less likely than spouse caregivers to be involved in personal care because intimate bodily contact between spouses is not considered taboo whereas the bathing and dressing of a parent by a child may have incestual connotations. In addition, sons are more likely than daughters to avoid performing personal care tasks, Montgomery and Kamo (1989) contend, not so much because these tasks are viewed as women’s domain but because they are viewed as inappropriate or taboo behaviors for sons to perform. This hypothesis, though plausible, is speculative as far as the empirical evidence goes. Thus far, no empirical research has directly tested this hypothesis. Among the few studies with data available (Stone, Cafferata, and Sangl 1987; Noelker and Wallace 1985), evidence suggests that although spouses are indeed consistently more likely than adult children to provide help with personal care tasks, the gender difference in husbands’ and wives’ involvement with intimate tasks and in the informal support they receive, though less remarkable than those between daughters and sons, remains significantly large. Gender appears to be the organizing principle in the provision of care involving intimate tasks even among spouse caregivers. If the intimate nature of personal care is indeed a concern for spouse caregivers as they assist, husbands should feel as uneasy as wives to relegate personal care tasks involving bodily contact with their spouses to other helpers unless there is a great need to call in help. The dual nature of personal tasks, characterized by both labor intensity and intimacy, needs to be examined. In addition, using cultural taboos as an explanation for sons’ inactive involvement in personal care is problematic. The hypothesis does not give a full account of why bodily contact between children and 23 parents is considered taboo for sons but less so for daughters. According to research on patterns of family intimacy (see Ahn and Gilbert 1992, Table 1), the taboo of intimacy is gender-specific only when it involves cross-gender parent-child dyads and the threshold of family intimacy, in fact, is lower for father-daughter dyads than for mother-son dyads. While the cultural taboo hypothesis proposed by Montgomery and Kamo (1989) may be able to account for the differences between spouses and adult children in personal care tasks, this hypothesis is problematic when used to account for sons’ less committed involvement than daughters in these tasks. An examination of parent- child dyads of the opposite gender and same gender can help increase understanding of the role cultural taboos play in the care provided by sons and daughters. In addition, from the review above about the gender differences in the help received by spouse caregivers, the intimate aspect of personal care may be overrated. Personal care tasks appear to be tasks of labor intensity as much as, if not more than, tasks of intimacy. Sophisticated research is needed to examine the pattern of caregiving in personal care by gender and family relation. Summary All in all, an increasing amount of caregiving research has tried to address gender differences in the support received and labor provided by primary caregivers. Several explanatory frameworks have been proposed for the gender differences found in bivariate analyses of caregiving research, but only few pieces of evidence have supported them. The limited empirical evidence that supports these explanatory frameworks may, in part, be associated with the ways the key concept and outcome 24 variables are operationalized. For example, education has been used as an indicator of power/resources (Finley 1989; Brody and Schoonover 1986), but it could be an indicator of gender-role ideology (Ross 1987). Employment status was used as a measure of time availability in one study (Stoller 1983) and power/resources in another (Finley 1989). In addition, tasks associated with different gender roles are frequently combined into one measure (e.g., Dwyer and Seccombe 1991; Finley 1989). The arbitrary categorization of tasks into the same task domain of work that is differently gendered introduces errors that may be responsible for some of the contradictory findings that exist. Most of all, as pointed out by Finley (1989), these explanatory frameworks have failed to recognize that gender has been an organizing principle of the social stratification system in which gender differences in caregiving have been institutionalized to the extent that ”it has virtually taken on a life of its own, unresponsive to household variation in potential efficiency, male power, and ideology” (England and Farkas 1986, p.99). Biological sex has been defined socially and forms the foundation for a stratification system. Gender now embodies a power differential and a hierarchical relationship in the sense that one gender becomes subordinate to the other. The work women do is usually repetitive, tedious, labor- intensive, time-demanding, and is not as rewarding and visible as the work men do. The ideological definition of women’s work as reproductive work, sustained by the interplay of patriarchy and capitalism, has shaped the context of the private and public and ideologically defined women’s place at home and women’s work as reproductive work. Nurturing is central to the definition of female gender roles and 25 domesticity and care activities are deemed to be women’s "natural" inclination despite women’s entry into the labor force. Caring, seen as "natural” for women, has become a ”social category through which one sex is differentiated from the other" (Graham 1983, p. 18). When caregiving tasks are congruent with women’s reproductive roles and require high levels of commitment in time, physical space, and intense labor, women’s nurturing, reproductive role, and sense of responsibility are invoked to elicit a response to the need of their elderly relatives. Consequently, it is women, not men, who are expected to take charge of nurturing and housework tasks, that are intensive and time-consuming. Therefore, although the caring-as-feeling that emphasizes the emotional and symbolic bonds between carer and the cared may deter spouses from accepting stereotypic gender roles, thus allowing them little discretion about which tasks to assume, spouse caregivers’ help seeking and receiving behaviors can be just as gendered as those of adult children when the tasks involved are labor-intensive and time-demanding. What is missing in the explanatory frameworks of power/resources, time, and gender-role ideology, therefore, is a central focus on the sex/gender system in which gender embodies a power relationship in both public and private spheres. Not only have gender differences become institutionalized but also the conception and exercise of time and power and the enactment of beliefs are so gendered that they ensure an unequal division of labor based on the gender of caregivers. Factors such as power/resources, time, and role ideology that are proposed in these frameworks thus should be placed in the context of the sex/gender system to understand their effects on how men and women provide care. 26 For example, the effect of time constraints imposed on caregivers has been found to be gender-specific. Whereas the presence of young children and the demands of employment reduced the volume of care sons provided to the elderly, no differences were found to exist between the care provided by working and nonworking daughters (Stoller 1983; Matthews and Rosner 1988; Reece, Walz, and Hageboeck 1983; Noelker and Poulshock 1982; Horowitz 1985b; Horowitz and Dobrof 1982; Enright 1991). Even if time constraints imposed by employment limited the effort of daughters invested in care, first, they did not affect the types of assistance employed daughters provided to their parents unless the tasks involved, such as personal care and cooking, required caregivers’ immediate physical presence (Brody and Schoonover 1986; Barnes, Given, and Given 1995) and, second, working daughters were still more likely than employed sons to be actively involved (Montgomery and Kamo 1989).5 The effect of propinquity is also likely to be gendered. According to Lingsom (1989), parents with daughters in the vicinity were much less likely than those with sons who lived close by to apply for formal assistance, suggesting that daughters are probably more likely than sons to be their immediate care providers even though both are equally accessible or flexible to respond to their parents’ need. 5Montgomery and Kamo’s (1989) data should be read with caution because the type of employment is not differentiated in their analysis. Caregivers who work part- time face different kinds of time constraints than those in full-time employment. Since daughters are more likely than sons to be in part-time employment, the differences may be a result of differences in the type of employment assumed by sons and daughters. 27 Although no caregiving research has added to our knowledge of the gender- specific effect of power/resources, the domestic labor research has suggested the differential effect of power/resources by gender on the household division of labor. For example, Blumstein and Schwartz (1991) concluded from their research that wives in general need to earn more income than husbands in order to acquire equivalent amounts of power in the family. Johnson (1976) and Miller and Cummins (1992) explain that, in a male-dominated society, the style of and relationship to power prescribed by cultural rules inhibits women from using their resources as effectively as men. Blumberg (1984), based on her general theory of gender stratification, explains how the economic power women acquire through market work at the micro level is likely to be devalued by the continuation of male power and prevailing gender ideology at the macro level. Thus, women’s micro-level is never as effective as is that of men. Researchers of domestic labor (Baxter 1992; Ross 1987; Seccome 1986; Huber and Spitze 1983) have also shown that the effect of gender-role attitudes is often gender-specific. For instance, it is husbands’ egalitarian attitudes, not wives’, that are associated with the division of labor. Chafetz (1990) explains that since gender ideologies provide fewer restrictions and are more "real” and rewarding for men than they are for women, men who favor egalitarian roles are willing to envisage change in the gendered division of labor within the family (Chafetz 1990). In contrast, women’s egalitarian gender ideology is unlikely to be effective without the help of 28 structural power or change in the values of men around them (Ross 1987).6 As a result, a woman who provides care may voice an egalitarian viewpoint but actually be pulled toward the macro-level ideology that equates caring with femininity (Coleman 1988). In conclusion, this study adopts a gender-difference approach to understand how the provision of care is organized on the basis of the gender of primary caregivers and how the gender organizing principle varies with the type of tasks involved and the family relationship of primary caregivers to care recipients. Then, a feminist stance is taken to explain why such gender differences in caregiving have been institutionalized and, accordingly, how power/resources, time, and role ideology, are, in fact, gendered and have to be situated in the context of the sex/gender system to understand their influence on how men and women provide care. Due to the limitations of data, to be detailed in Chapter 2, this study does not test the influence of the gendered mechanisms of time, power/resources, and gender- role attitudes on how men and women provide care. Nevertheless, some tentative hypotheses are proposed and preliminary data analysis of the gender-specific effects of power/resources, time, and gender-role beliefs for men and women are presented in Chapter 6 as a guideline for future research focusing on the reasons why gender differences occur in caregiving and the mechanisms that shape men’s and women’s caregiving behaviors. The following hypotheses, based on the gender-difference approach, are concerned with how the provision of care is organized according to the 6The concept of structural power and why women have restricted access to the structural power will be discussed in Chapter 6. Briefly, the structural power refers to the access to the labor market and politics in the public sphere. 29 gender of primary caregivers. Hypotheses Range of Task Participation 1. Spouse caregivers provide a wider range of tasks than do child caregivers, particularly in personal care tasks that require bodily contact. Daughter caregivers are more likely than son caregivers to take on tasks traditionally defined as women’s work whereas son caregivers are not any more likely than daughter caregivers to take on tasks that are extensions of their gender roles. Husband and wife caregivers do not differ in the level of task involvement unless tasks involved are repetitive, tedious, and time-demanding. Division of Labor 4. Spouse caregivers are more likely than child caregivers to provide care without the assistance of other caregivers, particularly when intimate personal care tasks are involved. Regardless of the family relation of primary caregivers to care recipients, in domains traditionally defined as women’s responsibilities, women are less likely than men to be dependent on secondary caregivers’ involvement for task accomplishments. In male—oriented task domains, there is no gender difference in the allocation of tasks between primary and secondary caregivers among either spouse caregivers or child caregivers. Cultural taboo 7. In personal care tasks characterized by both intimacy and labor intensity, spouse caregivers are more likely than child caregivers to get involved and to provide care without the assistance of other caregivers. [Same as Hypothesis 1.4]. Adult children who care for a parent of the same gender are more likely than those who care for a parent of the opposite gender to take on personal tasks and provide help without other helpers’ assistance. But, when the gender of child caregivers is taken into account, sons who care for a mother are less likely than daughters who care for a father to take on personal care tasks and to provide care without the help of other caregivers. 30 Summary In this chapter, empirical evidence and theoretical explanations regarding gender differences in caregiving were reviewed. The caregiving tasks men and women assume diverge along gender-defined lines that intersect with the family relationship of caregivers to the frail elderly. Spouses’ endeavors are consistently found to be greater than those of adult children probably due to the obligatory norms and expectations underlined in the marriage contract. In addition, whereas the task performance of husbands and wives are likely to be constrained by the norms of reciprocity in marital relationships more than by stereotypic gender roles, the division of caregiving labor, as evidenced in some caregiving research, appears to be gendered for spouses as much as it is for adult children. Several explanatory frameworks that have been proposed and tested in caregiving research were reviewed and criticized: power/resources, time availability, gender role ideology, specialization-of-task, and cultural taboo. These extant frameworks fail to recognize the structural nature of gender differences in the sense that the gender differences in caregiving have been institutionalized and thus taken on a life of their own. A feminist framework is called upon to explain persistent gender differences and to examine the gendered nature of the mechanisms involved in producing women’s disadvantage. CHAPTER 2 DATA AND METHODS The study involves an analysis of survey data from six panel home-care studies of primary caregivers who provided care to relatives with Alzheimer’s disease, cancer, or a broad range of physical impairments. Only data from intake, i.e., the first of several interviews during a year of follow up, are employed in the analysis. This chapter has several purposes: (1) to describe the studies from which the data were drawn and the characteristics of these sub-studies, (2) to discuss the Operationalization of the outcome variables, the range of task participation and the two division of labor indices: task exclusivity and relative contribution, (3) to delineate the Operationalization of predictors, and (4) to summarize the analysis plan. The Six Panel Home Care Studies The surveys were conducted in central Michigan at different points in time, between 1987 and 1991.7 All six studies relied on convenience samples in which 7The studies were funded by the following grants: ”Caregiver Responses to Managing Elderly Patients at Home” (National Institute on Aging, NIA #1 R01 AGO6584, Charles W. Given & Barbara Given, Principal Investigators), "Caregiver Response to Managing Elderly Patients at Home" (National Institute on Aging, N IA R01 AGO6584-06, Charles W. Given & Barbara Given, Principal Investigators), "Impact of Alzheimer 's Disease on Family Caregivers” (National Institutes of Mental Health, NIMH #1 R01 MH41766, Clare Collins, Principal Investigator), "Family Homecare for Cancer -A Community-Based Model" (National Cancer Institute, NCNR #1 R01 NR01915, Charles W. Given & Barbara Given, Principal Investigators), and ”Family Homecare for Cancer Patients” (American Cancer Society, PBR-32, Charles W. Given & Barbara Given, Principal Investigators). 31 32 family members were recruited either through hospitals or through community agencies and support groups. All studies were limited to self-identified primary caregivers, i.e., caregivers who considered themselves to provide most of the care to their frail relatives. Two homecare studies of the elderly, labeled CGl for the earlier homecare study of long-term caregivers and CG2 for the later homecare study of new caregivers, were conducted in the late 19805 and early 19903 respectively, and focused primarily on caregiving experiences with the elderly who had functional limitations or physical impairments. Two dementia studies, labelled ALZl and ALZZ, conducted in 1987 and 1989 respectively, focused on family members who cared for patients diagnosed with Alzheimer’s disease or related dementias. An additional two studies, labelled CANl and CAN2, emphasized the impact of cancer on patients and the family members who provided care to them. Because the two cancer studies, conducted independently in 1988 and 1990, employed the same methods and criteria to recruit their samples, no distinction and comparison will be made between these two studies. Similarly, the two dementia studies employed the same instruments, methods, and criteria for sample selection and thus are treated together. Across all studies, participants were recruited through a variety of means. The CGZ study recruited patient-caregiver dyads through thirteen hospitals in metropolitan areas throughout Michigan. Nursing and discharge planners were employed to identify and recruit patients who had new needs for assistance within the home following their discharge from an acute care setting. All other studies (CGl, CAN, and ALZ) located family caregivers through mailings distributed by local chapters of the Alzheimer’s Association, community-based cancer treatment centers, and other community health 33 and home care agencies. Criteria for recruitment varied slightly from study to study. The method of recruitment and recruitment criteria for these studies are described in detail in Figure 2.1. The CG] study required care recipients to be 64 years old or older while the cancer study imposed no age restriction, although the majority of these care recipients are also elderly. The dementia study recruited care recipients who needed assistance with at least one instrumental activity of daily living (IADL), such as cooking or shopping, and one activity of daily living (ADL), such as eating or dressing, while caregivers in the cancer study were recruited as long as the recipients of care were dependent in one ADL or presented with cancer symptoms. Of 1,688 primary caregivers who completed the first-wave interviews across all studies, 61 percent were spouses, 26 percent were children, 5 percent were children- in-law, and the remainder (8%) were other family members. Caregiving wives accounted for 66 percent of spouse caregivers, and daughters and daughters-in-law constituted 90 percent of child caregivers. Although these studies are not based on probability random samples, these statistics about the participants are comparable to those of national studies with one exception; the proportion of spouses in this study is somewhat higher than might be expected on the basis of a national profile of caregivers (Stone, Cafferata, and Sang] 1987). Because this study is concerned with caregiving labor provided by family members, the sub-sample used meets the following criteria: (1) care recipients were 55 or older and needed assistance in daily or health care activities; (2) primary 34 Figure 2.1 Recruitment Criteria for Caregiver-patient Dyads Caregiver-patient diads were enrolled in the specific study if they met the following criteria: C61: Caregiver I study (N830? at intake) 1986-87 (a): the patient was at least 64 years of age; (b): the patient was dependent in at least two activities of daily living (ADLs) or instrumental activities . of daily living (IADLs); and ' (c): the caregiver was a family member who self-acknowledged that slhe provided the most care. (Community health agendas and home care agencies in west Michigan provided assistance with sample recruitment.) C62: Caregiver II study (II-628 at intake) 1990-91 (a): The patient was age 55 or older; (b): the patient had a recent hospitalization. because of an event or disease progression. that qualified him/her for skilled home care; (c): the caregiver self-acknowledged him/herself to be the caregiver in the family who provided the most care: (d): the caregiver assisted with at least one IADL activity; and (e): the caregiver had increased the level of involvement in at least one of ADL activities, or two instrumental activities of daily living (lADLs). or one healthcare activity. such as incision or catheter care or injection. during the 3 months prior to intake into the study. (First time family caregivers who provided help to elderly family members with a broad range of diagnoses were recruited on their discharge from 27 hospitals in Michigan.) ALZl: Alzheimer Study I (hi-229 at intake) 1987-88 a ALZZ: Study II (N8101 at intake) 1989-90 (a): The patient was diagnosed with Alzheimer’s disease or other progressive dementia; (b): the patient was age 55 or older; (c): the patient was dependent in at least one IADL activity and one ADL activity; (d): the caregiver was self-identified as the family member providing the most care to the relative with dementia; and (e): the caregiver recipient was residing in the home at the time of data collection. (Sample recruitment was done through an extensive mailing distributed by eight local chapters of the Alzheimer's Association. the Michigan Association of Adult Day Care Centers, and health agencies in southwest Michigan.) CAN‘I: Cancer Study I (N8303 at Intake) 1988-89 8: CANZ: Study ll (Ill-121 at intake) 1990-91 (a): The patient was between 20 and 80 years of age diagnosed with either a solid tumor or lymphoma: (b): the patient was in active treetrnent for new or recurrent disease at study outset; (c): the patient was dependent in at least one ADL or was symptomatic according to a standard list of symptoms; and (d): the caregiver was self-identified as the family member who provided the most care. (The sample of caregivers of cancer patients was recruited with the help of six major community- based cancer treatment centers in west Michigan.) Note: Activities of Daily Living (ADL) include caregiving activities such as bathing, grooming, and toileting. Instrumental Activities of Daily Living (lADL) include activities such as cooking and transportation. 35 caregivers were either spouses or children and children-in-law of the elderly? and (3) primary caregivers provided at least some care to the elderly in need of help. In total, 1,387 cases met these criteria. Spouses constituted 62 percent of the sub-sample while children (31%) and children-in-law (6%) made up the remainder. Again, two-thirds (66%) of these spouse caregivers were wives while among children, daughters and daughters-in-law made up 90 percent of the primary caregivers. As expected, sons and sons-in-law constituted the smallest group in the sample (N=54). Characteristics of Separate Study Samples The sample characteristics of the care recipients and caregivers across sub- studies (CG1,CG2,CAN,ALZ) for the 1,387 cases are presented in Table 2.1. In general, the sociodemographic characteristics of caregivers in the cancer study differ most from those in the other homecare studies. Caregivers in the cancer study are more likely than caregivers in other studies to be new caregivers, spouse caregivers with young children at home, financially better off, and to provide few hours of care in supervision and direct care. This low number of hours of care is related to cancer care recipients’ relatively low level of functional dependency, particularly in areas such as personal care, administrative tasks, and mobility tasks. Both homecare studies of physically impaired elderly (CGI and CG2) focused on the homecare of the elderly with a variety of physical impairment and circulatory problems -- with 80 percent of cases in CG2 study and 75 percent in CGl study associated with orthopedic problems, strokes, or cardiovascular illnesses. But the two studies did have one notable difference. The CGl study was primarily a long-term aThirteen grandchildren ( 10 granddaughters and 3 grandsons) were also included. 36 6.: 6.: 6.: 6.: 6.: am 6... 3 a... a.“ 3 s. .9963 6.30... 6.9 6.9 6.9 6.9 6.9 cm a... as 6.. 5 w... .2 .9...me 5.8.6.522 6.9 6.9 6.9 6.9 6.9 am 3 an a." 6." ea .2 699.338: 6.: 6.9 6.9 6.9 :9 cm L 3 6.." 9a m.» a." s. 699.63 98 653.. 6.9 6.: 6.9 6.: 6.: cm 6.6 6.: a... 6.: 3 s. 699.3. .33 656530 6.8.8....“— 6659 69:9 6qu: .2 K: 666: am 6.9.8 .528 s. 63:" 623 ES... 2:" 83a .2 «58... 2238.. E .9 .9 .9 .9 am 66 2 2. 3 2 .2 8< l l l l a. 86 3 SN 2 E .5569... 3.9.5. 2: EN 9 E n m 2 EN .850 l 2: in F F N .6 3 as” 362.29.. 6.6835 69.55 a m P v m 8 N 6 6 mm 398368.26 : a. mu 2 3 6: x. 6: a mom 838.2. 3 x: 2. EN 3 69.. 6 N6. .6 96 8:6... mama 5%.... =9. 3. «m 9.: 3. :m 8 62 F. m: mason. .9 In 3: in 3: 66 3.. v8 .9 69 880.3353; . z: .6 66 66 .6 66 66 66 66 66 66. .6. .66.... 5.2.5 66.. 66 66. .6 .66 66 6.6 .6 .66 66 66.. 86.6.2 66 66 6. .6 66 66. 66 66 66 666 666.656 .oz 66 66 66 66. .6 6.6 66 66. 66 .66 66.66.... 6. .6 6. .6 6 66 6 66 6. 66. 6.6.6.6.. .6 .6 6. .6 66 66. 6. 66 .6 666 656-..... 6.65.2656 66 66 .6 6.6 6 6 66 66. 66 666 6.66.. 6 A 6.8 .6 62.6.8 6 6 6 6 6 66 6 .. 6 66 .56 66 66 . 66 66 66 666 .6 6. . 66 6.6 6.66666 66 66 66 6. 6. 66. 6. 66 .6 666 2.666.... 66 66 66 .6. 66 6.6 .6 6.. .6 6.6 6...... 6.668.666 6. 66. 66 6.6 .6 666 66 .6. 66 .66 66:86 66 66. . 6. 666 t. .66 66 666 6. 666. 26...... .6... 6.6 .6... 6.6 .6... 6.6 .6... 666 .6... .66. 6668 .o .8562 2.8 N... «oo .oo .36. .636 8.66.66.66.20 6629.60 59:00. _..N 038. 38 care study in which 64 percent of respondents were cared for by relatives for more than two years. Less than one percent of the care recipients improved and required no care within one year of follow up. The CGZ study recruited caregivers whose relatives had just been discharged from hospitals, and thus it focused on caregivers who had just assumed the caregiving role or who faced elderly relatives’ new needs for assistance. In contrast to the care recipients in the CG1 study, 24 percent of those in the C62 study had no need for care three months after hospital discharge and 38 percent had no need for care within a year. This explains the higher functional dependency level (=11.5) of care recipients in the C01 study and lower dependency level (=9) of those in the CG2 study. In fact, caregivers in the CG1 study were more similar to caregivers in the Alzheimer’s study than they were to caregivers in the C62 study. Both tended to care for relatives with high levels of functional dependency, co-reside with frail relatives, and to be long—term care providers. Nevertheless, care recipients in the Alzheimer’s study were somewhat more likely to be cared for by a spouse (two-thirds of them by a wife) and their caregivers provided longer hours of care per day (13.6 vs. 11.2 for the CG1 study) than did their counterparts in the C61 study. Variations in sample characteristics among the studies may be attributed to both sample recruitment criteria and the nature of diagnoses. For example, the high level of functional dependency of care recipients in the CG1 study and Alzheimer’s study is in part due to the recruitment criteria. The CGl study and Alzheimer’s study recruited caregivers whose relatives were functionally dependent in at least two activities while the cancer study required care recipients to have only one functional 39 dependency activity. Diagnosis also played a role in the differences among caregivers and care recipients. For example, the cancer patients were usually diagnosed at a younger age than were participants with other diagnoses and therefore tended to have a living spouse who cared for them. In contrast to all other homecare sub-studies (ALZ,CGI,CGZ) in which wives and daughters were the major source of caregivers, husbands in the cancer study accounted for nearly one-third of the caregivers and, together with wives, were the primary sources of care, a reflection of the fact that many of these cancer patients were diagnosed at a young age. Additionally, cancer patients often continue to take care of their own hygiene and move around until the terminal stage of their illness. These attributes explain the overall low functional dependency of cancer care recipients and their low need for help with personal care and mobility tasks but high need for assistance in routine and daily housework. In contrast, the cognitive problems of Alzheimer’s patients and the physical impairments of the subjects in the CG1 study demanded a round-the-clock care from caregivers. Thus, child caregivers in these studies largely lived with the care recipients and provided longer hours in supervision and direct care than did caregivers in the homecare study of new caregivers (CG2) or the cancer study. Outcome: Task Participation and Division of Labor Researchers who study family household labor are often criticized for lacking a consistent operational definition of household tasks and proper estimates of the division of labor (Berk 1985; Nickols and Metzen 1982). What tasks should be included? Can one assign the same importance to each household task? How does one 40 approach household tasks that share the same characteristics? Is a task allocation measure a better estimate of participation than a time measure, and how should the frequency dimension of tasks be incorporated in a task measure? These methodological concerns are also applicable to studies of caregiving labor. The discussion in the section which follows includes (1) the way caregiving task measures are constructed, (2) the use of caregiving domains over individual caregiving tasks in data analysis, and (3) the Operationalization of outcome variables. Three indices are constructed to capture the complexity of caregiving behaviors. Range of task participation, or level of task involvement, measures primary caregivers’ degree of participation in tasks for which assistance is required. Two division of labor indices, relative contribution and task exclusivity, capture the relative contribution of primary caregivers in relation to all other caregivers’ effort and the degree to which primary caregivers perform caregiving tasks without the involvement of secondary caregivers. Construction of Caregiving Tasks What constitutes caregiving labor? Should caregiving tasks that share the same characteristics be conceptualized as a single task domain? Family caregiving tasks are different from household chores, which are usually performed on a routine, daily basis. Family caregiving begins when a relative becomes sufficiently impaired (physically or mentally) so as to be unable to perform daily activities. Caregiving can thus be identified by a finite set of tasks in response to an impairment (Dwyer and Coward 1992). Although what constitutes caregiving labor varies from one family to the other, researchers (Albert 1991; Cicirelli 1983) are in consensus about the list of 41 primary assistance activities that are most often mentioned by caregivers. This list includes tasks assumed in response to care recipients’ functional impairments, such as help with bathing, toileting, and moving around the house, tasks related to routine household responsibility, such as cooking, doing laundry, and money management, and tasks related to home health care, such as tube feeding and help with bed sores. In research on gender and caregiving labor, a global measure such as a caregiver’s total hours of care is often used, leaving the type of assistance unspecified (e. g., Enright 1991; Jutras and Veilleux 1991). When the types of assistance provided by caregivers are reported, they are usually examined in a task-by-task analysis (e. g., Stoller 1990; Stoller 1983; Dwyer and Seccombe 1991) or by some arbitrary categorization of tasks that are associated with different gender roles (e. g., Dwyer and Seccombe 1991; Dwyer and Coward 1991; Finley 1989). A task-by-task analysis can be troublesome and tedious in data analysis while the arbitrary categorization of tasks into the same task domain of work that is differently gendered introduces errors that may be responsible for some of the contradictory findings that exist. For example, Dwyer and Seccombe (1991) measured instrumental tasks by combining tasks that are associated with different gender roles: household tasks such as cooking and housework were grouped with tasks that are typically performed by men such as help with moving around the house and money management. Further, personal care tasks were often loosely defined, with personal care tasks in hygiene (e. g., feeding, bathing, dressing) and care in mobility (e. g., assistance with walking, moving in or out of bed) combined into one measure (Dwyer and Seccombe 1991; Finley 1989). 42 In this study, such problems were addressed explicitly. The dependent variables employed are constructed from twenty caregiving tasks (listed in Figure 2.2) that are time—consuming and frequently performed in informal care (Albert 1991; Cicirelli 1983). In most studies, particularly gerontological research, these tasks are commonly grouped into two categories: assistance with care recipients’ functional limitations in activities of daily living (ADLs), which include both personal care in hygiene and mobility help; and assistance with instrumental activities of daily living (IADLs) that include housework and administration of the household (Dwyer and Seccombe 1991; Dwyer and Coward 1991). By contrast, in this study, the twenty tasks for which data were available in all sub-studies are categorized into five task domains: (1) personal care in hygiene (called personal care hereafter); (2)mobility tasks; (3) housework; (4) household administrative tasks; and (5) skilled health care tasks. This categorization of task domains is based on the nature of the task demands involved and on the traditional gender role expectations associated with these caregiving tasks. The conceptualization of task groupings is also consistent with empirical evidence on the gender-based division of labor (Blood and Wolfe 1960; Horowitz 1985a). Personal care, such as bathing and dressing, characterized as nurturing yet "hands-on" tasks, are tedious, time consuming, and often demand round-the-clock attention. It is also a task domain that entails bodily contact and, therefore, is infused with an intimate content between primary caregivers and care recipients.9 Housework ”In contrast to most of the gerontological research, this study excludes eating, commonly included as part of personal care tasks, from analysis. Since eating does not place much intimate demand on caregivers for bodily contact, omitting it allows me to simplify the analysis of this dual-nature task domain. 43 Figure 2.2 List of Caregiving Tasks Total Twenty Caregiving Tasks administration of catherization (a) attending to incontinence with respect to stool attending to incontinence with respect to urination arranging transportation ' bathing ' cooking dressing eating (b) grooming help with bed sores (a) housework maintenance laundry lifting and turning in bed money management moving around the house moving in and out of bed shopping (c) toileting tube feeding (a) walking Task Domains Personal care (6) attending to incontinence with respect to urination attending to incontinence with respect to stool bathing dressing grooming toileting Housework (3) cooking housework maintenance laundry Administration of household (2) arranging transportation money management Mobility tasks (4) lifting and tuming in bed moving around the house moving in and out of bed walking Note: (a) Three health care tasks are dropped from task domain analyses due to the small number of aregivers who cared for the relative with these needs. (b) Shapping is dropped from housework task domain because of its ambiguious meaning shown in preliminary factor analyses. (c) Eating is dropped from personal care task domain because of analysis strategic concern. That is, since eating lacks intimacy content. dropping it from analysis simplies analysis of this dual-nature task domain characterized by both intimacy and labor intensity. 44 such as cooking and doing laundry share some of the same attributes as personal care tasks.10 Both are repetitive, tedious, and are done on a routine basis. The fundamental difference between personal care and housework lies in the amount of control over scheduling that each allows, aside from the difference in the demand of intimacy. Personal care requires more immediate assistance and physical presence as help is needed. In contrast, housework need not to be done at a certain time or place, thereby allowing caregivers more control over scheduling. Time availability, therefore, is more of a concern in personal care than it is in housework. Household administrative tasks, including transportation and money management, are intermittent tasks which involve making decisions and taking control. Mobility tasks such as walking and moving around the house require physical strength from caregivers and may best be categorized as a masculine job. Health care tasks such as administration of catheterization and tube feeding are dual-nature tasks which involve both nursing attention as well as medical technique. Since the number of cases in need of this care is small, focus on the health care domain is not feasible. It is therefore dropped from task domain analysis, although it will be retained in the global analysis because it represents the administration and technical skill aspect of caregiving chores. 10One of the housework tasks, shopping, was dropped from the task domain analysis. A factor analysis of caregivers’ relative involvement in four housework tasks showed that the factor loading for shopping was consistently low for the total sample and four caregiver subgroups. A factor analysis of caregiver’s relative involvement in housework tasks and administration tasks also showed that the factor loadings of the shopping item were high for both factors of housework and household administration, suggesting the ambiguity of this task in the division of labor measures. 45 All these task domains represent very different aspects of caregiving and are likely to produce different gender-based divisions of labor. Setting aside its intimate nature, personal care tasks, along with household chores, are an extension of women’s nurturing role, to which female caregivers are more likely committed than male caregivers. Mobility tasks and administrative tasks require either physical strength or "skill," which (especially older) women may be physically incapable or not be socialized to take on. A task-centered definition of caregiving labor and a subsequent categorization of caregiving tasks have many advantages. First, care is usually delivered only to people who become functionally dependent. Thus, the chores caregivers perform in response to functional dependency vary according to a care recipient’s needs. A task—centered analysis takes such variations into account. Second, a task-oriented definition of caregiving behavior highlights the labor-intensity of informal caregiving and addresses the need for revaluation of caregivers’ work at home (Abel 1990a). Third, not all caregiving tasks are alike in their demand for physical, emotional, and time commitments. There is a need to conceptualize caregiving tasks not just as a single domain but as one that comprises several dimensions with different expectations as to who should perform them. Fourth, a categorization of specific tasks along traditional gender lines allows empirical tests as to whether a gendered nature of caregiving behaviors exists. By analyzing the chores caregivers perform, one highlights the labor-intensity of informal caregiving. Nevertheless, as argued by Abel (1990a), caregiving is also a subjective experience embedded in intimate personal relationships. The task—oriented 46 approach of caregiving labor cannot address subjective experiences such as emotional attachment as labor is rendered or the stress and burden caregivers undergo, which are aspects of caregiving to which women are more susceptible than are men (Fitting et al. 1986; Kessler and Mcleod 1984). Therefore, it is acknowledged that the quantification of caregiving chores used in this study is insufficient to understand the encompassing nature of family caregiving and often underestimates women’s endeavors in caregiving. Operationalization of the Range of Task Involvement and the Division of Labor To measure a caregiver’s degree of involvement in tasks, primary caregivers were first asked if their relatives needed help with each of the twenty tasks in question. After the needs of care recipients were identified, primary caregivers were then asked to report how often they and all other caregivers as a group, including formal and informal helpers, provided assistance in each of the applicable tasks in a two-week period. Frequencies of involvement were recorded at an ordinal level: no involvement (0), once a week (1), several times a week (2), once a day (3), and several times a day (4). The scoring is converted into a quasi-interval measure by assigning a value of 0 to no involvement, 1 to once a week, 3.5 to several times a week, 7 to once a day, and 14 to several times a day. But how should one measure the degree of the task participation of primary caregivers and the division of caregiving labor between primary and secondary caregivers? What is the advantage of using a task measure over a time measure? How should the frequency dimension of tasks be incorporated in a task measure to reflect the relative importance and the routine of caregiving tasks? 47 In analyses of the division of labor, it should be noted that the allocation of caregiving labor to specific tasks is, analytically, distinct from the allocation of overall time to caregiving. Allocation of tasks, i.e., who does what and how often, indicates the degree to which tasks are shared or performed by separate providers. Allocation of time, on the other hand, concerns the hours of involvement and the degree of participation measured in time. As distinct as they are, however, what tasks are assumed and how much time is allocated are not independent from each other because time is often inherent in tasks as they are allocated to caregivers. As put by Berk (1985), task is the conceptual unit through which work is initially defined and allocated while time is the ultimate output of that division by task. While time spent in caregiving is logically related to the types of tasks performed, little is known about the inherent differences (or similarities) in the time demand of specific caregiving tasks. In this study, primary caregivers were asked to indicate only the frequency of tasks in which they were involved, not the hours of care they provided for each task. The data thus are inadequate to address the association between a time measure and a task measure or the departure of one from the other. Hence, the operationalization of primary caregivers’ involvement and share in the caregiving process is based on a task allocation measure and the frequency of the tasks being performed. To operationalize the division of labor in terms of a frequency weighted task allocation measure, as outlined in detail below, has two advantages. First, the frequency with which a task is performed in a certain period of time is, to some extent, a close proxy for time spent in performing a task (Warner 1986; Model 1981). 48 Second, the frequency approach, as Berk (1985, p.43) puts it, "reflects the routine, ordinary, and day-to-day efforts contributed to household labor." Compare two caregivers: one reported helping with dressing twice a day and took only 15 minutes each time; the other shopped once a week, a task which took 4 hours. The total effort from the first caregiver, 3.5 hours per week, is less than that of the second one. Yet when the routine, time fragmentation, and day-to-day effort are taken into account, the first caregiver’s endeavor, providing assistance twice a day on a daily base, should be considered greater than that of the second one, who provides assistance intermittently. To grasp the routine, repetitive dimension of caregiving tasks, as argued by Berk (1985), each time a task is done it should be counted as a unique task. This frequency-weighted approach to household labor, however, is not without problems. As explained by Berk (1985 p.43), "it is impossible to distinguish between the infrequent accomplishment of many tasks and the frequent accomplishment of only a few tasks. " With this frequency approach, the contribution of a household member routinely involved in a single task may be greater than the contribution of a household member who performs more tasks but less often. Nevertheless, as far as the routine and time fragmentization of caregiving chores is concerned, the contribution of a caregiver who undertook only a few tasks on a daily basis should not be considered less than that of a caregiver who took on more tasks that were accomplished once every other week. Gender differences in the task assistance of primary caregivers can be reflected in many ways: in the number of tasks primary caregivers assume; the degree to which task performance adheres to traditional role expectations; the degree to which tasks 49 are performed without the involvement of secondary caregivers; and the relative contribution of primary caregivers in relation to that of secondary caregivers. Three measures of caregiving behaviors --range of task participation, or level of task involvement, task exclusivity of primary caregivers, and primary caregivers’ relative contribution-- are used in this study to explore the complexity of caregiving and the role a primary caregiver’s gender and family relationship to the care recipient plays. Range of task participation is measured by the number of tasks primary caregivers participated in at least once during a two-week period, divided by the number of tasks applicable to the care of the functionally dependent relative. The percentage score has a range of 0 through 100. A score of 100 indicates that primary caregivers took part in all the tasks for which their relatives needed help while a score of 0 would indicate that primary caregivers turned all the tasks over to others. This global measure of range of assistance for all possible tasks as a result of the patients’ need, however, says little about role stereotypic task participation. Men and women may provide help with the same number and proportion of tasks, but in very different task domains. To examine the degree to which primary caregivers adhere to traditional gender roles, a task domain analysis of range of task participation will be used. If adherence to traditional gender roles exists, female caregivers will assume tasks in a women’s sphere while male caregivers will participate in tasks defined as part of a man’s domain. In other words, scores for the range of task participation in personal care and household work will be high among female caregivers while those in mobility and administrative tasks will be high among male caregivers. 50 Range of task participation, sometimes used in the caregiving literature to demonstrate gender differences in care providing behavior, however, does not indicate how tasks are allocated between primary and other caregivers and to what extent task allocation is based on primary caregivers’ gender and family relation. Primary caregivers may very well take on the same number and types of tasks, but their assistance in relation to what others provide may be at a totally different level. An analysis of the division of labor between primary caregivers and all other caregivers captures another dimension of caregiving behaviors. Two indices of the division of caregiving labor could be used. One is an index of relative contribution of primary caregivers, which measures primary caregivers’ proportionate volume of care. The other is a measure of task exclusivity of caregiving labor, a gauge of the degree to which tasks are undertaken by primary caregivers without the involvement of other helpers. These two indices are conceptually different but highly correlated. In early family household labor studies (e. g., Blood and Wolfe 1966, Clark, Nye, and Gecas 1978), and even in recent ones (Baxter 1992), five-point Likert scales of relative contribution were frequently used to measure the household division of labor. Respondents (wife or husband) were asked to indicate, on a five-point scale, their relative responsibility for particular tasks undertaken in the household. The wife’s relative contribution to housework was computed according to Formula 1. Formula 1: OZ! wife contribution l: 2'! maximum possible score I: i=applicable tasks. Contribution in 5-point Likert scale items coded 4=wife always does the work always, 3 =wife more, 2=equal, l=husband more, 0=husband always. 51 To address the deficiency of this method which assigns equal weights to tasks that vary in frequency and time, recent research weights individual tasks by frequency or hours of task performance based either on respondents’ report of task frequency or on time estimates from time budget studies of domestic work (Model 1981; Kamo 1988).11 The frequency-weighted relative contribution of labor is usually approached with the following formula (Formula 2) in which frequency is measured at an ordinal level (Warner 1986). Fgrmula 2: LJ. 1 wife contribution by relative scale X frequency 1: if 1 frequency of task 1 = i=applicable tasks. Relative contribution scale is coded as 1=wife always, .75=wife more, .5=equal, .25=husband more, 0=husband always. Frequency is coded at ordinal level as 4=dai1y, 3=4—6 times a week, 2:2-3 times a week, l=once a week, 0=task not done. The division of labor index constructed for this study is based on this second formula with a slight modification. Because data are available on the actual frequency of task accomplishment for both primary and secondary caregivers, this study employs a division of labor measure incorporating the actual frequency with which each task is performed rather than a relative contribution scale and ordinal-level frequency as used in Formula 2. The index of relative contribution used in this study is defined in Formula 3. “The problem associated with the use of time estimates from time budget studies has been discussed in time budget and family labor research (Andorka 1987). It is even a bigger problem in caregiving research when one uses the estimates of the average of hours typically allocated to caregiving tasks as weights for all caregiving households because hours spent on caregiving chores vary from one household to another depending on caregivers’ ages, social class, gender, family relation, and most of all care recipients’ functional dependency. 52 Formula 3: 2], frequency of task done by the primary caregiver Index of . . [=1 Relative Contribution: X 100 )5 1 frequency of task undertaken by all care providers 1: i=applicable task for all twenty tasks or in a given task domain. Frequency coded as O=no involvement, l=once a week, 3.5=several times a week, 7=once a day, 14=several times a day. This frequency-weighted relative contribution index, more precisely a relative frequency proportion index, is defined as the total frequency of tasks the primary caregiver accomplished divided by the total frequency of all tasks undertaken by all care providers for a care recipient. It represents the relative contribution effort in tasks, a measure of the extent to which caregiving tasks are divided between primary caregivers and all other carers. The closer the index approaches either 100 or 0, the more the labor is unequally divided, with primary caregivers performing either the majority or the minority of the labor. The index in the third formula is essentially the same as the weighted relative contribution index in the second formula and they are highly correlated with each other (correlation coefficients ranging between 0.9 and 0.95). But, the frequency-weighted approach of task accomplishment more directly captures the routine, fragmented, and day-to-day efforts of caregiving labor than the index in Formula 1 or Formula 2, as the frequency approach counts a task done each time as a unique task. The second measure of the division of labor employed in this study is an index of the exclusivity of a primary caregiver’s efforts, that is, the degree of tasks accomplished exclusively by primary caregivers, or accomplished without the involvement of secondary caregivers. It is defined as follows. 53 Formula 4: j frequency of task done exclusively by the primary caregiver j: Task Exclusivity index= . X 100 I! frequency of the task undertaken by all care providers i= i=applicable task for all twenty tasks or in a given task domain. Frequency coded as 0=no involvement, l=once a week, 3.5=several times a week, 7=once a day, 14=severa1 times a day. Task exclusivity of primary caregivers, also using a frequency-weighted approach, measures the degree to which caregiving tasks are accomplished without any help. The index score ranges from 0 to 100. An index score of 100 indicates primary caregivers did all the tasks within a given task domain without any unpaid and/or paid help. An index score of 0 indicates that all tasks undertaken for the care of the elderly are shared by primary and secondary caregivers, or that primary caregivers completely turn over the responsibility in a given task domain to secondary caregivers. Personal care and housework, perceived to fall within women’s domain, are likely to be performed exclusively by female primary caregivers while male primary caregivers will tend to share these tasks with others. The obverse will be observed for mobility and administrative tasks, activities that fall within men’s domain. The results of the reliability analysis of internal consistency of the three outcome measures are presented in Table 2.2. The reliability analysis tests the extent to which a caregiver’s task participation, relative contribution, or task exclusivity, varies simultaneously across all the tasks that are part of the scale/domain. A high Cronbach’s alpha would confirm the appropriateness of the categorization of tasks and a group of tasks as a one-dimensional task domain. But it should be noted that the reliability test is conducted on caregivers who provided care to relatives who were 54 Table 2.2 Reliablity of Task Participation, Relative Contribution and Task Exclusivity for Each Task Domain, by Relationship of Caregiver to Care Recipient Task Participation, Alpha ASK DOMAIN car:g"lvers Wife Husband Daugrhter Son Personal care 0.77 0.78 0.85 0.70 n.a. Housework 0.57 0.28 0.78 0.62 0.62 dministration of household 0.32 0.38 n.a. 0.41 n.a. Mobility 0.78 0.49 0.96 0.87 n.a. Relative Contribution, Alpha TASK DOMAIN cargguivers Wife Husband Daughter Son Personal care 0.93 0.94 0.95 0.90 0.97 Housework 0.79 0.71 0.77 0.80 0.81 dministration of household 0.57 0.52 0.72 0.55 0.59 Mobility tasks 0.94 0.90 0.96 0.95 0.96" Task Exclusivity, Alpha TASK DOMAINS careAgljlvers Wife Husband Da_ugfhter Son Personal care 0.92 0.93 0.95 0.88 1.00 Housework 0.81 0.82 0.78 0.78 0.84 dministration of household 0.70 0.72 0.78 0.62 0.70 Mobility tasks 0.94 0.96 0.91 0.93 1.00 Note: n.a. if all caregivers participated all tasks. Table 2.3 Number of Cases Applicable in Each Task Domain“ A MAI in caregivers Wife Husband Daughter Son Number of Cases in the Sample 1387 573 288 462 54 Personal care (6) 1011 416 192 389 34 (151) (54) (36) (56) (5) Housework (3) 1254 488 267 446 53 (957) (369) (208) (344) (36) dministration of household (2) 1179 472 222 434 51 (794) (313) (141) (307) (33) Mobility tasks (4) 621 228 129 242 22 (177) (57) (29) (84) (7) ‘No. of cases with at least one functional dependency in the task domain Note: The number in parentheses is the valid number of cases in the reliablility analysis (Table 2.2). i.e., the number of patients in need of help with all tasks in the task domain. 55 functionally dependent in all tasks within a given task domain. That is, not every care recipient had the same functional need and only caregivers whose relatives were dependent in all the tasks of a given task domain were included in the task-domain reliability test. The results may not be consistent with those based on caregivers whose relatives needed help with only one or two tasks within a given task domain. The number of valid cases used to construct the three outcome variables for each task domain is presented in Table 2.3. The number of cases used in the index’s reliability analysis is presented in parentheses. The data in Table 2.2 show that the alphas for task participation in personal care and mobility tasks were as high as .75, suggesting that if primary caregivers participated in one task during a two-week period, they more than likely assumed responsibility for other tasks in the same task domain. Alphas were also high for task exclusivity and relative contribution in all task domains, with the exception of administrative tasks, suggesting a high internal consistency of task allocation within a task domain. The unsatisfactory internal consistency for administrative tasks, particularly for the task participation index, is largely a result of having only two items in the composite scale of household administration (Carmines and Zeller 1979). The appropriateness of the categorization of tasks can be further tested by performing a reliability analysis for wives, husbands, daughters, and sons. The results show that the high alphas for the total sample in the task domains of personal care, housework, and mobility tasks also are consistently high for individual sub-groups. The absence of group variation in the scale’s reliability suggests that caregiving tasks in the same domain tend to vary simultaneously in the same direction, i.e., high in 56 one task if high in another task within the same domain, regardless of family relation or gender. The substantial group variations for the range of task participation in the housework domain and for the two indices of division of labor in the administrative task domain are probably a result of sampling fluctuation and the fact that fewer than three items are included in the domains. It should be noted here that the reliability analysis suggests not the proper use of task domain analysis but rather the internal consistency of the caregiving task groupings that are constructed in this study. In contrast to domestic labor, caregiving chores are unlikely to include a common set of tasks performed on a regular basis within a household.” Factor analyses therefore cannot be performed to confirm the conceptual groupings and the need of using task domain analyses to separate personal care tasks from mobility tasks and household work from administrative tasks. A final note is necessary about the outcome variables, to be detailed in Chapter 3. All three outcome variables are highly skewed and responses fall heavily and unevenly on the endpoints. Responses within the range of task participation, while on a continuum, will thus be dichotomously categorized: (1) those who participated in all the tasks in a given task domain, or full-scale task involvement; and (2) those who took on some or none of the tasks in a given task domain. Responses within the relative contribution index will be collapsed into three categories at an ordinal level: ( 1) primary caregivers performed all (100%) the chores, or exclusive care by primary caregivers, (2) primary caregivers performed more than half (50%-99%) of caregiving l2Although the twenty tasks used in this study are most often mentioned by caregivers, care recipients are unlikely to need help with all twenty tasks. 57 chores, a complementary pattern of primary caregivers’ assistance in which primary caregivers’ assistance was more than that of secondary caregivers, and (3) primary caregivers performed less than half (0%-49%) of the chores as compared to secondary caregivers, a supplementary pattern of primary caregivers’ assistance in which the amount of help provided by secondary caregivers was more than that provided by primary caregivers. Responses within the task exclusivity index are collapsed into three categories: (1) exclusive care by primary caregivers, same as the first category of the relative contribution index, (2) partial task sharing (1 %-99%), that is, primary caregivers were involved in at least one task in a given task domain without secondary caregivers’ assistance, and (3) total dependence on secondary caregivers, that is, total task sharing or exclusive care of the secondary caregivers wherein primary caregivers either shared every single task in a given task domain with, or completely turned the chores over to, secondary caregivers. The grouping of two seemingly odd categories, total task sharing and exclusive care of secondary caregivers, points to the extent of primary caregivers’ reliance on secondary caregivers to meet the needs of the elderly as a result of disability. A separate category for the exclusive care of secondary caregivers is desirable but unfeasible in this study because primary caregivers were the only sample in the study and were generally highly involved in the tasks with which care recipients needed assistance. Only a small percentage of the sample, four percent or less across caregiver groups, had secondary caregivers assume total caregiving responsibility for tasks in some specific task domains (see Table 3.7).[3 13The one exception was in personal care, in which 5.1 percent of daughters and 17.6 percent of sons completely turned over the caregiving responsibilities to other 58 Major Predictor Variables Six categories of predictor variables are considered in this study: (1) power and resources possessed by primary caregivers; (2) time availability of primary caregivers; (3) gender ideology held by primary caregivers; (4) characteristics of primary caregivers; (5) caregiving demands and situation; and (6) characteristics of care recipients. The concepts and empirical evidence for the first three have been discussed in previous chapter and this section explains how they are operationalized as measures. Other variables of importance that have been suggested in family caregiving research will also be discussed and included as control variables. Power/resources Two variables are used to measure power/resources: (1) primary caregivers’ adjusted household income; and (2) primary caregivers’ occupation. Adjusted Household income. Primary caregivers’ household income is measured at an interval level and expressed in constant 1991 figures. The use of household income rather than an individual’s personal income is a result of the limitations of the data. Thus, this measurement is based on the assumption that caregivers had access to the income of other family members with whom they resided, i.e., adult children had access to the income of parent(s) and husbands or wives had access to the incomes their spouses earned.“ caregivers. l"More discussion of the problem of using household income as a proxy for power, the issue of absolute power vs. relative power, and the conceptualization of power as the ability to mobilize resource vs. the ability to negotiate in a family dynamic, will be presented in Chapter 4. 59 Primary caregivers’ household income is adjusted by the number of persons residing in the caregivers’ household, using the Social Security Administration’s scaling factor (U.S. Bureau of Census 1993). The household income of a large-size family will be weighted down more than that of a small-size family because, with the same amount of household income, a large-size family would have fewer resources to mobilize or less power to negotiate than a smaller—size family. @cupation. Caregivers’ current occupation, or occupation before retirement if applicable, is divided into five categories: ( 1) professional, managerial, administrative; or technical; (2) sales and clerical; (3) labor and operatives; (4) unclassified occupation; and (5) never employed, thus without an occupation. Four dummy variables are accordingly created, with professionals as the reference group. Time availability Time availability is measured by three variables: work schedule conflict; competing family responsibility; and propinquity. Work schedule conflict. To measure the time demands on a caregiver and the degree of conflict between a caregiver’s work activities and caregiving, primary caregivers’ employment status (coded 1=full time and O=other) will be used. Preliminary findings showed that respondents employed part-time and those who were not working did not differ in their caregiving behaviors, probably because part-time employment already reflects an adjustment to a caregiving situation and is therefore no different from non-employment in respect to primary caregivers’ task participation and the division of labor (see also Boaz and Muller 1992). Based on this preliminary finding, no distinction is made between caregivers who worked part-time and those 60 who were not employed. Competing family responsibility. This concept is measured by the presence or absence of young children under 15 in a caregiver’s household, although an adult child’s marital status is a more common measure in family caregiving research.15 Although this study uses a somewhat high cut-off point for the age of children, domestic labor studies have demonstrated that school—age children can be just as time demanding as pre-school children (Coverman 1985). Sixteen is the age minors can obtain a driver license, thereby becoming less time demanding especially with respect to transportation needs. This variable is applicable to adult children caregivers only. Propinquity. Propinquity, or the accessibility of help, is another dimension of time availability in the context of caregiving. Primary caregivers who live with the elderly are more accessible to care for a recipient’s need than are caregivers who live apart from them. The impact of propinquity especially holds true for hands-on tasks and obligatory routine chores that require a caregiver’s presence. Empirical evidence has suggested that geographical proximity is a primary determinant of parental care (Litwak and Kulis 1987). Co-resident children provide greater amounts of personal 15Some researchers (Horowitz 1985b) suggest that an adult child’s spouse is an important source of help while others (Litvin et al. 1995; Stoller 1983; Lang and Brody 1983) imply that a spouse is a burden and a source of competing family responsibility that constrains a caregiver’s ability to care for a parent and decreases the number of hours an adult child can devote to parent care. Preliminary analysis in this study showed that the presence of a spouse had little to do with adult children’s task participation and the division of labor; thus it was dropped from the measures of time availability. The weak explanatory power of the presence of a caregiver’s spouse may be related to the contradictory and confounding nature of this variable. The presence of a spouse may be a double burden for younger caregivers but not for older caregivers who have reached a life cycle stage where there is less demand to assume household responsibilities and more role sharing between spouses, thus a source of help (Albrecht, Bahr, and Chadwick 1979). 61 care than do children who live elsewhere (Soldo and Myllyluoma 1983). Propinquity is a dichotomous variable coded 1 as living together and O as living apart. The measure of propinquity is applicable only to adult children caregivers as all spouse caregivers lived with the care recipients. Gender Ideology Educational level is used as a proxy-indicator of a primary caregiver’s beliefs about gender. It is hypothesized that primary caregivers who are less educated are more likely than educated ones to adhere to traditional gender role ideology, to see caregiving as women’s responsibility and, as a result, to conform to gender- stereotypic roles. The lack of a direct measure of gender ideology and the use of education as a proxy indicator has significant drawbacks, mainly arising from the inconclusive role education plays in shaping an individual’s gender role beliefs. Some researchers (Farkan 1976; Ross 1988) perceive education as a mechanism that transmits to people a set of liberal values which emphasize equality and encourage egalitarian gender values and behaviors. It is believed that higher education engenders beliefs in gender equality and, therefore, encourages an egalitarian division of labor. Educated people are thus more likely to embrace egalitarian gender roles than are those who are less educated. Other researchers (Huber and Spitze 1983) consider education to be an indicator of power and argue that educated persons have more resources and power to minimize their involvement than do less educated ones. Family labor research usually does not support education being conceptualized as power, at least on the part of husbands 62 (Ross 1987; Hardesty and Bokemeier 1989). When comparing earning and education as measures of power, economic earning power usually decreases husbands’ contribution to housework while higher education of husbands usually promotes an egalitarian division of labor. The counter effects of income and education throw into doubt the efficacy of education as a measure of power, and, as explained by Hardesty and Bokemeier (1989), are probably due to the fact that education, though usually raising earnings, imparts values that influence the division of labor in ways that are opposite to earnings. Although it is acknowledged that education has an effect that is over and above gender role attitudes (Ross 1988), in this study, education is assumed to promote a value of gender equality that encourages an egalitarian division of labor. Characteristics of Primary Caregivers Four characteristics of primary caregivers are considered in this study: (1) gender; (2) family relationship to the care recipient, (3) health condition; and (4) age. we; is coded 1 for female and 0 for male. Here the variable is used to delineate gender differences in caregiving behaviors and to understand how this crucial social category affects an individual’s group placement in systems of inequality such as in the case of caregiving. Family Relationship. Since more than 95 percent of the caregivers in the sub- studies from which the data are drawn are either spouses or adult children, only those two relationships are considered in the analysis. Primary caregivers’ relationship to the frail relative is coded 1 for spouses and O for adult children. Mm. The health condition of primary caregivers is directly related to their ability to help frail relatives. A deteriorating health condition restricts caregivers from 63 participating in some chores and imposes a need to call upon others to share responsibilities. Health is measured as self-reported on a four-point scale, from excellent(= 1) to poor (=4). Age. Caregivers’ age, measured in years, can also be a proxy for health. As the risk of poor health usually increases with age, age can be used as a supplementary variable to measure the effect caregivers’ health may have on their ability to help frail relatives (Enright 1991). Caregiving Demands and Situation The caregiving situations in which primary caregivers are placed, affect, to a great extent, how they respond and provide help to an elderly relative. Four aspects of caregiving circumstances and environment are considered: (1) available helpers; (2) need level of the elderly or the demands of care; (3) diagnosis of the relative for whom care is provided; and (4) duration of care. Available helpers is measured as the number of people over 18, other than the primary caregiver, living in the same household with the functionally impaired relative. A primary caregiver without immediately available helpers will be more likely to experience a caregiving burden in respect to the division of labor than one with available helpers. Additional adults co-residing in the elder’s household have been found to be an immediate source of support and to lessen the demands on primary caregivers, particularly child caregivers (Noelker and Wallace 1985 ; Stommel and Kingry 1991; Tennstedt, McKinlay and Sullivan 1989). Need level of the elderly is defined as the number of caregiving tasks for which an elderly relative needs assistance. Level of need measures the extent of the demands 64 of care put on primary caregivers as a result of the functional dependencies of their relatives. Past research has suggested that as functional dependency increases so do the hours of care and the scope of assistance (Stoller 1983; Dwyer and Coward 1991; Dwyer and Seccombe 1991; Stoller and Pugliesi 1991). Studies of helping networks and the division of labor have also found that a positive relationship exists between the functional capacity of the elderly and the size and scope of helping networks, suggesting that as care recipients’ functional needs increase, primary caregivers are less likely to be the sole provider and more likely to receive assistance than those who care for less disabled relatives (Stoller and Earl 1983; Stoller and Pugliesi 1988; Stommel, et al. 1995). Diagnosis is coded as a dummy variable, with Alzheimer’s as 1 and cancer and other physical impairments as 0. Although the sociodemographic characteristics of caregivers and recipients in the cancer study were distinctly different from those in other sub-studies (CGl,CGZ,ALZ), the preliminary data analysis revealed that the organization of assistance and the division of labor showed a distinct pattern only for Alzheimer’s patients when compared to other diagnoses such as cancer or physical impairment. Hence, cancer is combined with other physical impairments as a single group of diagnoses. The fact that Alzheimer’s rather than cancer patients differ from all other diagnoses is probably due to the fact that the organization of help is not so much driven by crisis as by the availability of formal organizations with which physically impaired and cancer patients tend to be provided following hospital discharge, a support system Alzheimer patients usually do not experience (Stommel, Collins and Given 1994). 65 Duration of care. Since the actual length of time for which primary caregivers had been caregiving was not asked in the later homecare study of new caregivers (CGZ), to be prudent, duration of care is measured as a dichotomy, rather than at a continuous level, coded 1 for more than two years in duration of caregiving, or long- term caregivers, and O for two years or less, or new caregivers. The cross-sectional study of the time factor (long—term caregivers vs. short—term caregivers) examines how the length of time a caregiving role has been performed is related to the range of task participation and the division of labor, and how the effect of time on care providing behaviors differs for women and men and for spouses and adult children. Thus far research has provided few guidelines about the cut-off point which differentiates new caregivers from long-term caregivers. Some (Albert 1991) define new caregivers as those who provide care one year or less while others (Stone, Cafferata, and Sangl 1987; Enright 1991) consider long-term caregivers to be those who provide care for more than five years. The use of two years as a demarcation point in this study is purely a statistical decision. Preliminary findings suggest that duration of care for a two-year period is better than that of a one-year or five-year period to differentiate new caregivers from long-term caregivers and to predict task participation and the division of labor. Care Recipient’s Characteristics The relevance of two care recipients’ characteristics, gender and marital status, to adult children’s caregiving behaviors has been suggested in the caregiving literature (Dwyer and Coward 1991). 66 Gender of the elder is coded 1 if female and 0 if male. The importance of the gender of the elder in the selection of primary caregivers has been mentioned in caregiving research by Lee, Dwyer and Coward (1993) and Stoller (1990), who suggest that the elderly tend to name caregivers of the same gender. It is unknown if the care provided by primary caregivers will vary on the basis of the gender of the parent. Further, little is known about the role the gender of the elder plays in the type and nature of the tasks caregivers perform. As implied in the cultural taboo hypothesis proposed by Montgomery and Kamo (1989), daughters are likely to be more comfortable helping mothers than fathers with personal care tasks such as bathing and toileting while the reverse is the case for sons. Marital status of the elder is coded 1 if the parent is married and 0 if the parent does not have a spouse resident in the household. Stoller (1983) has suggested that the presence of a spouse is negatively related to adult children caregivers’ level of involvement. Although the selection of children over a spouse as primary caregiver usually suggests that the health of the spouse is deteriorating, an infirm living spouse, though not the primary caregiver, can still be a major source of assistance, probably more so for tasks that require intimacy than for other tasks. Data Analysis The data analysis will be performed in three steps in order to identify the relationship of gender and family relation to task participation and the division of labor and to delineate the group-specific combination of factors that contribute to how men and women provide care. 67 The first step examines how the background characteristics of primary caregivers, the demands of caregiving, and caregiving situations are themselves related to the gender and family relation of care providers. The purpose here is to estimate the extent to which these characteristics, when later entered as predictors, are themselves associated with the gender and family relation of care providers. The second step involves specification and testing of models about the effect of variables common to both spouses and adult children and variables applied to adult children’s caregiving context on their task involvement and division of labor. The purpose of the analysis is to identify gender as well as relational effects in caregiver behaviors. That is, it seeks to determine if male caregivers are different from female caregivers and spouses from adult children in their caregiving behaviors after all other effects are partialed out. In addition, since duration of care emerged as a significant determinant of caregivers’ behaviors in the preliminary data analysis, family relation is allowed to interact not only with gender but also with duration of care to see (1) if the effects of gender are the same among child caregivers as among spouse caregivers; and (2) if the effects of family relation are the same among short—term caregivers as among long-term caregivers. Another interaction term, gender by duration of care, is removed from the model specification because, in the preliminary data analysis, the term was statistically insignificant for all three indices and for all task domains. That is, the effects of gender do not differ between short-term and long-term caregivers; whatever gendered patterns are found among short-term caregivers are likely found among long-term caregivers. 68 This step also tests the effects of gender-role ideology, time availability, and power/resource, assuming that they are the same across caregiver groups. Nevertheless, one problem associated with this assumption, as one can immediately see, is the missing specification of the differential effect of role ideology by gender that is well documented in domestic labor research (Baxter 1992; Ross 1987; Seccome 1986; Huber and Spitze 1983). In contrast to the gendered effect of power/resources or time availability, that differs only in magnitude, the effect of role ideology differ in direction for men and women. That is, in domains traditionally defined as women’s responsibility, men, especially the young cohort of men exposed to social changes in gender relations since the 1960s, should take on more tasks if they hold an egalitarian ideology while women who voice an egalitarian viewpoint should do less; the direction of the relationship should be opposite in domains that are extensions of men’s gender roles. The lack of specification of the differential effects of role ideology by gender in a model that contains both male and female samples is thus problematic. Because of data limitations, specifically the small number of son caregivers, it is not possible to include this interaction effect in the final model. In addition, in a model that specifies interaction effects, coefficients for the dummy variables are no longer the estimates of the average effects of certain variables for the total sample but the estimates for certain subgroups, depending on which subgroup constitutes the baseline group (Hardy 1993). The more interaction terms one considers, the more complicated, the calculation and table presentation become. To remedy the situation, this study, first, presents a model without specification of the interaction effect of education by gender, acknowledging that any statements regarding 69 the gender role ideology hypothesis may be ambiguous in wording as the statements may not distinguish the experience of women from that of men. Then, the interaction effect of education by gender will be added in the model in a separate run; results regarding the gender-by—education effect will be stated in the text, although figures are not presented unless notable changes surface, to add to our knowledge of the hypotheses of gender ideology. The third step involves preliminary analyses of the role power/resources, time, and ideology play in the provision of care by women and men, respectively. Limitations in the use of a separate equation model approach to identify the mechanism that influences caregiving behaviors for each of the caregiver groups are acknowledged. That is, limited conclusions can be drawn from the separate runs about how the effects of predictors for one group are different from another since estimated parameters and significance tests associated with the predictors, drawn from separate runs, can be compared within, but not across, caregiver subgroups. Finally, a note about the main statistical technique used in this study. Since all three outcome variables are highly skewed and responses fall heavily and unevenly on the endpoints, the normality assumptions of ordinary least square regression is not met. Responses on a continuum are thus collapsed into dichotomous or polytomous variables and logit modeling is used to test the effects of variables, mixed with continuous and categorical response, on the odds of an event occurrence (Hosmer and Lemeshow 1989; Demaris 1992). With respect to the index of range of task participation, a binary logistic regression is used to predict the likelihood of primary caregivers being involved in all 70 tasks in a given task domain versus some or none of the tasks. With respect to the indices of task exclusivity and relative contribution, a polytomous logistic regression rather than an otherwise more appropriate ordinal logistic regression is used because the proportional odds assumption of ordinal logistic regression was often violated in the preliminary data analyses that used ordinal logistic modeling (Demaris 1992). In the multinomial logistic regression analyses of the task exclusivity index, with total dependence (i.e., either sharing every single task in a given task domain with, or completely turning the chores over to, secondary caregivers) as the contrast group, two sets of odds ratios are predicted: (1) the odds of primary caregivers being an exclusive care provider versus total dependence on secondary caregivers’ assistance and (2) the odds of primary caregivers being a partial task-sharing care provider (i.e., assisting in at least one task without the involvement of secondary caregivers) versus total dependence on other helpers. With regard to the relative contribution index, with the supplementary care (i.e., primary caregivers providing less than half of the total care) as the contrast category, two sets of odds ratios are formed: (1) the odds of primary caregivers providing exclusive care versus supplementary care, and (2) the odds of primary caregivers providing complementary care versus supplementary care. Summary This study aims to examine, first, the roles the gender and family relationship of primary caregivers play in the range of task participation and the division of labor as a result of an elderly relative’s disability, and second, the mechanisms that are involved in producing how men and women provide care. Secondary data analyses were performed on a sub-sample of primary caregivers drawn from six panel studies 71 conducted between 1986 and 1991 in central Michigan. This chapter provided the rationale for using task domain analyses rather than task—by-task analyses and the rationale for the way the tasks are categorized. Also discussed was the operationalization of predictors and three outcome variables: task involvement level, task exclusivity and relative contribution. Binary and multinomial logit models are used to test the hypotheses outlined in Chapter 1. CHAPTER 3 DESCRIPTIVES OF THE SAMPLE AND CAREGIVING PATTERNS The family is the primary source of long-term care assistance for the frail elderly (Dwyer and Coward 1992). The majority of the disabled elderly live in their communities (U.S. Senate, Special Committee on Aging, 1987) and receive most of their support from family and friends (Scanlon 1988). As noted in Chapter 1, the gendered nature of caregiving is reflected, first, in the predominance of women as primary caregivers and, second, in the inequitable gender division of labor, especially among child caregivers. While the fact that wives make up a disproportion of spouse caregivers may be attributed to demographic reasons (i.e., differential longevity of husbands and wives), the fact that daughters constitute a large majority of child caregivers reveals, not the innateness of nurturance in women, but rather the gendered nature of the social, demographic, and economic factors that contribute to the gender composition of primary caregivers among adult children (Graham 1983; Aronson 1992; Urgerson 1983; Fisher and Tronto 1990; Finch and Grove 1980). This chapter is organized into two parts. In the first part, the major characteristics of the care recipients and caregivers in the sample and the gender and relational differences in social, demographic, and economic factors are presented. This will help to delineate the gendered nature of the factors that are later entered into the model to predict caregiving behaviors. In the second part of the chapter, preliminary findings about the gender and relational differences in the caregiving of 72 73 primary caregivers in terms of what they do and how the work is divided between them and secondary caregivers are presented. The Frail Elderly and Family Care Providers Characteristics of the Sample In Table 3.1, the sociodemographic and the caregiving situations in which primary caregivers are placed, such as functional dependency and diagnosis, are presented for the total sample as well as for the four caregiver subgroups. The elderly cared for by a family member had a mean age of 73 years, somewhat younger than care recipients in a national sample (78 years of age in Stone, Cafferata, and Sangl, 1987). The majority of care recipients were women (51%), married (68%), and lived with those who provided them care and/or with other members (86%). The elderly cared for by adult children were about ten years older than those cared for by spouses (78 vs. 69), thus, somewhat more functionally dependent (10.1 vs. 9.1). The household incomes of the elderly cared for by adult children was higher than those of the elderly cared for by a spouse ($33,340 vs. $27,830). In addition, care recipients cared for by adult children tended to be widowed (75 %), divorced or separated (9%), and fewer than one—fifth (15%) had a living spouse. This finding is consistent with the hierarchical caregiver selection model; adult children usually assume primary responsibility for caregiving when the spouse of a parent becomes unavailable (Cantor 1979; Shanas 1979; Stoller and Earl 1983). The gender composition of the frail elderly, that is, women constituting 33 percent of the elderly whose care was provided by a spouse and 81 percent of those whose care was provided by a child, illustrates to some extent the demographic 74 Table 3.1 Major Characteristics of Care Recipients by Relationship of Caregiver to Care Recipient All Caregivers relationship to care recipient Characteristics of care recipients carggivers Wife Husband Daughter Son “Number of cases 1387 573 288 472 54 Gender Male 49 100 0 % 20 15 % Female 51 0 100 80 85 Marital status Married 68 100 100 % 16 15 % Widowed 28 0 0 75 72 Divorced/Separated/Never married 4 0 0 9 13 Living Arragement Patient alone 9 0 0 % 23 21 % Patient and caregiver alone 55 79 82 13 38 Patient, caregiver, and others 31 21 18 51 32 Patient with others. caregiver apart 5 0 0 13 9 Diagnosis Alzheimer's 25 27 28 % 21 20 % Cancer 21 25 28 14 19 Other 53 48 44 65 61 Age M 73 70 68 78 77 SD (9.0) (7.3) (7.8) (8.8) (9.8) Household Income M 29900 26200 31090 33240 34240 (in constant dollars) SD (19510) (14440) (18130) (24250) (22220) No. of dependencies Total tasks (20) M 9.4 9.1 9.0 10.3 8.7 SD (4.7) (4.6) (4.8) (4.7) (4.5) Personal Care Tasks (6) M 2.6 2.5 2.4 2.9 2.1 SD (2.1) (2.1) (2.2) (2.1) (2.1) Housework (3) M 2.6 2.6 2.6 2.7 2.7 SD (0.8) (0.8) (0.8) (0.7) (0.6) Administration Tasks (2) M 1.5 1.5 1.4 1.7 1.6 SD (0.7) (0.7) (0.8) (0.6) (0.6) Mobility Tasks (2) M 1.3 1.1 1.2 1.6 1.0 SD (1.6) (1.6) (1 .6) (1.7) (1.5) No. of additional adults M 0.5 0.3 0.2 0.8 0.5 in the household SD (0.7) (0.6) (0.4) (0.8) (0.7) 75 Table 3.2 Major Chracterlstics of Primary Caregivers by Relationship of Caregiver to Care Recipient amfier's relationship to care recipfint I All haracterlstics of caregivers caregivers Wife Husband Daughter Son Number of cases - 1387 573 288 472 54 Race: whites ' 91 % 93 % 97 % 87 % 85 91: non-whites 9 7 3 13 15 Duration of Care > 2 years 32 % 35 % 31 % 29 % 28 % <= 2 years 68 65 69 71 72 Employment Full-time 21 % 9 % 16 % 35 % 52 % Part-time 10 8 6 14 13 Retired 44 55 73 15 18 Not employed 25 28 5 37 17 Occupation: Professional 25 % 15 % 40 % 27 % 32 % Clerical lSales 38 50 14 39 33 Operative lLabor 12 6 30 7 22 Other 6 2 15 4 7 Non employed 20 27 2 23 6 Marital status: Married 86 % 100 % 100 % 68 % 43 % Never married 7 0 0 13 46 Widowed/Divorced 7 0 0 1 9 1 1 Had children under 15' 29 % -- -- 30 % 17 96 over 15 71 — -- 70 83 Livinsg arrangement -together' 65 % - -- 64 % 69 96 apart 35 -- - 36 31 Helpers: availibe 54 % 47 % 52 % 63 % 61 % not available 46 53 48 37 39 Age M 60 66 69 49 47 SD (12.6) (7.4) (7.6) (10.6) (12.1) Household income M 32700 26200 31090 40760 37810 (in constant dollars) SD (19670) (14440) (18130) (22650) (20180) Adjusted household income M 22750 19200 23370 26220 25840 (in constant dollars) SD (13220) (10700) (13770) (14500) (12930) Years of education M 12.9 12.4 12.7 13.5 13.8 SD (2.8) (2.8) (3.4) (2.5) (2.3) Self-rated physical heath M 2.0 2.1 1.9 1.9 1.8 (1 =excellent 4=poor) SD (0.8) (0.7) (0.7) (0.7) (0.8) Hours of care per day M 10.7 11.6 10.8 9.9 6.8 SD (8.7) (8.7) (9.1) (8.5) (7.6) % of secondary carers' help to M 38.5 24.9 28.7 58.7 58.3 primary caregivers' help SD (135) (104) (83) (181) (133) No. of helpers M 1.3 1.2 1 .3 1.5 . 1.4 SD (1.8) (1.7) (1.8) (1.8) (1.5) : For Ella caregivers only (N=525) 76 composition of the elderly at large. The gender gap among spouses in age and life expectancy (Binstock and Shanas 1985; Rosenwaike 1985; Coward, Horne, and Dwyer 1992), thus a greater likelihood of women surviving their husbands, to a large extent, explains why women are less likely to become a care recipient cared for by a husband but constitute the majority of care recipients when care is provided by adult children (Lee, Dwyer, and Coward 1993). Elderly care recipients diagnosed with Alzheimer’s (25 %) and cancer (21%) made up almost half of the sample while the remainder had a variety of physical impairments. Among twenty possible functional dependencies, the elderly needed assistance with, on average, nine tasks. Or, to put it in a different way, 77 percent of the elderly as reported by caregivers had difficulty with at least one personal care task or mobility task, 99 percent had difficulty with at least one housework or administrative task, and 16 percent had difficulty with one or more health care tasks (figures not presented in Table 3.1). Almost two-thirds (62%) of caregivers in the sample were spouses and three- fourths (75%) were women: daughters constituted 34 percent of all caregivers and 90 percent of the child caregivers and wives constituted 41 percent of all caregivers in the samples and 67 percent of the spouse caregivers (see Table 3.2). As mentioned above, the gender composition of spouse caregivers can be largely explained on demographic grounds in terms of the gap between husbands and wives in age and life expectancy (Binstock and Shanas 1985; Rosenwaike 1985; Coward, Horne, and Dwyer 1992). The gender composition of adult children, however, appears to reflect, not women’s natural feelings of compassion and connectedness, but the unequal 77 gender division of labor sustained by the patriarchy and structural forces of labor market (Aronson 1992; Finch and Grove 1980; Graham 1983; Fisher and Tronto 1990; Urgerson 1983). Women’s disadvantageous positions in the household and work place often give rise to dependency, poverty, and powerlessness, and thereby legitimate their role in taking care of an ill parent (Fisher and Tronto 1990; Graham 1983). The predominance of daughters in the caregiving work force may also be a result of parents’ preference of adult children of the same gender to take care of them, a fact that explains why more daughters than sons are chosen as primary caregivers because surviving mothers make up the largest group of care recipients (Lee, Dwyer, and Coward 1993; Stoller 1990; Finley, Roberts, and Banahan 1988).16 Many of the caregivers were themselves in the later stage of life --averaging 60 years of age in this sample -- and most were married (86%). Given this age distribution, it is not surprising that two-thirds of caregivers were either retired or unemployed. In general, the sample consists of largely white, middle-class, well- educated caregivers. Ninety percent of the sample were white, nearly half of the sample had at least some college education, one-fourth of the employed were professionals, and one-third of the employed worked in clerical or sales occupation. The average household income for the sample was $32,700 per year, in constant 1991 figures. 16Although parents prefer adult children of the same gender to take care of them, daughters become the preferred source of support for both older male or female recipients as their health deteriorates (Coward, Horne, and Dwyer 1992). 78 The majority of the caregivers who participated in this study were short-term caregivers; less than one-third of the caregivers had provided care for more than two years --6.8 years on average» at intake into the study. On average, caregivers spent 11 hours per day supervising and caring for activities.17 Nevertheless, caregivers did not provide care by themselves. More than half of the sample indicated the availability of helpers, i.e., 1.3 helpers, and these secondary caregivers’ hours of help constituted, on average, 39 percent of the hours care primary caregivers provided. The caregivers are not a homogeneous group. There are gender as well as relational differences in their characteristics. As expected, compared to spouses, adult children, a younger age cohort of caregivers, had more education and were more likely to be in the labor force and less likely to hold jobs in gender-specific occupations. In addition, adult children provided less hours of care and received more help from others in terms of both the number of helpers and hours of help than did spouses. As to economic status, the household incomes of spouses were significantly lower than those of adult children, mainly because spouse caregivers were in the later stage of the life cycle and were largely retired from work. Wife caregivers’ household incomes were lower than those of husband caregivers and those of any other caregiver groups. This may well result from a situation in which husbands, who usually are the major wage earners in the household, become ill, causing the household income level to plummet. When compared to the effect of the illness of husbands, the frailty of wives had a lesser impact on the living standard of the household unit. This income 17The median is eight hours of care per day. 79 differences between husband and wife caregivers reflect the prevailing gender differences in the earning opportunities and capacities of an older age cohort of Americans. Compared to husband caregivers, wife caregivers were less educated and less likely to ever have been in the labor force. In addition, the occupations spouse caregivers held were very gender specific. Husbands were primarily concentrated in professional and operative/labor occupations while clerical and sale categories accounted for half of wives’ employment. Despite the gender gap in socioeconomic status among spouse caregivers, there were no obvious differences between husbands and wives with respect to care characteristics. About the same proportion of caregiving wives and husbands (35 % vs. 31%) cared for their spouses for over two years. Husband and wife caregivers provided the same number of hours of care (10.8 vs. 11.6), were equally likely to live alone with frail spouses (79% vs. 82%), and to have secondary helpers (1.2 vs. 1.3 helpers), although husbands received slightly more hours of help from others in relation to what they provided (25% vs. 29%, respectively). Finally, there was no gender difference in the level of functional dependencies of the elderly cared for by spouse caregivers (9.0 and 9.1 for husbands and wives, respectively). Sons and daughters, on the other hand, were different in many ways. Both were middle aged (49 and 47 for daughters and sons, respectively). Daughters, however, were more likely than sons to be caught in the middle of competing obligations between caring for the frail parents and their own family responsibilities in household maintenance and child-rearing, a phenomenon characterized by Brody (1981, 1990) as pressures on the "woman in the middle." Over two-thirds (68%) of daughters were 80 married and 30 percent had children under 15. In contrast, only about two-fifths (43%) of sons who assumed primary responsibility were married and 17 percent had young children at home. The low marriage rate among sons --46 percent of sons never married-- is especially striking, given that over 80 percent of men between 30 and 65 years of age had been married (US. Bureau of the Census: 1990 Census of Population). The role of caregiving has been added to women’s traditional roles as wives, homemakers, and mothers, thus intensifying demands on their time and energy. Despite the comparability of educational level achieved by sons and daughters in the sample (57% of daughters and 61% of sons were college educated), daughters who assumed primary caregiving responsibility were less likely than sons to be in the labor force; 37 percent of daughters in this sample, as compared to only 17 percent of sons, were not working. The departure in employment patterns may mainly be a result of the gendered effect of assuming the caregiving role (Abel 1990b). Men and women resolve the conflict between waged work and caregiving in different ways. Daughters are more likely than sons to quit work to become caregivers while sons resolve the conflict by reducing their caregiving responsibilities (Stone, Cafferata, and Sangl 1987). Although daughters were less likely than sons to be employed, the household incomes of daughters were higher than those of sons ($40,760 vs. $37,810) and the highest among all caregiver groups, a reflection of daughters’ marital status rather than their employment status. When marital status was taken into account, the income profile changes considerably (see Table 3.3). Among married child caregivers, the 81 Table 3.3 Child Caregivers' Household Income and Their Parents' Living Arrangement by Marital Status Daugfiter Son Caregivers' (unadjusted) household income Married M 47400 48060 SD (21613) (16895) Not married M 26590 29670 SD (17846) (19045) ICarIgivers' adjusted household income Married M 29151 29032 SD (13829) (9804) Not married 19960 23300 SD (13951) (14627) MARITAL STATUS OF CHILDREN AND THEIR PARENTS ADULT CHILDREN Married Single PARENT PARENT Total Married Single Total Mam'ed Single IParents' iivirlg arrangement Percentage distribution Together with caregivers 60.9 39.6 64.7 71.8 65.5 73.0 Apart from caregivers 39.1 60.4 35.3 28.2 34.5 27.0 ..Apalt and alone 25. 9 0.0 29.8 15.7 0.0 18.5 ..Apart with other 13.2 60.4 5.5 12.5 34.5 8.5 82 annual household income of sons was about the same as that of daughters ($48,060 vs. $47,400). The household income of single sons, on the other hand, was much higher than that of single daughters ($29,670 vs. $26,590). The same profile persisted even when the family size of caregivers was taken into account. Consistent with findings in previous research (Stoller 1990; Montgomery and Kamo 1989), sons in this sample were more likely to take care of parents in relative good health than were daughters who cared for parents. In fact, the elderly who were cared for by their daughters were the most impaired. The number of functional impairments for parents cared for by a daughter was 10.3 as compared to 8.7 for parents cared for by a son and to 9 for the elderly cared for by a spouse. The difference in functioning is mostly a result of the greater likelihood of a daughter taking care of a parent who was impaired in self-care activities such as personal care and mobility tasks. The impairment in these functional domains usually suggests the severity of the health of the elderly and the intense demands of time and energy put on caregivers.18 The gender difference in the functional dependencies of the frail parents for whom sons and daughters cared is also consistent with findings in other research in which daughters are found to be the preferred source of support by both male and female parent care recipients when their health deteriorates (Coward, Horne, and Dwyer 1992). 18As suggested in the literature, there is a hierarchical and cumulative relationship between self-care activities, such as personal care and immobility, and routine household work (Kempen and Suurmeijer 1990; Spector et al. 1987). Patients need help with routine household work before having difficulty with self-care activities. 83 As to living arrangement, or proximity, about two-thirds (65%) of adult children in the sample shared the household with frail parent(s). In contrast to evidence from other research (Coward 1987; Stone, Cafferata, and Sangl 1987), however, daughters in the sample did not seem to be any more likely than sons to live with dependent parents. About two—thirds (64%) of daughters co-resided with elderly parents as compared to 69 percent of sons who did so. These living arrangements were related more to the marital status of the care recipients and caregivers than to the gender of primary caregivers (see the lower panel of Table 3.3). Nearly three— fourths (71.8%) of unmarried children, in contrast to 60.9 percent of married children, resided with a frail parent. When the marital status of both caregiving children and the frail parent was taken into account, about two-thirds (60.4%) of married children lived apart from parents who had living spouses. But if a parent was widowed or divorced, only one-third (35.3%) of married children lived apart. And, there was even less chance that children who were not married lived apart from their spouse-absent parents (27%) or allowed them to live alone (18.5%). As mentioned above, and consistent with previous findings (Tennstedt, McKinlay, and Sullivan 1989), spouse caregivers, particularly wives, provided more hours of care per day and received substantially less assistance from secondary caregivers than did adult children. In contrast to this relational difference, gender gaps in some of the care characteristics, on the other hand, were not as large as one would expect on the basis of theory and some previous findings (Enright 1991; Tennstedt, McKinlay, and Sullivan 1989). With the exception of hours of care in supervising and caring (9.9 and 6.8 for daughters and sons, and 11.6 and 10.8 for 84 wives and husbands), husbands and wives, and daughters and sons, were fairly similar in terms of the availability of helpers, the number of helpers, and the hours of help per week they received from secondary caregivers relative to what they provided. The Role Responsibility of Caregiving Caregivers in this sample consisted of self-identified primary caregivers who considered themselves to provide the most care in a family to a relative. Caregivers who assumed the primary caregiving responsibility, however, did not necessarily held the same sense of obligation to care for a relative or react in the same way to the caregiving role. Two indices, role responsibility and positive reaction to caregiving role, were constructed to measure a primary caregiver’s obligation, feeling, and commitment toward the caregiving role. Constructed from a five-point Likert scale, the role obligation score is the average score of a caregiver’s perception of duty to care for the elderly while the index of positive reaction to caregiving role measures the feeling of primary caregivers toward the caregiving role. Significance tests are presented in Table 3.4. Here the data show that the main effects, gender and family relation, and the interaction term between the two were all statistically significant at alpha < 0.05. Spouses felt a stronger sense of responsibility to take care of a family member who became ill than did adult children. Male caregivers also differed from female caregivers, but the gender difference was found primarily among spouse caregivers. Husband caregivers felt more obligated than wife caregivers or any other caregiver group to care for a frail relative. In contrast, sons who assumed the primary caregiving responsibility were not different from daughters in their perception of role 85 Table 3.4 Caregivers' Attitude toward Caregiving Role by Relationship of Caregiver to Care Recipient Hus- Test of Significance Total Wife band D'ter Son (F Value) Index (1237) (523) (264) (404) (46) Gender Relation GXR Positive Reaction M 3.7 3.6 4.0 3.5 3.6 9.1“ 11.4‘“ 4.1' toCaregiving Role SD (0.9) (0.9) (0.8) (0.9) (0.8) Role Responsibility M 4.0 4.0 4.3 3.8 3.8 9.2“" 42.3“" 8.8” SD (0.7) (0.6) (0.5) (0.7) (0.6) items Constructed for the Indices of Caregiver's Attitude toward Caregiving Role Positive Reaction to Caregiving Role, Alpha-=36 1. I enjoy caring for __ 2. Caring for _ makes me feel good. 3. I feel privileged to care for _ iRole Responsibility, Alphas. 74 1. Caring for _ is important to me. 2. I believe it is my responsibility to care for _. 3. i could not live w myself if i just quit caring for _. 4. I really want to care for _. 5. I will never odo__ enou m p < 0.001; " p < 0.01; * p < 0.0 86 obligations. A similar pattern was found in the index of positive reaction to caregiving role. Spouse caregivers, husbands in particular, reacted more positively to the caregiving role than did child caregivers. Whereas husbands were more likely than wives to feel positive about their caregiving role, no gender difference was found among adult children who assumed the major caregiving responsibility. The lack of gender difference among adult children in the two indices is consistent with past research (Finley, Roberts, and Banahan 1988; Montgomery and Kamo 1989; Finley 1989) which found that sons and daughters who assumed the primary caregiving responsibility do not differ in their perceived obligation and feeling toward the caregiving role. In contrast, huge differences were found between caregiving wives and husbands, with husbands expressing a stronger sense of obligation and privilege about their caregiving role than wives. It is then interesting to know whether male caregivers’ strong sense of role responsibility, which is not any less, if not more, than that of their female counterparts, was transformed into their assistance of an ill family member, as measured by the range of task participation and the division of labor. Past research has suggested that although men and women share a common sense of obligation to care for the elderly, they have a different perception and expectation of the way they will fulfill their felt obligation. As suggested by Montgomery and Kamo (1989) and Finley (1989), gender inequality in the caregiving division of labor is associated less with felt obligation than with the way men and women are socialized to transform the obligation into actual caregiving. 87 Caregiving Involvement and Task Allocation In this section, preliminary findings about gender and relational differences in caregiving behaviors in terms of what primary caregivers do and how the work is divided between them and secondary caregivers are presented. The purpose of the preliminary presentation of the three indices, task involvement, task exclusivity, and relative contribution, is to give an overall picture of the patterns of the care provided by men and women, and by spouses and adult children, and thus to highlight the multi-dimensional facets of caregiving behaviors and the importance of gender as well as relational factors in understanding family caregiving. Range of task participation demonstrates the degree to which caregivers respond to care recipients’ functional needs whereas the two division of labor indices address how tasks are allocated between primary and secondary caregivers. Multivariate analyses of the three indices are presented in the following chapters. Task Participation The task-by-task analysis in Table 3.5 shows overall the active participation of primary caregivers. Over 94 percent of the primary caregivers are involved in each of the tasks for which the elderly need assistance --with the exception in bathing where only 88% of caregivers are involved. In general, husbands are as likely as wives to take on caregiving tasks of all kinds, except for bathing and doing laundry, in which wives are somewhat more likely than husbands to take part, and for transportation, for which husbands are more likely than wives to assume responsibility. In contrast, there is a much bigger gender difference among adult children in task participation, but primarily in female-oriented tasks. Overall, daughters are as likely as wives to be Table 3.5 Task Participation of Primary Caregivers by Relationship of Caregiver to Care Recipient‘ % CAREGIVERS PARTICIPATING TASKS DURING THE PAST TWO WEEKS CHILD CAREGIVERS Daughter Son All Hus- caring for caring for TASKS cmivers Wife band D‘tr Son Mother Father Mother ether“ Personal care tasks ressing 98.3 98.8 97.4 98.8 90.5 99.5 96.2 86.7 100.0 grooming 94.6 96.6 93.4 94.6 75.0 94.6 94.2 73.3 80.0 bathing 88.0 93.1 86.3 86.6 53.1 89.3 76.5 57.7 33.3 toileting 96.9 96.9 97.9 97.6 84.6 99.2 91.4 90.0 66.7 urination" 98.1 99.2 96.4 97.8 100.0 98.1 96.4 100.0 100.0 stooi‘” 98.4 100.0 98.3 97.0 100.0 97.1 96.9 100.0 100.0 Housework tasks housework 95.8 98.9 96.6 93.3 86.3 92.4 97.4 86.4 85.7 laundry 94.4 96.8 90.7 95.3 82.0 95.7 93.8 83.3 75.0 cooking 98.6 99.5 99.6 97.9 90.0 97.7 98.7 87.9 100.0 drnlnlstrative tasks money management 94.3 99.7 100.0 86.6 88.9 89.3 76.1 90.0 83.3 nsportation 95.7 94.3 99.0 95.2 97.9 95.0 96.2 97.6 100.0 Mobility tasks Iking 98.3 96.4 98.7 99.4 100.0 99.3 100.0 100.0 100.0 moving around the house 97.8 97.0 98.5 98.0 100.0 98.3 97.1 100.0 100.0 moving in and out of bed 97.1 96.5 97.8 97.0 100.0 96.9 97.3 100.0 100.0 lifting and turning in bed 97.5 98.3 96.6 97.7 90.0 98.0 96.8 87.5 100.0 'Number of valid cases varies from task to task. “Number of valid cases for sons caring for fathers in need of help with any of the tasks is 5 or less. "Number of valid cases less than 10 for son caregivers. 89 involved in either masculine or nurturing tasks, with the exception of assuming responsibility for bathing and money management, in which, as expected, wives are more likely than daughters to take part. Sons, on the other hand, have substantially lower involvement than daughters in almost every time-consuming and labor-intensive personal care and housework task (75% for sons vs. 94.6% for daughters helping with grooming and 82% for sons vs. 95.3% for daughters assisting with laundry). Consistent with findings in some caregiving research (Lee, Dwyer and Coward 1993), the gender of parents plays a role in children’s involvement, but only in certain tasks.19 Daughters caring for fathers are as likely as daughters caring for mothers to help with tasks their parents found difficult to perform on their own. In areas such as money management, bathing, and toileting, however, daughters caring for fathers indeed appear to be less likely than those caring for mothers to get involved. Although task participation of child caregivers, at least among daughters in this sample, appears to be associated with the gender of a parent, it may not be sufficient to account for why daughters are more likely than sons to be involved in caregiving tasks. When adult children caring for a parent of the opposite gender are compared, daughters who care for fathers are far more likely than sons who care for mothers to take on personal care tasks such as dressing, grooming, and bathing, and household work tasks such as laundry and cooking. 19Due to the insufficient number of sons caring for fathers in the sample (less than 10 cases), a cross—group or within-group comparison of adult children who cared for parents of the opposite gender is limited to two sets of contrast: daughter-mother dyads vs. daughter-father dyads and daughter-father dyads vs. son-mother dyads. 90 When the range of task participation is examined with the index of task involvement level, that is, the percentage of tasks caregivers take on out of total tasks for which assistance is required, the data in Table 3.6 again show the overall task commitments of primary caregivers. On average, primary caregivers are involved in 95.9 percent of all caregiving tasks for which assistance is required. Even in a hands- on task domain like personal care, primary caregivers are involved in 93.3 percent of tasks. The range of task participation, however, varies according to the gender and family relationship of caregivers. Overall, wives do not seem to differ from husbands in the range of tasks they take on, although wives are slightly more likely than husbands to provide assistance with personal care and housework tasks, and husbands are somewhat more likely than wives to assume responsibility for administrative tasks. For adult children, on the other hand, sons participate in far fewer tasks than do daughters. On average, sons participate in only 87.2 percent of the total tasks for which their parents need assistance, as compared to 94.3 percent of task participation for daughters and about 97 percent for spouses. In addition, sons limit their participation to tasks traditionally defined as men’s work while daughters, like spouses, tend to extend their participation beyond tasks associated with their traditional gender roles. Daughters are involved in 98.1 percent of mobility tasks and 91.5 percent of administrative tasks, hardly less than the task involvement of sons (98.9% of mobility tasks and 95.1% of administrative tasks). In the personal care and housework domains, however, sons participate far less in tasks for which assistance is required than do daughters (70.7% vs. 91.2% in the personal care domain and 85.2% 91 Table 3.6 Range of Task Participation on a Continuous Scale by Relationship of Caregiver to Care Recipient % OF TASKS CAREGIVERS PARTICIPA TED All TASK DOMAIN N caregflers Wife Husband Daughter Son Total tasks 1387 ' M 95.9 97.6 96.8 94.3 87.2 SD (10.2) (8.1) (9.7) (11.0) (16.8) Personal care 101 1 M 93.3 96.9 93.5 91 .2 70.7 SD (21.1) (13.9) (19.5) (24.6) (38.5) Housework 1254 M 96.1 98.5 95.7 94.8 85.2 SD (14.9) (7.6) (15.7) (17.6) (26.5) dministration 1179 M 95.1 96.6 99.1 91.5 95.1 SD (18.3) (15.6) (9.5) (23.5) (15.0) obility 621 M 97.9 97.2 98.5 98.1 98.9 SD (12.3) (13.7) (10.7) (12.3) (5.3) CHILD CAREGIVER Daughter Son All child caring for caring for ASK DOMAIN N caregivers Mother Father Mother Father‘ irotal tasks 526 M 93.6 94.7 93.0 87.1 87.6 SD (11.9) (10.9) (11.4) (17.1) (16.1) Personal care 403 M 89.5 92.9 84.5 70.5 71.7 SD (26.7) (22.3) (31.7) (40.9) (27.0) Housework 499 M 93.8 94.4 96.9 84.8 87.5 SD (19.0) (18.3) (14.2) (27.0) (24.8) 485 M 91.9 92.4 87.8 95.4 93.8 SD (22.7) (22.9) (25.4) (14.7) (17.7) 264 M 98.1 98.1 97.9 98.5 100.0 SD (11.9) (11.7) (14.4) (6.1) (0) " No. of valid cases for sons caring for fathers in any task domain is less than 10. 92 vs. 94.8% in the housework domain). Since the distribution of the range of task participation is highly skewed and falls unevenly at the right endpoint (with a skewness value as high as 3.3 and kurtosis 11.7), the responses are collapsed into two categories: caregivers taking on all the tasks in a given task domain and those participating in only some or none of the tasks in the domain. The sample distribution of the collapsed categories across groups is presented in Table 3.7. The patterns found in the collapsed categories are consistent with the patterns found in the continuous scales shown in Table 3.6. Spouses are more likely than adult children, and women are more likely than men, to be involved in all caregiving tasks for which assistance is required. Nearly nine-tenths of wives (87.6%) take on all caregiving tasks, as compared to 83.7 percent of husbands, 70.8 percent of daughters, and 51.9 percent of sons. The most striking differences are found between daughters and sons in the nurturing and hands-on task domains of personal care (85.6% vs. 52.9%) and housework (90.4% vs. 71.7%). The notion that personal care tasks associated with intimacy raise the taboo thresholds for adult children seems to be confirmed by the low task participation of child caregivers in this domain. Aside from the relational gap, gender differences in the task involvement level in the personal care domain remain large among spouses and, particularly, adult children. A large majority (85.6 %) of daughters, as compared to about half of sons (52.9%), and 93.3 percent of wives, as compared to 85.4 percent of husbands, are fully involved in the personal care tasks for which assistance is required. Sons’ lack of participation in personal care tasks is most striking; 17 .6 percent of sons, as compared to 5.1 percent of daughters, 2.1 percent of husbands, 93 Table 3.7 Range of Task Participation in Collapsed Categories by Relationship of Caregiver to Care Recipient Caregivers All TASK DOMAIN N participated caregivers Wife Husband Dauflbr (%) (%) (%) (%) Total tasks 1387 All tasks 79.7 87.6 83.7 70.8 Some tasks 20.3 12.4 16.3 29.2 Personal care 1011 All tasks 87.6 93.3 85.4 85.6 Some tasks 8.9 5.5 12.0 9.2 None of the tasks 3.5 1.2 2.6 5.1 Housework 1254 All tasks 92.1 96.1 91.8 90.4 Some tasks 6.9 3.9 7.5 7.8 None of the tasks 1.0 0.0 0.7 1.8 ministration 1179 All tasks 92.5 94.9 99.1 86.9 Some tasks 5.2 3.4 0.0 9.2 None of the tasks 2.3 1.7 0.9 3.9 Mobility 621 All tasks 96.3 94.7 97.7 97.1 Some tasks 2.6 3.9 1.6 1.7 None of the tasks 1.1 1.3 0.8 1.2 CHILD CAREGIVER Daughter Son Caregivers All caring for caring for TASK DOMAIN N participated caregivers Mother Father Mother Father‘ (%) (%) (%) (%) Total tasks 526 All tasks 68.8 72.3 64.5 52.2 Some tasks 31.2 27.7 35.5 47.8 Personal care 403 All tasks 82.9 88.1 75.7 57.1 Some tasks 10.9 7.8 14.9 21.4 None of the tasks 6.2 4.1 9.5 21.4 Housework 499 All tasks 88.4 89.5 94.1 71.1 Some tasks 9.6 8.6 4.7 24.4 None of the tasks 2.0 1.9 1.2 4.4 Administration 485 All tasks 87.2 88.8 79.1 90.7 Some tasks 9.3 7.2 17.4 9.3 None of the tasks 3.5 4.0 3.5 0.0 Mobility 264 All tasks 97.0 96.9 97.9 94.1 Some tasks 1.9 2.1 0.0 5.9 None of the tasks 1.1 1.0 2.1 0.0 'Valid cases for sons caring for fathers in any task domain are less than 10. 94 and 1.2 percent of wives, have secondary caregivers completely take over the responsibility. When the gender of the frail parent is taken into account, the difference between caregiving daughters and sons remains large. Whereas daughters caring for fathers are less likely than daughters caring for mothers to be involved in all tasks in the personal care domain (76% vs. 88 %), they are still far more actively involved in personal care tasks than are sons who care for mothers (57%) or fathers (33%). The gender of parents plays a role in adult children’s level of involvement but may not be sufficient to account for the low involvement of sons in personal care tasks, a point to be made in the multivariate analysis presented in the next chapter. Division of Labor: Relative Contribution and Task Exclusivity The measure of level of involvement that was discussed above reflects one dimension of what primary caregivers do, but it does not indicate how work is actually divided between primary caregivers and other helpers. Two division of labor indices constructed for this study, relative task contribution and task exclusivity of primary caregivers, serve to capture other dimensions of task involvement: how much primary caregivers are involved in relation to the amount provided by other caregivers, and how tasks are shared between them and secondary caregivers. Preliminary findings of the caregiving division of labor are presented in the following order: task—by-task analyses of relative contributions scored on a five-point Likert scale (primary caregiver always, primary caregiver more, equal, other more, other always), task domain analysis of two division of labor indices on a continuous scale and in collapsed categories. 95 Five-point Likert Scale of Division of Labor Indices The task-by-task relative contributions of primary caregivers on the five-point Likert scale are shown in Table 3.8. Consistent with findings in previous research (Tennstedt, McKinlay, and Sullivan 1989), adult children rely on secondary caregivers more often than do spouses and are less likely to undertake tasks all by themselves. Spouses are different from adult children in the relative task contribution in almost every caregiving task, from personal care tasks to housework to mobility tasks. In addition to the relational gap, men who assume caregiving responsibility also differ from women in how tasks are divided. The gender gap pattern in the relative contribution scale can be found among both spouses and adult children, but more so in female-oriented tasks than in male-oriented tasks. Although both husbands and wives participate in a wide range of tasks for which their ill spouses need to be cared, they are different in the way they share tasks, primarily female-oriented tasks, with other helpers. Data in Table 3.8 show that in tasks traditionally characterized as women’s responsibility, wives are much more likely than husbands to perform them by themselves. In tasks characterized as part of men’s domain, husbands are only slightly more likely than wives to undertake them alone. The gender gap in child caregivers’ use of extra help is even more remarkable. Sons are more likely than daughters to rely on other caregivers’ help, particularly in personal care tasks such as dressing, grooming and bathing and household tasks such as laundry. In female-oriented tasks, daughters, though receiving more assistance than do spouses, are still more likely than sons to undertake the tasks alone. In masculine- 96 Table 3.8 Relative Task Contribution Scale by Relationship of Caregiver to Care Recipient‘ CHILD CAREGIVER Daughter Son“ regiving All Hus- caring for caring for sks caregLvers Wife band D'tr Son Mother Father Mother Fathe L (%) (%) (%) (%) (%) (%) (%) (96) (96 resslng Caregiver always 49 60 51 34 24 35 29 27 1 Caregiver more 32 30 29 35 38 33 44 33 I Equally 13 7 14 21 10 23 15 7 1 Other more 5 2 3 9 19 9 8 20 1 Other always 2 1 3 1 10 1 4 13 I qGreoming Caregiver always 40 57 34 28 10 31 17 13 I Caregiver more 26 22 32 25 3O 23 33 33 I Equally 23 15 24 30 20 30 33 13 . I Other more 6 2 4 11 15 10 12 13 I Other always 5 3 7 5 25 5 6 27 I iaamrng Caregiver always 48 60 50 37 9 40 27 12 . Caregiver more 17 19 13 18 9 17 22 12 I Equally 16 10 17 23 13 24 22 12 17 Other more 7 5 7 8 22 8 6 23 1 Other always 12 7 14 13 47 1 1 24 42 67 Toileting Caregiver always 42 52 40 33 39 37 17 40 Caregiver more 28 29 32 25 31 22 34 40 I Equally 24 14 23 35 8 35 34 0 Other more 4 2 3 5 8 5 6 10 I Other always 3 3 2 2 15 1 9 10 FUrination Caregiver always 39 46 29 36 44 38 25 43 I Caregiver more 27 28 29 24 22 23 29 29 I Equally 29 22 36 33 22 31 39 14 I Other more 4 3 2 5 11 6 4 14 I Other always 2 1 4 2 0 2 4 0 I Stool Caregiver always 41 48 29 42 22 46 28 14 I Caregiver more 19 20 31 13 33 1 1 19 29 I Equally 33 27 33 39 33 37 44 43 I Other more 5 5 5 4 11 3 6 14 I Other always 2 0 2 3 0 3 3 0 I Shopping Caregiver always 60 68 67 48 50 48 45 47 71 Caregiver more 14 10 16 19 12 18 21 9 = Equally 21 17 14 28 34 29 25 40 I Other more 3 2 3 3 2 3 5 2 I Other always 2 2 1 3 2 3 4 2 I ework Caregiver always 52 66 48 42 31 42 44 34 1 . Caregiver more 24 21 33 22 22 21 30 23 1 . Equally 17 1 1 14 24 28 24 22 27 = Other more 3 0 2 5 6 6 3 2 3 Other always 4 1 3 7 14 8 3 14 1 - ‘ 97 Table 3.8 oont'd CHILD CAREGIVER ' Daughter Son“ regivlng All Hus- caring for caring for ks caregivers Wife band D'tr Son Mother Father Mother Fathen I (%) (%) (%) (%) (%) (%) (%) (%) (%) Laundrying Caregiver always 65. 76 61 60 34 61 56 33 Caregiver more _ 13 11 11 15 1O 14 21 9 1 Equally 14 8 16 16 34 17 1O 36 Other more 3 1 3 5 4 4 6 5 Other always 6 3 9 5 18 4 6 17 2 LCooklng Caregiver always 56 73 57 38 33 39 35 36 1 Caregiver more 29 22 32 34 28 32 43 24 Equally 10 4 7 16 20 16 15 15 Other more 5 1 3 10 10 11 7 12 Other always 1 1 1 2 10 2 1 12 My Mgt Caregiver always 72 82 80 58 61 63 42 60 Caregiver more 1 1 12 23 9 6 9 13 7 Equally 11 6 8 18 19 17 18 23 Other more 1 1 O 2 3 1 3 0 1 Other always 6 1 0 13 1 1 11 24 1O 1 Trans- Caregiver always 53 59 64 43 48 43 42 42 Caregiver more 17 15 20 18 4 17 19 5 Equally 24 20 15 32 40 31 33 44 14 Other more 2 1 1 3 6 3 3 7 Other always 4 6 1 5 2 5 4 2 Walking Caregiver always 33 41 43 23 27 24 19 25 33 Caregiver more 29 29 31 30 27 30 29 33 Equally 32 24 25 41 27 4O 45 25 Other more 4 4 0 6 20 6 7 17 Other always 2 4 1 1 0 1 0 O IGettlng Caregiver always 30 36 37 21 20 22 17 22 around the Caregiver more 34 35 34 33 40 34 31 44 House Equally 29 21 25 38 20 36 43 22 Other more 5 5 2 6 20 6 6 11 1 Other always 2 3 2 2 O 2 3 0 Transferring Caregiver always 37 40 50 25 42 27 19 40 Ira/out of Caregiver more 33 35 30 32 25 32 32 30 Bed Equally 24 19 17 32 25 31 35 30 Other more 4 3 0 8 8 7 11 O Other always 3 4 2 3 0 3 3 O Mug Caregiver always 31 35 36 26 30 27 26 38 Caregiver more 34 40 31 31 20 32 26 25 Equally 28 20 28 37 20 35 42 12 Other more 4 4 2 4 20 4 3 12 Other always 3 2 3 2 10 2 3 12 ' Number of valid cases for each task varies. " Number of valid cases for sons caring for mothers in need of help with any of the tasks is less than 20. and number of valid cases for sons caring for fathers in need of help with any of the tasks is less than 5. 98 oriented mobility tasks with the exception of transferring in and out of bed, however, sons do not seem to rely on others any less than daughters. Regarding the issue of taboos in caregiving, a close look at three personal care tasks -- dressing, grooming, and bathing-- further reveals the dual nature of tasks characterized by both intimacy and labor intensity. Spouses tend to undertake tasks requiring bodily contact without help whereas adult children are more likely to rely on others’ participation. The pattern to some extent confirms the notion that personal care tasks are tasks of intimacy that distinguish spouses from adult children in care providing. Nevertheless, spouses’ experiences in these tasks are not unvarying as there are gender gaps between spouses in regard to their use of outside help. Wives often help their spouses with dressing, grooming, and bathing by themselves whereas husbands rely more, though rarely totally, on other caregivers’ efforts. Division of Labor Indices on a Continuous Scale The indices of relative contribution and task exclusivity (see Formula 3 and 4 in Chapter 2), incorporated with the task frequency dimension of involvement, are constructed to measure the proportion of primary caregivers’ assistance in tasks in relation to that of secondary caregivers and the extent to which care activities are performed exclusively by primary caregivers. Data in Table 3.9 show that, on average, wives perform 88.3 percent of all care activities, as compared to 85.5 percent for husbands, 72.7 percent for daughters, and 65.3 percent for sons. With regard to the task exclusivity index, wives perform 64.1 percent of all care activities alone, as compared to 55.2 percent for husbands, 35.2 percent for daughters, and 31.9 percent for sons. Without exception, spouse caregivers’ relative contribution and 99 Table 3.9 Two Indices of Division of Labor on a Conflnuous Scale by Relationship of Caregiver to Care Recipient % TASK BtCLUSMTY . All Hus- FI’ASK DOMAIN N caregivers Wife band Dagghter Son Total tasks 1387 M 51.2 64.1 55.2 35.2 31.9 SD (40.1) (38.7) (39.8) (36.1) (34.2) Personal care 101 1 M 49.8 64.2 48.6 36.5 23.0 SD (46.0) (44.0) (45.9) (43.9) (37.8) Housework 1254 M 54.9 70.2 55.1 41.1 29.7 SD (44.6) (41 .2) (44.0) (43.4) (40.0) ministration 1 179 M 60.4 68.5 69.8 47.7 51 .0 SD (45.3) (42.8) (43.4) (45.5) (47.2) obility 621 M 35.8 41.2 45.1 25.4 40.9 SD (46.2) (47.6) (47.6) (41 .7) (50.3) % RELATIVE CONTRIBUTION All Hus- TASK DOMAIN N caregivers Wife band Daughter Son Total tasks 1387 M 81.5 88.3 85.5 72.7 65.3 SD (19.9) (16.7) (17.4) (20.9) (24.5) Personal care 101 1 M 77.9 87.5 79.1 68.9 60.0 SD (27.6) (20.5) (26.1) (30.0) (35.5) Housework 1254 M 83.7 91.9 86.5 75.3 64.8 SD (22.7) (15.2) (19.4) (26.1) (29.2) dministration 1 179 M 82.3 87.2 90.0 74.1 72.4 SD (24.5) (21.0) (17.7) (27.6) (29.5) Mobility 621 M 75.7 79.1 81.7 69.7 72.9 SD (24.3) (23.7) (21 .4) (24.5) (30.1) 100 task exclusivity indices scores are higher than are those of child caregivers in all task domains. In addition to the relational gap, there are also gender-specific patterns of labor division for both spouse and child caregivers. In task domains traditionally defined as women’s work, wives’ relative contribution effort ranks somewhat higher than that of husbands (87.5% vs. 79.1% in the personal care domain and 91.9% vs. 86.5% in the housework domain). In masculine domains, however, the relative contribution of husbands appears only slightly higher than that of wives (90% vs. 87.2% in the administration domain and 81.7% vs. 79.1% in the mobility domain). The gender- specific patterns of labor division for spouse caregivers are even more evident in the task exclusivity index. The gap in the task exclusivity between wives and husbands is large in areas such as personal care (64.2% vs. 48.6%) and housework (70.2% vs. 55.1%) but are not found in areas such as administrative (68.5% vs. 69.8%) and mobility domains (41.2% vs. 45.1%). Husbands and wives may be equally involved in the tasks required for assistance, but husbands are more likely than wives to call for help, share tasks with others, and contribute less than what others provide. In many ways, the difference between sons and daughters in the relative contribution scale resembles the gender difference pattern between spouses, although the gender gap among adult children seems to be larger. In domains traditionally defined as men’s responsibility, the relative contribution of sons is not any greater than that of daughters (72.4% vs. 74.1% in the administration domain and 72.9% vs. 69.7% in the mobility domain). In female-oriented domains, however, the relative contribution of daughters is far more than that of sons (68.9% vs. 60% in the 101 personal care domain and 75.3 % vs. 64.8% in the housework domain). On the other hand, the task exclusivity patterns along gender-defined line can be found in both female-oriented and, to a lesser extent, male-oriented task domains. Daughters perform 36.5 percent of personal care tasks and 41.1 percent of housework tasks by themselves as compared to sons who perform on their own 23 percent of personal care tasks and 29.7 percent in housework tasks. In the mobility task domain that requires physical strength, daughters, on the contrary, are far less likely than sons to assist without the help of others (25.4% and 40.9%), although the overall relative contribution efforts of daughters are not any less than those of sons (69.7% vs. 72.9%). Division of Labor Indices in Collapsed Categories Since the distributions of relative contribution and task exclusivity indices also fall heavily on the end points, two indices of division of labor on a continuous scale are collapsed into three categories to be used later in the multivariate analyses of logistic regression (presented in Chapter 5). Responses for the relative contribution index were collapsed into three categories: (1) primary caregivers perform all (100%) care activities, or exclusive care; (2) primary caregivers perform more than half (50%—99%) of caregiving activities, or a complementary pattern of care; and (3) primary caregivers perform less than half of caregiving activities (0%-49%), or a supplementary pattern of care. Responses for the task exclusivity index that measures the degree to which primary caregivers assist with tasks without the involvement of other helpers were also collapsed into three categories: (1) exclusive care by primary caregivers (100%), i.e., primary caregivers perform all tasks without any help; (2) 102 partial task sharing (1 %-99%), i.e., primary caregivers are involved in at least one task in a given task domain without others’ assistance; and (3) total dependence of primary caregivers on secondary caregivers (0%), which includes the categories of primary caregivers sharing every task with others’ help and secondary caregivers exclusively performing all tasks. For the mobility domain, since only less than nine percent of primary caregivers adopted the pattern of partial task sharing, partial task sharing and total dependence are combined into one group: dependence on secondary caregivers. The sample distributions of the collapsed categories of relative contribution and task exclusivity are presented in Table 3.10 and Table 3.11, respectively. The collapsed relative contribution index in Table 3.10 shows that for all tasks combined, only about one-fourth (26.7%) of primary caregivers are exclusive care providers, completing all tasks alone; the majority of primary caregivers (62.9%), as expected, are engaged in the pattern of complementary care, providing more than half of all task demands. The patterns found in the sample distribution of the collapsed categories are similar to those found in the index on a continuous scale. Spouses are more likely than adult children to be the only care provider. Fewer than seven percent of spouses identifying themselves as primary caregivers assist in less than half of the care activities, as compared to 17.4 percent of daughters and 27.8 percent of sons who are in the primary caregivers’ role but provide supplementary pattern of care. This relational gap in the relative contribution can be found in all task domains, although it is somewhat less evident in mobility tasks that require physical strength for task accomplishment. 103 Table 3.10 Relative Contribution in Collapsed Categories by Relationship of Caregiver to Care Recipient Pattern of All Hus- TASK DOMAIN N Relative Contribution caregivers Wife band D'tr Son (%) (%) (%) (%) (%) Total tasks 1387 Exclusive care 26.7 38.7 30.9 11.4 9.3 Complementary care 62.9 56.5 62.2 71.2 63.0 Supplementary care 10.4 4.7 6.9 17.4 27.8 Personal care 1011 Exclusive care 40.9 55.8 38.0 28.2 11.8 Complementary care 39.5 36.1 43.2 41.5 38.2 Supplementary care 19.7 8.2 18.8 30.4 50.0 Housework 1254 Exclusive care 44.0 60.7 42.3 29.8 18.9 ' Complementary care 42.3 33.2 47.9 48.2 47.2 Supplementary care 13.7 6.1 9.7 22.0 34.0 Administration 1179 Exclusive care 53.5 62.1 65.3 39.2 45.1 Complementary care 28.6 27.1 27.0 32.3 17.6 Supplementary care 17.9 10.8 7.7 28.6 37.3 Mobility 621 Exclusive care 32.9 38.2 41.1 22.7 40.9 Complementary care 39.9 40.8 40.3 40.1 27.3 Supplementary care 27.2 21.1 18.6 37.2 31.8 CHILD CAREGIVER Daughter Sons Pattern of All caring for caring for TASK DOMAIN N Relative Contribution careglv_ers Mother Father Mother Father” (%) (%) (%) (%) (%) Total tasks 526 Exclusive care 11.2 11.1 12.9 8.7 12.5 Complementary care 70.3 71.2 71.0 63.0 62.5 Supplementary care 18.4 17.7 16.1 28.3 25.0 Personal care 403 Exclusive care 26.8 30.5 18.9 14.3 0.0 Complementary care 41.2 38.6 52.7 32.1 66.7 Supplementary care 32.0 30.8 28.4 53.6 33.3 Housework 499 Exclusive care 28.7 29.6 30.6 20.0 12.5 Complementary care 48.1 47.1 52.9 48.9 37.5 Supplementary care 23.2 23.3 16.5 31.1 50.0 LAdministration 485 Exclusive care 39.8 41.1 31.4 39.5 75.0 Complementary care 30.7 31.3 36.0 20.9 0.0 Supplementary care 29.5 27.6 32.6 39.5 25.0 Mobility 264 Exclusive care 24.2 24.7 14.6 41.2 40.0 Complementary care 39.0 39.7 41.7 35.3 0.0 Supplementary care 36.7 35.6 43.8 23.5 60.0 " Number of valid cases for sons caring fofiather in any task domain is less than 10. 104 As mentioned above, husband caregivers in this sample have a strong sense of caregiving commitment, as seen in their wide range of involvement ranging from hands-on tasks to intermittent tasks. When other caregivers’ contribution is taken into account, husbands’ involvement in the time-consuming and labor-intensive tasks, as compared to that of wives, are no longer impressive. In the personal care domain, husbands are less likely than wives to be the sole care provider (38% and 55.8%, respectively); nearly one-fifth (18.8%) of husbands, as compared to only 8.2 percent of wives, have secondary caregivers assume the majority of caregiving chores. In the housework domain, while both husbands (9.7%) and wives (6.1%) are unlikely to be engaged in a supplementary pattern of care providing, wives are more likely than husbands to be the sole care provider, performing all tasks alone (60.7% and 42.3%, respectively). In contrast to the pervasiveness of spouses as the sole care provider, adult children are unlikely to accomplish caregiving without outside help. Only 9.3 percent of sons and 11.4 percent of daughters, as compared to 30.9 percent of husbands and 38.7 percent of wives, are the sole care provider. While both are equally unlikely to be the sole provider for the care of their parents, sons are more likely than daughters to have secondary caregivers complement their role as primary caregivers. Only 17 .4 percent of daughters, as compared to 27.8 percent of sons share less than half of all task demands. The gender gap in the relative contribution is most apparent in female- oriented task domains, particularly in personal care. For example, in the personal care domain, daughters are more likely than sons to be the sole care provider (28.2% vs. 11.8%, respectively) whereas sons are more likely than daughters to provide care in a 105 supplementary patterns (50% vs. 30.4%, respectively). In male-oriented task domains, particularly the mobility domain, sons, expectedly, tend to be an exclusive provider whereas daughters are likely to provide care with other caregivers’ help, in both complementary and supplementary patterns. A gender-stereotypic division of labor, revealed by the collapsed relative contribution scale, may be more evident for child caregivers than for spouse caregivers. Finally, personal care tasks that involve high degrees of bodily contact indeed raise the taboo thresholds for adult children, who are unlikely to be the sole provider in the tasks. The gender gap, obvious for both spouses and adult children, however, suggests that intimacy may not be the only factor that caregivers concern as they assist with personal care (55.8% for wives vs. 38% for husbands, and 28.2% for daughters and 11.8% for sons, as exclusive care providers). Among adult children, the cultural taboo hypothesis can be examined by the comparison of child caregivers caring for parents of the opposite gender. The within group comparison of daughters caring for parents of the opposite gender suggests that the gender of a parent plays a certain role in primary caregivers’ use of others’ assistance. Whereas both groups of daughters are highly likely to adopt a pattern of supplementary care in the personal care domain (30.8% and 28.4% for daughter-mother dyad and daughter-father dyad, respectively), daughters caring for mothers are much more likely to be the sole care provider in the helping networks (30.5% vs. 18.9% for daughter-father dyad). Data in the cross-group comparison of children caring for parents of the opposite gender, on the other hand, suggest that taboo may not be the only reason for sons’ low-level relative contribution in personal care. Daughters caring for fathers and sons caring for 106 mothers are equally unlikely to assist with personal care tasks without any help (18.9% vs. 14.3%), but their approach to the use of outside assistance is quite different. Daughters remain the major source of care (52.7% vs. 32.1% for sons) whereas the majority of sons are in a pattern of supplementary care, providing less than half of total assistance (53.6% vs. 28.4% for daughters). With regard to the task exclusivity index, the data in Table 3.11 also show both relational and gender gaps in the task sharing patterns of primary caregivers. For all tasks combined, although both spouses (51.9%) and adult children (70.1%) primarily adopt the pattern of partial task sharing, compared to adult children, spouses are relatively likely to provide tasks alone (38.7% and 30.9% for wives and husbands vs. 11.4% and 9.3% for daughters and sons) and somewhat unlikely to share with secondary caregivers every task in which they are involved (11% and 13.9% for wives and husbands vs. 18.4% and 20.4% for daughters and sons). Similarly, in all task domains except the mobility domain, in which all caregiver groups tend to undertake the tasks with others’ help, the majority of spouses are engaged in the pattern of exclusive care whereas adult children are, in the main, totally dependent on secondary caregivers. Although spouses tend to perform tasks in a given task domain all by themselves, it is not unlikely for spouses to share every single task with, or completely turn the chores over to, others when there are others in their helping networks. Take personal care tasks as an example. More than half (55.8%) of wives and about one-third (38 %) of husbands provide the tasks alone whereas most of daughters (46.3%) and sons (67.6%) totally depend on others to help. Nevertheless, this is not to say spouses do not maximize the extra help they 107 Table 3.11 Task Exclusivity in Collapsed Categories by Relationship of Caregiver to Care Recipient JPlattem of All Hus- TASK DOMAIN N Task Exclusivity caregivers Wife band D'tr Son (%) (%) (%) (%) (%) Total tasks 1387 Exclusive care 26.7 38.7 30.9 11.4 9.3 Partial task sharing 58.8 50.2 55.2 70.1 70. Total dependence 14.5 1 1 .0 13.9 18.4 20.4 Personal care 1011 Exclusive care 40.9 55.8 38.0 28.2 11.8 Partial task sharing 22.6 20.4 21.9 25.5 20.6 Total dependence 36.6 23.8 40.1 46.3 67.6 Housework 1254 Exclusive care 44.0 ' 60.7 42.3 29.8 18.9 Partial task sharing 27.4 21.1 29.6 33.2 26.4 Total dependence 28.5 18.2 28.1 37.0 54.7 dministration 1179 Exclusive care 53.5 62.1 65.3 39.2 45.1 Partial task sharing 16.3 15.5 10.8 20.0 15.7 Total dependence 30.2 22.5 23.9 40.8 39.2 Mobility 621 Exclusive care 32.9 38.2 41.1 22.7 40.9II Dependence on secondary 67.1 61.8 58.9 77.3 59.1 caregivers CHILD CAREGIVER Daughter Sons Pattern of All caring for caring for TASK DOMAIN N Task Exclusivity carggivers Mother Father Mother Father‘ (%) (%) (%) (%) (%) Total tasks 526 Exclusive care 11.2 11.1 12.9 8.7 12.5 Partial task sharing 70.2 72.3 61.3 69.6 75.0 Total dependence 18.6 16.6 25.8 21.7 12.5 Personal care 403 Exclusive care 26.8 30.5 18.9 14.3 0.0 ‘ Partial task sharing 25.1 25.4 25.7 21.4 16.7 Total dependence 48.1 44.1 55.4 64.3 83.3 Housework 499 Exclusive care 28.7 29.6 30.6 20.0 12.5 ‘ Partial task sharing 32.5 34.1 29.4 26.7 25.0 Total dependence 38.9 36.3 40.0 53.3 62.5 . dministration 485 Exclusive care 39.8 41.1 31.4 39.5 75.0 Partial task sharing 19.6 19.3 23.3 18.6 0.0 Total dependence 40.6 39.7 45.3 41.9 25.0 Mobility 264 Exclusive care 24.2 24.7 14.6 41.2 40.0 ' Dependence on secondary 75.8 75.3 85.4 58.8 60.0 caregivers ' Numbero vadli cases or sons caring or ther in any task domain is less than 10. 108 receive. When there are others in their helping networks, spouses are just as likely, if not more likely than children, to be totally dependent on outside help as they are to partially share tasks with other helpers (23.8% vs. 20.4% for wives and 40.1% vs. 21.9% for husbands). The tendency of adult children to totally depend on other helpers, and spouses to be exclusive care providers and task sharers with other caregivers are equally pronounced in the housework and administration domains. In addition to the relational gap, men are also different from women in the way they use other caregivers’ help, but the differences vary to a great extent depending on the type of tasks involved. In male-oriented task domains, wives are hardly different from husbands in the task exclusivity patterns (62.1% vs. 65.3% as exclusive care providers and 22.5 % vs. 23.9% in total dependence on secondary caregivers in the administration domain; and 61.8% vs. 58.9% with other helpers involved in the mobility tasks). In female-oriented task domains, particularly in personal care, on the other hand, the gap is wide between husbands and wives. Husbands are likely to share all care tasks with secondary caregivers whereas wives tend to perform tasks all by themselves. In personal care tasks that involve a high degree of bodily contact between care recipients and caregivers, 62 percent of husbands, as compared to 44.2 percent of wives, share tasks with others in varying degrees. Husbands’ total reliance on the help of others is especially striking. Four- tenths (40.1%) of husbands, in contrast to only 23.8 percent of wives, completely share the tasks of intimacy with other helpers. As for adult children, the gender gap in the task exclusivity pattern, not shown in all tasks combined, surfaces in the task domain analysis. In female-oriented task 109 domains, daughters, though never as likely as spouses to be exclusive care providers, are more likely than sons to assist with tasks without help while sons are more likely to be totally dependent on secondary caregivers to care for their parents. In the personal care domain, for example, over two-thirds of sons (67.6%), as compared to 46.3 percent of daughters, are totally dependent on other caregivers’ involvement whereas daughters are far more likely than sons to provide care all by themselves (28.2% vs. 11.8% for sons). Gender differences between adult children in male- oriented administrative tasks, on the other hand, are small (39.2% vs. 45.1% as exclusive care providers), although daughters are much less likely than sons to perform physically demanding mobility tasks without others’ help (22.7% vs. 40.9%). Finally, the large percentage of spouse caregivers who are exclusive care providers indeed is consistent with the cultural taboo hypothesis which posits that spouses are reluctant to seek help with tasks that involve bodily contact. Nevertheless, the hypothesis is put in doubt as one compares husbands’ degree of task sharing with that of wives in personal care tasks and compares spouses’ task sharing in the personal care domain with that in other domains. The data show quite a gender gap between spouses in the task exclusivity pattern (40.1% of husbands vs, 23.8% of wives totally relying on other caregivers’ help). In addition, spouses’ total task sharing in personal care that is infused with intimate content is not any less, if not more, than that in the housework or administration domains. Wives are as likely to totally depend on other caregivers’ help in personal care tasks (23.8%) as they are in the housework (18.2%) and administration (22.5%) domains. This pattern is even more dramatic for husbands, who are more likely to totally rely on others’ assistance 110 with personal care (40.1%) than they are for help in housework tasks (28.1%) and administrative tasks (23.9%). The gender gap between spouses and the comparable level of spouses’ task sharing in personal care and other domains suggest intimacy may not be the only factor affecting caregiving in personal care. The cultural taboo hypothesis in the context of adult children, to some extent, is supported in the preliminary data analysis. There is a higher degree of total dependence on secondary caregivers for daughter-father dyads (55.4 %) than for daughter-mother dyads (44.1%). Nevertheless, the cross-group comparison of children caring for parents of the opposite gender suggests that whereas intimate personal care tasks are supposed to raise the taboo thresholds equally for adult children caring for parents of the opposite gender, and thus to increase the need to call in help, secondary caregivers play a much more important role for sons than for daughters (64.3% for son-father dyads vs. 55.4% for daughter-mother dyads in total dependence on other caregivers’ help). In all, the preliminary data analyses of the pattern of caregiving behaviors showed, first, primary caregivers are actively involved in the tasks for which assistance is required, with the exception of personal care tasks that are labor- intensive and time-demanding. Primary caregivers, though not alone in providing care, are rarely totally dependent on others’ assistance as they assume primary caregiving responsibility. Second, family relationship plays a pivotal role in the ways primary caregivers provide care. In all task domains, with the exception of the mobility domain that requires physical strength to carry out task demands, spouses are more likely than 111 adult children to be involved and, usually, to provide exclusive care. Nevertheless, it is not uncommon for spouses to share tasks, including tasks of an intimate nature, when there are others in their helping networks. In addition, the gender-stereotypic task participation and division of labor, primarily seen in domains traditionally defined as women’ sphere, are pronounced for both spouses and adult children. Lastly, whereas the intimate nature of personal care tasks indeed discourages spouses from seeking help, the juxtaposition of personal care tasks as both intimate and labor intensive cannot be ignored. Spouses’ task sharing with secondary caregivers in personal care are not any less, if not greater, than that in domains such as housework or administration that have no bearing on bodily contact. Further, the intimate nature of personal care tasks does not restrict husbands’ help seeking and task sharing behaviors as much as it restricts wives’, suggesting intimacy may not be the only factor dominating caregiving in personal care. Additionally, taboo appears to be a concern for children caring for parents of the opposite gender, at least for daughters where sufficient number of cases allowed the comparison. The cross-group gender differences in the caregiving indices between adult children caring for parents of the opposite gender, however, also cast doubts on the taboo hypothesis. The data show that the bodily contact involved in the intimate personal tasks is not considered as much a taboo for daughters as it is for sons. Summary Spouse caregivers and child caregivers in this study differed in their sociodemographic and care characteristics. Spouses were older, less educated, more likely to hold a gender-specific occupation, and to be largely retired from work than 112 were children who provided care to a parent. In addition, spouse caregivers provided more hours of care and received fewer hours of assistance from other helpers than did child caregivers. Consistent with findings in previous research (Miller 1990), caregiving daughters and sons were very different in both sociodemographic characteristics and caregiving experience. Daughters were more likely to care for a parent who was severely impaired and to provide more hours of care than were sons. Daughters, achieving the same level of educational attainment as did sons, were more likely than sons to be confined to the private sphere and to be caught in the middle of competing obligations between caregiving and their own family responsibilities. In contrast, husbands and wives who provided care had relatively few differences in their caregiving experience. They did not differ in the length of the caregiving role or the impairment level of the spouse for whom they cared. The primary difference lies in their socioeconomic status. Caregiving wives tended to be less educated, out of the labor force, to hold an occupation that is gender-specific, and to have lower household incomes than caregiving husbands and any other caregiver groups. Finally, the preliminary data analysis of the outcome variables gave an overview of how care is provided depending on the primary caregiver’s gender and relationship to the care recipient. The preliminary analysis of three facets of caregiving behaviors, range of task participation, relative contribution, and task exclusivity, highlights the complexity of caregiving, the role gender and family relation play in it, and how the relationship varies according to the type of tasks involved. Multivariate analyses of the three indices are presented in the following chapters to see if the patterns found in 113 the preliminary findings hold when variables associated with primary caregivers’ gender and family relationship to care recipients are controlled. CHAPTER 4 GENDER STEREOTYPIC TASK PARTICIPATION Research on gender and caregiving has suggested that the types of tasks sons and daughters perform are often reflections of stereotypic gender roles. Daughters are involved in a broad range of tasks and provide routine and hands-on assistance whereas sons are primarily involved in intermittent tasks and offer occasional help (Horowitz 1985a; Finley 1989). With few exceptions (e. g. Dwyer and Seccombe 1991; Dwyer and Coward 1991), much existing research lacks a multivariate analytical framework to separate the effects of gender from other effects that might be attributable to gender. In addition, existing research, generally, does not distinguish between primary and secondary caregivers (e.g., Finley 1989; Dwyer and Coward 1991; Jutras and Veilleux 1991).20 Thus, it sheds little light on the degree to which the gender stereotypic task participation found among child caregivers as a whole also holds true for sons who are primary caregivers and husbands who have a strong commitment to their caregiving role. 20In research where primary and secondary caregivers were both recruited in the sample but were not distinguished in the data analysis, the gender stereotypic task participation among adult children highlights more of the divergence of work between primary and secondary caregivers than of the gender differentiation of task participation per se. Primary caregivers, the role of which is usually relegated to daughters, are care providers assuming hands-on care in a routine caregiving style. Secondary caregivers, roles usually assume by sons, are care managers, giving commands and intermittent help in a backup style of care (Archbold 1983). 114 115 Employing a multivariate analytical approach, this chapter is devoted to identifying factors that contribute to primary caregivers’ task participation. The focus is on understanding gender and relational effects on the range and the types of tasks primary caregivers perform while controlling for known confounding effects, such as time, power/resources, and gender beliefs. Without considering the differential effects of time, power/resources, and ideology by gender, this multivariate analytical approach also tests the effects of power/resources, time, or gender ideology under the assumption they operate the same way for men and women. The findings from the data analysis are presented in three parts. First, the full- sample analyses of all tasks combined and four task domains are presented. In the second part, the data analysis is based on a large set of variables applicable to adult children’s caregiving context. The testing of the cultural taboo hypothesis regarding adult children’s task involvement in personal care tasks is presented in the third part of the chapter. Range of Task Participation for All Caregivers The aim of the first part of this chapter is to identify the effects of primary caregivers’ gender and family relationship to care recipients on the range of tasks provided, while known confounding effects are controlled. As mentioned in Chapter 3, many characteristics of care recipients and primary caregivers, such as age, impairment level, and employment status, that are likely to exert influence on caregivers’ provision of assistance, are themselves associated with the gender and family relationship of caregivers to care recipients. In a multivariate analytical approach, the effects of gender and family relation can be disentangled from the 116 effects of other factors. The multivariate analyses also test the average effects of gender ideology, time availability, and power/resources under the assumption that these effects are not gendered and thus are constant across all caregiver groups. Nevertheless, as mentioned in Chapter 2, an additional testing of the differential effect of role ideology by gender in a separate analysis will be presented to redress the deficiency of the model without a specification of such an interaction term. One other factor, duration of care, that emerged as a primary factor in understanding caregivers’ range of task participation is also discussed at length. Two interaction terms -- gender and family relation, and family relation and duration of care -- are tested to determine: (1) if the effects of gender are the same for child caregivers and spouse caregivers; and (2) if the effects of family relation are the same for short-term as well as long-term caregivers. Another interaction term, that of gender and duration of care, was removed from the final model specification since preliminary data analyses showed that the term is statistically insignificant for all outcome indices and for all task domains. That is, the effects of gender do not differ between short-term and long-term caregivers; whatever gender difference patterns are found among short-term caregivers are also likely to prevail among long- term caregivers. Data analysis and interpretation of the results are based on multiple logistic regressions using multiple independent variables common to all caregiver groups. The five outcome variables include all tasks combined that are required for assistance and the four previously identified task domains. The index of the range of task participation, constructed on a continuous scale, is dichotomized into two categories: 117 caregivers involved in all the tasks with which care recipients need help and caregivers yielding at least one of the tasks to other helpers. To ease interpretation and simplify data presentation, only odds ratios are presented throughout the study and those reported in the text are all statistically significant at alpha < 0.05 unless otherwise specified. Odds ratios are recovered by taking the natural antilogarithm of the logit coefficient (i.e., exponentiating the logit coefficient) estimated in the equation. In this exponentiated form, values exceeding 1.00 indicate an independent variable raises the odds of primary caregivers taking on all tasks, while values less than 1.00 indicate a decreased odds. Lastly, odds ratios only indicate the association between independent and outcome variables. The Wald statistic, a better source of information on the relative importance of a variable, is obtained by taking the estimated parameter divided by its estimated standard error (Hosmer and Lemeshow 1989; Peduzzi, Hardy, and Holford 1980). To simplify the presentation of data in tables, the Wald statistic is not reported in the tables but is used as a tool when the relative importance of variables is discussed in the text. Finally, it should be noted that, in a model that specifies interaction effects, coefficients for the dummy variables are no longer the estimates of the average effects of certain variables for the total sample but the estimates for certain subgroups, depending on which subgroup constitutes the baseline (Hardy 1993; DeMaris 1991; Lemeshow and Hosmer 1989). Table 4.1 shows the coefficients used to predict task participation for various subgroups and the various contrasts of differences between 118 Table 4.1 Coefficients Used to Predict Task Participation for Relation-by-gender and Relation-by-duration-of-care Subgroups in the Model 103 (OINTIG,R,DC)=B0 + B1*Gender + B2*Relation + B3*Duration-of-care + B4*G*R + B5*R*DC, Where Male, Child, and Short-term Caregivers Are Base-line Groups Female Male Long-term caregivers Spouses B0+B1+BZ+B3+B4+BS B0+B2+B3+BS Children B0 + Bl + B3 B0 + B3 Short—term caregivers Spouses B0+B1+B2 +B4 B0+B2 Children B0 + B1 B0 Note: The coefficients, B1-B3, in the equation estimate not the average effects of gender, family relation, or duration of care for the total sample but the effects for certain subgroups, depending on which subgroup is the base line group. The estimated log—odds for gender, family relation, and duration of care for various subgroups are as follows: Bl: Gender effect among adult children B1 +B4: Gender effect among spouses B2: Relational effect among short-term male caregivers B2 +B5: Relational effect among long-term male caregivers B2 +B4: Relational effect among short-term female caregivers B2+B4+B5: Relational effect among long-term female caregivers BS: Duration-of—care effect among adult children B3 +b5: Duration-of-care effect among spouses 119 Table 4.2 Odds Ratio of Caregivers Participating in All Tasks. by Task Domain L Personal Administration Predictorg'? Total Tasks Care Housework Mobility‘i nder (Female vs. Male) among adult children 3.32 "" 8.37 “" 3.72 m 0.93 2.50 among spouses . 1.54 2.23 ' 1.95 0.24 0.75 Family Relation (Spouse vs. Child) among short-term male cgs. 13.06 "" 27.66 "' 9.44 ”' 9.86 ' 2.95 among long-term male cgs. 3.16 " 3.00 4.00 ' 28.59 “ among short-term female cgs. 6.06 m 7.37 m 4.96 m 2.52 ' 0.88 among long-term female cgs. 1.47 0.80 2.10 7.29 ” Duration of Care (Long-T vs. Short-T) 2.94 among adult children 3.33 “" 5.12 m 2.18 ' 1.73 among spouses 0.81 0.55 0.92 5.00 " er X Relation 0.46 ’ 0.27 " 0.53 0.26 0.30 Relation X Duration of Care 0.24 '" 0.11 m 0.42 2.90 esources [Time lideology j. Household Income 0.99 0.99 0.98 ' 1.00 1.01 upafion Clerical 1.08 1.15 1.13 0.73 1.08 Labor 1.25 1.26 1.47 0.74 3.67 Other occupation 1.09 1.67 1.18 0.83 1.00 No occupation 1.23 1.45 1.31 0.93 1.07 Employment 0.61 " 0.65 0.46 " 1.26 0.31 Educational Level 0.95 0.93 0.88 " 1.03 0.92 Other Characteristics of Caregiver 0.88 0.93 0.66 ' 1.03 0.67 0.97 ‘“ 0.94 '*" 0.97 * 1.00 0.94 Caregiving Situation Demand of care 0.93 m 0.96 1.04 0.98 0.72 ltlonal helpers 0.89 0.90 1.45 0.61 ” 0.96 Diagnosis: Alzheimer's 1.42 2.18 "' 0.84 1.03 0.65 Likelihood Ratio 1116.90 594.57 557.00 455.30 150.14 Degree of Freedom 1267 930 1141 1073 577 Probability 1.00 1.00 1.00 1.00 1.00 "’p<.001;”p<.01;'p<.05 Note: 1. For the mobility domain, interaction effect of duration of care by family relation is unable to obtain due to insufficient cases in design cells. Only main effect of duarion of care and two sets of relational effects. one for men and the other for women, are estimated. Therefore, effects in the rows of Family Relation“ represent the average relational effects for male and female caregivers. respectively. 2. Coding scheme as follows. Gender. 1=female,0=male: Relation: 1=spouses.0=children; Duration of care: 1 >2 years,0=<=2 years: Adjusted caregiver household income: in dollars: Occupation: professionals as reference group: Employment: 1=full time. 0=part time or not employed; Educational level: in years Health:4=poor, 1=excellent; Age: in years; Demand of care: # of functional dependencies: Additional helpers: in numbers; Diagnosis: 1=Alzheimer's. 0=cancer or physical impairment. 120 subgroups that can be derived from these coefficients.21 Since coefficients in the model represent not the average effects but the contrasts for certain subgroups, all relevant contrasts of differences are presented in the upper portion of Table 4.2. For example, the odds ratio for the ”gender effect among adult children" estimates the gender differences between daughters and sons in estimated odds of caregivers participating in all tasks. The odds ratio for the "family relation effect among short- term male caregivers" estimates the degree to which husbands are more likely than sons to participate in all caregiving tasks when taking on a short-term caregiving role. The Divergence in Female Tasks and Convergence in Male Tasks among Adult Children In her discussion of gender differences among adult children in their role performance as primary caregivers, Horowitz (1985a, p.614) concluded that 21The estimates of the standard errors, needed to assess the significance or confidence levels for the estimated odds ratios, can be computed by the following formula if standard errors of the coefficients and correlation between variables are known: Var(B,+B2) = Var(B,)+Var(B2)+2Cov(B,,B,), where Cov(B,,Bz) = Corr(B,,B,)*Se(B,)*Se(B2). And, the endpoints of the 100(1-alpha)% confidence interval of the odds are: exp[(Bl+BZ) + 214,19“,2 * SE(B1+B2)] For example, to estimate the significant effect of gender among spouses within the personal care domain, we need the coefficient/standard error of the gender effect for children (Bl =2.1242 and SE=0.4534), the coefficient/stand error of the interaction effect of gender by relation (B4=-l.3227 and SE=0.5521), and their correlation coefficient (Corr(Bl,B4) =-0.7837). The odds ratio of the gender effect for spouses: exp(Bl+B4)=exp(2.1242+(-l.3227))=exp(0.8015)=2.23 Its standard error: [(0.4534)2+(0.5521)2+2"‘(0.4534)(0.5521)(-0.7837)]“2 =0.3435 Confidence interval at alpha < .05: exp(0.8015 i 1.96*0.3435) = (1.14,4.37) 121 Sons and daughters are more likely to diverge in the provision of specific types of care. With the exception of health care, daughters were significantly more likely to help their parents with all services calling for "hands-on” assistance. [W]hen the task was less gender-specific or tended to be male-oriented..., sons did not significantly differ from daughters in their involvement. This statement, supported by evidence in most of the family caregiving research on adult children’s caregiving behaviors (Dwyer and Coward 1991; Montgomery and Kamo 1989; Stoller 1983), also summarizes the findings of this study in regard to the task involvement among child caregivers. The logit modeling for total tasks shows that the odds that daughters are involved in all caregiving activities are 3.32 times the same odds for sons. This gender effect for the total task activities, however, does not reflect the pattern in every task domain. In task domain analyses, the data show there is a strong tendency of gender-stereotypic participation in tasks within female-oriented task domains but not within male-oriented task domains. Sons and daughters are likely to differ in the level of participation in tasks that are an extension of women’s reproductive roles but not in tasks that are congruent with men’s gender roles. Daughters are 8.37 times more likely than sons to be involved in all tasks within the personal care domain and 3.72 times more likely in the housework domain. In contrast, for the administrative and mobility task domains that are traditionally defined as men’s responsibility, daughters do not differ from sons in their level of task performance. These findings confirm those of previous research that focused on child caregivers’ behaviors. That is, sons who are primary caregivers tend to participate in tasks that are extensions of their masculine roles whereas daughters who are primary caregivers tend to adopt the roles of both care providers and care managers, working 122 in areas congruent with their reproductive roles and extending their care to tasks in male-oriented domains (Jutras and Veilleux 1991). Normative Expectations of Marriage and Spouses’ Task Participation To what extent is the gender effect on the range of task participation found among child caregivers also present among spouse caregivers? In the handful of research that addresses gender inequality in caregiving labor among spouse and non- spouse caregivers, family relation has been suggested as a primary confounding variable that is important in understanding the relationship between gender and caregiving behaviors (Dwyer and Seccombe 1991; Cantor and Little 1985). The normative expectations of marriage in American society emphasize the obligation of husbands and wives to fulfill each other’s instrumental needs while providing emotional satisfaction, especially when a spouse becomes ill (Johnson 1983). The strong sense of role responsibility among spouse caregivers, therefore, is likely to prompt both husbands and wives to take part in tasks for which assistance is required by their ill spouses. Neither husbands nor wives are likely to relinquish caregiving responsibility in tasks required for assistance. The data in this study confirm this postulate to a great extent. The gender effect among spouse caregivers is not statistically significant for any task domain except personal care. Husbands are as likely as wives to perform housework tasks in which their spouses are in need of care, and wives are not any less likely than husbands to take on administrative and mobility tasks that are congruent with masculine gender roles. There are also no gender differences between husbands and wives in the involvement level in all tasks combined. 123 In the personal care domain that is categorized by both labor intensity and intimacy, however, husbands are not as committed or as actively involved as are wives, a pattern consistent with findings available in a few caregiving research studies (Stone, Cafferata, and Sangl 1987; Noelker and Wallace 1985). The odds that wives take on all the personal care tasks required for assistance are 2.23 times higher than the same odds for husbands whereas husbands tend to hand over to other helpers some personal care tasks, such as bathing and grooming, that require intimate bodily contact with care recipients. Caregivers as a Heterogeneous Group Caregivers are not a homogeneous population. Gendered expectations of the caregiving role influence how men and women provide care in a caregiving situation. Similarly, the strong normative expectations of marriage and sense of role responsibility emphasized for spouse caregivers might also affect their reaction to caregiving situations in ways different from adult children. Researchers of family labor and caregiving have focused on gender differences in caregiving behaviors. Whether the family relationship of primary caregivers to care recipients plays a role in care provision remains to be explored. Those few studies that have explored this issue found that spouse caregivers devote larger amounts of time to caregiving and show higher levels of task involvement than do child caregivers (Dwyer and Seccombe 1991; Enright 1991; Stone, Cafferata, and Sangl 1987). In addition, it has been suggested that the effect of gender by family relation on task participation is differential (Seccombe 1991; Enright 1991; Chang and White-Means 1991). Nevertheless, the lack of multivariate analytical approaches and problematic measures 124 of outcome variables in these studies call for systematic research to examine how range of task participation is dependent on the family relationship of caregivers to care recipient and if gender intersects with family relation to influence caregivers’ task participation. The data in this study overall confirm the findings of Dwyer and Seccombe (1991) and Enright (1991) who suggest that family relation may confound the association between gender and task involvement. The data in Table 4.2 (see the rows of Family Relation) suggest that for all subgroup comparisons, spouses are more likely than adult children to be involved in total tasks for which their relatives require assistance. In general, husbands are more likely than sons, and wives more likely than daughters, to take on caregiving tasks required for assistance. The effect of family relation, though profound, is not omnipresent. Various factors should be taken into account: the type of tasks required for assistance, gender of primary caregivers, and duration of care. In all task domains except mobility, the range of task involvement by spouses is wider than that by adult children. The lack of relational effect in the mobility domain is not surprising because it is a task domain requiring physical strength, which elderly spouse caregivers usually lack. In addition, although spouses are more likely than adult children to be involved in the two female-oriented task domains, the relational effect appears to be more pronounced in the personal care domain than in the housework domain, at least among those new to the caregiving role. For example, among those who have provided care for only a short time, husbands have an estimated odds of taking on all tasks in the personal care domain 27.66 times that for 125 sons as opposed to the odds ratio of 9.44 in the housework domain. Similarly, among female caregivers, short-term wife caregivers are 7.37 times more likely than daughters to be fully involved in all required tasks within the personal care domain, as compared to the odds ratio of 4.96 in the housework domain. Although both domains are associated with female gender roles, personal care tasks demand more labor-intensity, time commitment, and most of all, intimate contacts between caregivers and care recipients, all of which explain why the relational effects appear to be stronger for personal care than for housework. One of the research questions in this study is whether spouse caregivers are influenced by gender roles in the same way as adult children, that is, is the effect of gender a function of family relationship? The data show that the interaction effect of gender and family relation on the range of task participation is statistically significant for total task activities and the personal care domain (odds ratios=0.46 and 0.27) but not for the other task domains. For all total task activities and the personal care domain, first, the relational effect is much weaker among female caregivers than it is among male caregivers, and second, the gender gap in the task participation for spouse caregivers is not as wide as it is for child caregivers; husbands are much more likely than sons, particularly when new to the caregiving role, to participate in all tasks, as opposed to the narrower gap in task participation between wives and daughters. For example, in total task activities, the odds ratios for relational effect are 13.06 for male caregivers vs. 6.06 for female caregivers who are new to the caregiving role, and 3.16 for male caregivers vs. 1.47 for female caregivers who have provided long-term care. In addition, daughters are much more likely than sons to 126 take part in tasks included in total task activities and the personal care domain, as opposed to the comparatively small (or insignificant) gap in the task participation between husband and wife caregivers. Odds ratios of the gender effect are 3.32 for children and 1.54 for spouses in total task activities, and 8.37 vs. 2.23 in the personal care domain. Finally, although spouses are more likely than adult children to take part in the caregiving tasks for which assistance is required, the relational effect varies, to a great extent, with the length of time caregivers have performed the primary caregiving role. The interaction effect between family relation and duration of care is discussed below. Duration of Care as a Primary Factor Research on caregiving and family labor has suggested, but not fully examined, the relevance of time to the care primary caregivers provide (Montgomery and Kamo 1989; Stoller 1990). In their study of parental care by adult children, Montgomery and Kamo (1989) found that caregivers tend to progressively assume more intense and time consuming tasks as their caregiving career is extended. The relationship between duration of care and family caregiving is, however, subject to the gender of primary caregivers. Daughters seem to be more likely than sons to engage in the most demanding personal care tasks and for a longer period of time. It is unknown how spouses’ experiences fit into the relationships among gender, duration of care, and multi-faceted caregiving chores. How does the length of time for which family members have performed the primary caregiving role affect the range of tasks they take on? In what way and in what task domains do gender and family relation 127 confound the association between duration of care and range of task participation? The data in Table 4.2 reveal that duration of care indeed plays a role in the range of tasks with which caregivers assist; but its effect is primarily observed among child caregivers for all total task activities and the labor-intensive, time-demanding task domains of personal care and housework. Sons and daughters who provide long- term care are more likely than those in short-term caregiving role to fully participate in tasks within the total task activities, the personal care domain, and the housework domain (odds ratio=3.33, 5 .12, and 2.18, respectively). In contrast, spouses’ range of task participation has little to do with how long they have provided care. Husbands and wives who provide long-term care are as likely as those new to the caregiving role to take on the tasks for which their elderly spouses need assistance. The only exception to the lack of effect of duration of care among spouses is found for administrative tasks, in which spouses in a long-term caregiving role are five times more likely than those new to the caregiving role to fully take part. As for mobility tasks, group-specific estimates of the effects of duration of care are unavailable due to the insufficient number of cases in the design cells, but the overall effect of duration of care is statistically significant, suggesting that an increase in the length of care indeed increases the likelihood of primary caregivers taking part in mobility tasks. Are there gender or relational differences in the effect duration of care has on the range of task participation? Preliminary data analysis showed that across all task domains, there is no differential effect of the duration of care by gender on the range of tasks with which primary caregivers assist. The gender effect found among short- term caregivers, in both direction and in incremental quantity, tend to hold for long- 128 term caregivers. In other words, long-term son caregivers are more likely than their short-term counterparts to take on a wider range of tasks, particularly hands-on tasks; and so are daughters as the length of their caregiving is extended. Thus, there is no interaction effect involving gender and duration of care presented in the final model specification. The other interaction term, duration of care by family relation, examines whether the effect of duration of care for spouse caregivers is the same as for child caregivers. The data in Table 4.2 show that the effect of duration of care appears to be a function of family relation, but this effect is primarily observed for total task activities and for the personal care domain (odds ratios=0.24 and 0.11, respectively), within which, as the length of caregiving is extended, sons and daughters both are likely to take on a wider range of tasks whereas husbands’ or wives’ engagement in tasks is not related to the length of time they have provided care. In other words, adult children who assume the primary caregiving role for a long period of time became much more involved in caregiving tasks, thus narrowing the gap between them and long—term spouse caregivers’ task involvement.” Finally, it should be noted that in contrast to their progressive involvement in total tasks and hands-on tasks such as personal care and housework, adult children in the long-term caregiving role do not necessarily take on a wider range of tasks in the administration domain. Long-term spouse caregivers, on the other hand, tend to take a 22It may also be a sample attrition effect: those child caregivers who stay longer in the caregiving role are more like spouse caregivers. Since the effect of duration of care may present either change over time or attrition, any statement in this study that suggests a time effect should be read with caution. 129 very active part in administrative tasks. Thus, in contrast to the narrower gap in the participation of hands-on tasks between spouses and adult children who provide long- term care, the gap in administrative tasks, such as transportation and money management, in fact becomes wider, and occurs in a reverse direction, between long- term spouse and child caregivers. Husbands are more likely than sons, and wives are more likely than daughters, to take part in administrative tasks when they are in the long—term caregiving role than in the short-term ones. Odds ratios of relational effect are 28.59 for long-term male caregivers and 9.86 for their short-term counterparts, and 7.29 for long-term female caregivers and 2.52 for their short-term counterparts. In other words, adult children who perform primary caregiving roles for a long period of time tend to take on intensive and time-consuming tasks, but not necessarily transportation arrangement and money management. In contrast, long-term spouse caregivers are more actively involved in administrative tasks, but not necessarily in time-demanding and labor-intensive personal care or housework tasks. Power/resources, Time Availability, and Gender Ideology Thus far, the data have suggested the intertwining relationship of the level of caregivers’ task participation with their gender, family relationship to care recipients, and the length of time they have provided care. Other than these three factors, the effects of which have been discussed above, power/resources, time availability, and gender ideology, as pointed out by Finley (1989) and Montgomery (1992), have also been suggested as explanations for caregivers’ involvement in caregiving chores and, moreover, as alternative explanations for the gender differences found in analyses of care provision. The multivariate analytical framework employed here allows the 130 examination of the partial effects of power/resources, time availability, and gender ideology and the relative magnitude of these effects compared to that of gender in the model. This multivariate approach also helps to explore to what extent gender differences in caregiving are due to the power/resources caregivers possess, the time constraints imposed on caregivers, and the gender role beliefs held by caregivers. Proponents of the perspectives of power/resources, time availability, and gender ideology (Berardo, Shehan and Leslie 1987; Coverman 1985; Ross 1987) postulate that an increase in the power/resources and time constraints of caregivers and a stronger beliefs of traditional gender roles, particularly among male caregivers, are all likely to lead to a lesser involvement in caregiving tasks, especially hands-on tasks, among primary caregivers. Because women are less likely than men to be in the labor force and to possess social resources and because women have been socialized to nurturing roles, they experience less competing demands and role conflicts, possess less power in family negotiation and for formal service purchase, and show more concern for family than do men. They thus are more likely than men to be engaged in time-consuming and in labor-intensive tasks and tasks congruent with the nurturing roles women learn through socialization processes. The data in Table 4.2 show that the three hypotheses received only slight support. When the effects of power/resources, time constraints, and gender beliefs are controlled, female caregivers are still more likely than male caregivers to participate in caregiving chores, particularly among child caregiver groups and when tasks are labor-intensive and time-demanding. The Wald statistic, in fact, shows that caregivers’ gender, aside from family relation, is the most important factor predicting the range of tasks in 131 which caregivers participate. Nevertheless, this is not to say that power/resources, time availability, and gender ideology do not independently exert influences on the range of tasks caregivers perform, although their effects are neither as inclusive nor as pervasive as those of gender or family relation. All hypotheses receive support, primarily in the housework domain. The increase in the level of caregivers’ household income decreases the likelihood of their involvement in housework tasks but caregivers’ occupations have little to do with the range of tasks they perform. Caregivers employed full-time are less likely than those employed part-time or not employed to take part in all tasks required for help and tasks in the housework domain (odds ratios=0.61 and 0.46, respectively). An increase in level of education decreases the odds of caregivers participating in all housework tasks by a factor of 0.12 (=1.00-0.88). With regard to the testing of role ideology hypotheses, as mentioned in Chapter 2, the lack of specification of the interaction effect of education by gender may be problematic because, according to the role ideology hypothesis and findings in domestic labor research (Baxter 1992; Ross 1987; Seccome 1986; Huber and Spitze 1983), less educated male caregivers should perform fewer tasks congruent with female gender roles than do their educated counterparts whereas less educated female caregivers should take part in women’s domain more often than do their educated counterparts. An interaction effect of education and gender is thus added to the model and tested for its significance. All the tests, including within total task activities and all task domains, however, indicate (data not shown) that the interaction effect is not significant for either the total sample or samples of spouse caregivers and child 132 caregivers. Several explanations can be proposed for the insignificant findings. First, education as a proxy of gender role ideology may not actually reflect attitudes toward women’s and men’s place in the society of elderly spouse caregivers, who witnessed social changes in gender relations in their later years and thus are likely to adhere to traditional gender-role beliefs regardless of their educational level. As for adult children, education may be a proper proxy for a young cohort of primary caregivers, but the number of sons in this study is too small to provide sufficient statistical power. Other predictors In this study, several variables associated with caregivers’ health characteristics, age and self-reported health, and with the caregiving situation are entered into the model as control variables. Three variables that measure the caregiving situation, aside from duration of care, are care recipients’ functional dependency level, or demands for care, additional helpers residing in the care recipients’ household, and diagnosis of care recipient. It is generally believed that the health of caregivers and the functional dependency of care recipients impose restrictions on caregivers’ abilities to take part in caregiving tasks. Adults who reside in the care recipients’ households are likely to relieve the task burden of primary caregivers. Finally, the diagnosis of care recipients may also affect caregivers’ ability to perform tasks for which assistance is required. The data in Table 4.2 reveal that the characteristics of caregivers and the caregiving situation indeed exert an influence on the range of assistance provided to care recipients, although these effects are not omnipresent in all task domains. The 133 older the caregivers, the less likely they are to take on tasks required for assistance. The effect of age is found for all task domains except for the intermittent domain of household administration, probably because the latter requires little physical strength and age is, thus, an irrelevant factor. The effect of self-rated health, not as consistent as the effect of age, is supported only for the housework domain (odds ratio=0.66). The demands for care, or the number of daily activities for which care recipients are in need of help, is negatively related to the level of involvement in total task activities and in physically demanding mobility tasks (odds ratios=0.93 and 0.72, respectively). Consistent with findings in other research (Stommel et al. 1995), the more dependent the care recipients, the less likely primary caregivers are to get involved in all care tasks, suggesting that as the level of functional dependency of care recipients increases, more help from other caregivers is received. The effect of additional helpers and the diagnosis of care recipients also receive limited support in administrative and personal care domains, respectively. An increase in the number of additional helpers residing in the care recipients’ households decreases the likelihood of primary caregivers participating in all administrative tasks (odds ratio=0.61). Primary caregivers of care recipients with Alzheimer’s or dementia are more likely to perform all care tasks alone in the personal care domain than are those caring for relatives diagnosed with cancer or physical impairment (odds ratio=2.18). Adult Children’s Range of Task Participation Adult children’s caregiving contexts are somewhat different from those of spouses. In contrast to elderly spouses, adult children, who usually are middle aged, face more role conflicts and time demands, involving not only caregiving but also 134 their own family responsibility. Their ability to provide care depends to a great extent on their geographic proximity to their frail parents. In addition, two characteristics of care recipients, gender and the availability of a living spouse, can also affect the way adult children provide care. Caregiving literature has suggested that women are more likely than men to reside with a parent for whom they care (Coward 1987; Soldo and Myllyluoma 1983; Stone, Cafferata, and Sangl 1987), to be caught in the middle of family responsibilities (Brody 1981, 1990; Spitze and Logan 1990), and to be named designated caregivers by their parents, particularly mothers who make up the majority of care recipients (Coward, Horne and Dwyer 1992). Therefore, the observed gender differences in caregiving may be a result of other factors attributable to gender. Once these factors have been controlled, gender differences may no longer hold. A large set of variables common to the adult children’s caregiving context is employed to re- examine the gender effect. In addition, caregivers’ occupation is collapsed into two categories, professionals and all other, for the following two reasons: the trivial effect of the variable in full-sample analyses, and, most importantly, the zero frequencies of some occupational categories in response cells for some task domains that cause computation problem of parameter estimates (Hosmer and Lemeshow 1984; SAS Manual 6.01). The data in Table 4.3 show that the findings in adult-children-only analyses are consistent with those in the full-sample analysis. Gender differences persist for total task activities and two female-oriented task domains after all known confounding factors applied to adult children are controlled. In tasks congruent with traditional male gender roles, gender differences are not statistically significant. 135 Table 4.3 Odds Ratio of Child Caregivers Participating in All Tasks, by Task Domain [ Personal Administrflon Predictors'2 Total Tasks Care Housework Mobility"! aregiver‘s Gender (Female vs. Male) 3.14 m 8.56 "'" 4.36 ”" 1.13 2.06 Resources IT ime /Ideology j. Household Income . 0.99 0.98 " 0.99 0.99 1.01 Professional occupation 0.98 0.79 0.90 1.32 0.73 Employment 0.84 0.66 0.45 ' 1.73 0.89 Propinquity 2.10 " 1.48 4.90 “" 1.36 5.43 Dependent children 0.49 " 0.70 0.75 0.55 0.54 Educational Level 0.97 1.08 0.83 ' 0.98 0.65 Other Characteristics of Caregiver Health 1.37 1.34 0.98 1.25 1.87 ge 0.96 "' 0.95 " 0.94 “ 1.00 0.98 Caregiving Situation Duration of care 3.30 '" 3.26 ” 2.28 ' 1.91 - Demand of care 0.96 1.06 1.06 0.97 0.73 " ditionai helpers 0.96 0.97 1.68 0.64 0.56 Diagnosis: Alzheimers 0.86 1.63 0.40 ' 0.73 1.83 Care Recipient's Characteristics ender 1.32 2.81 " 0.63 1.55 0.13 Marital status 0.77 0.50 0.47 0.66 — Likelihood Ratio 392.28 270.67 256.28 307.27 51.01 Degree of Freedom 478 363 452 439 236 Probability 0.99 1.00 1.00 1.00 1.00 ”‘p<.001;"p<.01;'p<.05 Note: 1. Estimates of duration of care and care recipient's spouse not obtainable. 2. Coding scheme as follows. Gender: 1=female, 0=male; Adjusted Caregiver household income: in dollars: Occupation: 1= professionals. 0=else; Employment: 1=full time. 0=part time or not employed: Pr0pinquity: 1=live together, 0=live apart; Young children: 1=children under 15. 0=no children under 15; Educational level: in years; Health: 4=poor. 1=excellent; Age: in years; Duration of care: 1 > 2 years: 0= <= 2 years: Demand of care: # of functional dependencies; Additional helpers: in numbers; Diagnosis: 1=Alzheimer's, 0=cancer or physical impairment; Marital status: 1=married; 2=widowed. 136 The findings about power/resources, time availability, and gender ideology are overall consistent with those from the full-sample analysis. An egalitarian gender ideology is likely to decrease primary caregivers’ likelihood of participating in the housework domain, but not in other task domains, including the personal care domain congruent with women’s traditional reproductive roles. The impact of employment, a measure of work schedule conflict, remains, but added to it are the effects of caregivers’ competing family responsibilities and propinquity. Adult children with dependent children are less likely those without to take on all caregiving tasks (odds ratio=0.49). And, child caregivers who live with elderly parents are more likely than those who live apart to take on all task activities required for assistance and tasks in the housework domain (odds ratios=2.l and 4.9, respectively). Adult children’s household income, on the other hand, no longer exerts an influence in the housework domain; rather, the effect is found in the most time-demanding and labor-intensive personal care tasks. With regard to factors related to the caregiving situation, as expected, duration of care in the adult-children-only model display the same pattern found in Table 4.2, both sons and daughters tend to extend their involvement in total tasks and in tasks calling for hands-on assistance, although long-term daughter caregivers are still more likely than their son counterparts to take on these tasks.23 The effect of the demands of care found in the full-sample model receives only limited support after variables 23As noted earlier, the interaction effect of gender and duration of care was not statistically significant in the preliminary data analysis. Thus, regardless of the length of time primary caregivers have performed the caregiving role, daughters are more likely than sons to take on all tasks in the hands-on domains. 137 such as propinquity and dependent children are introduced. Finally, the diagnosis of Alzheimer’s is now associated with the lesser time-demanding domain of housework, in which caregivers whose parents are diagnosed with Alzheimer’s have lower odds of fully participating than do those whose parents have other diagnoses. As to care recipients’ characteristics, gender and availability of a living spouse do not play a significant role in caregivers’ participation in all types of task domains as some caregiving research has suggested (Dwyer and Coward 1991; Coward, Horne and Dwyer 1992; Horowitz 1985b). The presence of an elderly’s spouse does not relieve the task burden put on adult children.“ Nor is parents’ gender related to the range of tasks adult children provide. The only exception is the significant effect of parents’ gender in the personal care domain. Caregivers caring for mothers are 2.81 times more likely than those caring for fathers to take on all personal care tasks. The significant effect of parents’ gender pertains to the hypothesis of cultural taboo but needs to be examined further before a conclusion is drawn; as past research has showed, the taboo of intimacy is gender-specific only when it involves cross—gender parent-child dyads (Ahn and Gilbert 1992). In the section that follows, the cultural taboo hypothesis proposed by some researchers (Montgomery and Kamo 1989; Montgomery 1992) to explain the lower task participation of sons as compared to daughters is examined. 2"Given the odds ratio of 0.5, the insignificant effect of care recipients’ marital status could be a sample size problem. 138 Cultural Taboo and Adult Children’s Involvement in Personal Care Researchers who propose the cultural taboo hypothesis hold that personal care tasks that involve intimate contact between caregivers and care recipients raise taboo thresholds for adult children, particularly sons, as they provide care (Montgomery and Kamo 1989; Montgomery 1992). As suggested by Ahn and Gilbert (1992), however, the taboo against intimacy is gender-specific only when it involves cross-gender parent-child dyads. The data in Table 4.4 explore the taboo hypothesis among children with a sophisticated design. Data in the first model, drawn from the column of personal care in Table 4.3, specify no interaction term of gender of caregivers and of care recipients, that is, assuming there is no difference in the task involvement of daughters and sons regardless of the gender of the parent. The data highlight the relevance of gender of both parents and adult children, but how the effect of parents’ gender differs for sons and daughters is unknown. In the second model, the interaction term is specified and various contrasts are compared. If the cultural taboo hypothesis holds, the following contrasts of differences in adult children’s task participation in the personal care domain should be observed: (1) the range of task assistance provided to a father should be higher for sons than for daughters, and vice versa when care recipients are mothers; (2) daughters’ range of task participation should be higher when they care for a mother than for a father, and the task involvement level of son-mother dyads should be lower than that for son-father dyads; and (3) there is no difference in the range of task participation between daughter- father dyads and son-mother dyads as tasks involving bodily contact should raise the same level of taboo thresholds for cross-gender children caring for parents of the 139 Table 4.4 Odds Ratio of Child Caregivers Participating in All Tasks in the Personal Care Domain Model '1 Model *2 Predictors'3 1 2 'Caregiver's Gender (Female vs. Male) 8.56 m E. caring for father 8.59 " caring for mother 8.56 ”* ‘Care Recipient's Gender (Mother vs. Father) 2.81 *"’ cared by sons 2.82 mred by daughter 2.81 ‘ Cross-gender of Caregiver & Care Recipient 3.04 * (daughter-father vs. son-mother dyad) Caregiver's Gender X Care Recipient's Gender 0.99 Resources IT ime /Ideology 9 dj. Household Income 0.98 " 0.98 " Professional occupation 0.79 0.78 0.66 0.66 1.48 1.47 Dependent chlldren 0.70 0.71 Educational Level 1 .08 1 .09 Other Characteristics of Caregiver 1. 34 1 .34 0. 95 *' 0. 95 ’ lCaregiving Situation Duration of care 3.26 “ 3.26 *‘ Demand of care 1.06 1.04 ddltlonal helpers 0.97 0.97 ' Diagnosis: Alzheimer’s 1.63 1.63 Care Recipient' 5 Characteristics 0.50 0.50 leelihood Ratio 270.67 271.60 Degree of Freedom 363 362 1.00 1.00 *“p< .01; “p < .01; * p < .05 Note: 1. Estimates in Model 1 are obtained from Table 4.3 2. Model 2 contains interaction term of caregivers gender and care recipient's gender. Varous contrasts related to the hypotheses of taboo hypotheses are presented in the upper portion of the table. See Table 4.5 for how they are derived. 3. For coding scheme see Table 4.3 140 Table 4.5 Coefficients Used to Predict Task Participation for Child Caregiver’s Gender (CGender) by the Gender of the Parent for Whom They Care (PGender) in the Model 108 (OINTI CGender,PGender)=B0 + Bl*CGender + B2*PGender + B3*CGender*PGender, Where Son and Father are Base-line Groups Daughter Son Mother B0+B1+BZ+B3 BO+BZ Father BO + B 1 BO Note: The coefficients, B1 and B2, in the equation estimate not the average effects of caregivers’ and care recipient’s gender for the total sample but thse effects for certain subgroups, depending on which subgroup is the base line group. The estimated log-odds for various subgroups relevant to the test of the taboo hypothesis are as follows: Bl: Log-odds for daughter-father vs. son-father dyads (Cargiver’s gender effect when caring for father) B1+B3: Log-odds for daughter-mother vs. son-mother dyads (Cargiver’s gender effect when caring for mother) B2: Log-odds for son-father vs. son-mother dyads (Care recipient’s gender when cared for by sons) BZ +B3: Log-odds for daughter-mother vs. daughter-fatherr dyads (Care recipient’s gender when cared for by sons) Bl-B2: Log-odds for daughter-father vs. son-mother dyads 141 opposite gender. Odds ratios of appropriate contrasts that test the above hypotheses can be seen in the upper part of Table 4.4. Table 4.5 illustrates the coefficients used to predict task participation of various subgroups and the contrasts of differences for various subgroups one can generate~ from these coefficients. The data in Table 4.4 in a limited way support the cultural taboo hypothesis. The gender of parents affects the range of tasks with which daughters assist. The odds of daughters assisting with all personal care tasks are 2.81 higher when the care recipient is a mother as opposed to a father. The insignificant effect of care recipients’ gender --and in a reverse direction-- for sons (odds ratio=2.81, p > .05) may be in part a result of sampling fluctuation as there were only six cases of son- father dyads in the sample. Although the taboo hypothesis is partially supported, at least among daughters, all other contrasts provide evidence of the significance of adult children’s gender as the primary determinant of their participation in these intimate and yet labor-intensive tasks. Daughters participate in a wider range of tasks than do sons regardless of the gender of care recipients for whom they care (odds ratios=8.59 when caring for fathers and 8.56 when caring for mothers). The cross-group comparison of children caring for parents of the opposite gender reveals that daughters caring for fathers still have higher odds than sons caring for mothers (odds ratio=3.04) of assisting with all personal care tasks that involve intimate bodily contact. The differences in the range of task participation between daughter-father dyads and son-mother dyads raise questions as to why the bodily contact involved in the intimate personal tasks is not considered as much a taboo for daughters as it is for 80118 . 142 Concluding Remarks and Discussion In this chapter, the sources of variation in the range of tasks for which primary caregivers provide assistance to their frail relatives were identified. Multivariate analyses were based on a common set of variables shared by all caregiver groups and a larger set of variables applied to adult children’s caregiving context. Gender was treated as a sociodemographic background characteristic that affects group placement in aspects of task performance. The Intersection of Gender and Family Relation The Wald statistics showed that family relation is the foremost predictor of the range of tasks with which primary caregivers assist in all task domains, with the exception of tasks in the mobility domain, that are probably too physically demanding for elderly spouses to perform. The active involvement of spouses in care for an ill partner highlights the obligatory component and norm of reciprocity in the marriage contract. Therefore, it is not surprising that the task involvement of husbands is as extensive as that of wives, with the exception of assistance in the area of personal care tasks, in which the odds of participating in all personal care tasks are twice as large for wives than for husbands. Whereas spouses’ task involvement is likely to be prescribed by the normative expectations of marriage, the tasks performed by adult children depends on their gender and the gendered nature of tasks. Daughters are more likely than sons to help their parents with services calling for hands-on assistance but they are not likely to relinquish responsibility in the male-oriented task domains as sons are in the female-oriented task domains. In contrast to this divergence in caregiving behaviors between sons and daughters in female-oriented 143 task domains, a convergence of behaviors is observed in male-oriented task domains. Duration of Care Duration of care emerges in this study as an important factor in understanding caregivers’ task involvement, at least for the child caregiver group. Regardless of gender, adult children in long-term caregiving roles are more likely than those in short—term caregiving roles to be involved in all tasks for which assistance is required and in the labor-intensive and time-consuming task domains such as personal care and housework. In contrast, the length of time spouses have performed the caregiving role does not have any impact on the range of care they provide to their partners, except in the administration domain, a result that confirms the overall commitment of both husbands and wives to care for their spouses. The data also revealed an interesting pattern with regard to the type of tasks spouses and adult children are likely to undertake as their caregiving lengthens. Over times, both daughters and sons tend to progressively assume intense and time- consuming tasks but not necessarily in intermittent tasks that involve money management and transportation arrangement. In contrast, when providing long-term care, spouses are likely to assume responsibility in administrative tasks but are not more actively involved in labor-intensive and time-consuming tasks than are those new to the caregiving role. Long-term caregiving sons’ active involvement, as compared to the involvement of short-term son caregivers, to some extent contradicts the general conception about sons who provide care. It has been suggested (Stoller 1990) that sons usually relinquish primary caregiving responsibility or shift the burden in the area of hands-on 144 tasks to formal help or other siblings when elderly parents continue to need assistance. It is difficult to say whether the contradictory pattern, shown in this study, is due to the sample fluctuation in the small size of long-term son caregiver sample (N=15) or to their unique characteristics. Two speculations can be made about the unique characteristics of sons who provide long-term care. First, sons usually are not the designated primary caregivers in a family. Those who do stay in the caregiving role to provide long-term care to parents may have a very strong sense of role responsibility and hence play a relative active role in task assistance. Second, sons who are willing to be long-term caregivers tend to have less power/resources than their short-term counterparts. They thus become the designated primary caregivers providing long-term care to their parents and therefore are more actively involved in hands-on tasks than are those new to the caregiving role. As far as the first speculation goes, the data that examine the relationship between caregivers’ attitudes toward role responsibility and the duration of caregiving do not quite support the claim. Table 4.6 shows that across all caregiver groups, long- term caregivers, in fact, feel less role obligation toward caregiving than do short-term caregivers, and sons are not any different from spouses or daughters in this regard. As to the second speculation, some selected sample characteristics, presented in Table 4.7, partially support the claim. Long-term caregivers are older, somewhat poorer, and less likely to be employed full-time than are short-term caregivers. The differences, however, are not invariable among relational groups. Sons who care for parents for a longer period of time, 6.4 years on average, appear to be very different from the rest of the groups who provide long-term care. 145 Table 4.6 Sample Mean and Significance Test of the Index of Caregivers’ Felt Role Responsibility to Caregiving by Relationship of Caregiver to Care Recipient and Duration of Care Hus- Total Wife band D’ter Son Index (1237) (523) (264) (404) (46) Duration of Care Total Sample M 4.00 4.00 4.33 3.81 3.81 SD (.66) (.64) (.54) (.68) (.62) < = 2 years M 4.06 4.09 4.39 3.83 3.87 SD (.62) (.62) (.52) (.64) (.61) > 2 years M 3.89 3.85 4.21 3.76 3.66 SD (.67) (.65) (.55) (.74) (.66) Test of Significance F Value Gender 6.64** Family Relation 36.30*** Duration of Care 8.82** Gender X Relation 9.25** Gender X DC 0.14 Ralation X DC 0.37 G X R X DC 0.81 ’” p < 0.001;”p < 0.01; ’p < 0.05 Note: See Table 3.4 for the items constructed for the index of caregiver’s role responsibility to caregiving 146 .3 8.: 3.: .v: .3 .3 .3 .3 .3 2.: am ”N 1N 9N EN N.N N.N ad «N SN SN .2 29.3.5... 05 c. 9598 .0 .oz .5: .N.: .5: Cu: 8.: 3.: .m: 6.: .m: 6.: am a... Q: a... oé N... a... N... .6 N... .3 2 .N. 9.8... 3.322 8. .3 .3 .3 .3 .3 c... .3 c... 2... am o.N v... N... o... a... n... Hp v... .3 tr .2 .N. 9.3.. Sausage“... .3 .3 .3 .5 c... .3 .3 .3 .3 .3 am o.N 9N oN EN EN 9N oN «N ed RN .2 .m. v3.6.25... RN. 6.: 2N. ._..N. :N. 8.: :3 3.: 3N. 8N. om o6 v.9 a.» 9N ad a... a... 9... ed N .2 .m. 9.3... 88 _ucoflod 1 .3. .3. .3. .3. .3. .3. .3. .3. .3. .3. on ad. N.» «.3 ad «.3 d... 9.: o.» «.3 a.» .2 .8. 5.5.8 .30... 5:35.30 .3325“. .3. 8.... .3. 8.3 8.8 .0... .5. .3... 8.... as. aw N» as 3 t. 2. so K 2 2. N» 2 oo< L. x o 2 l l l l m o tees tsetse .eoso 5.; scan. 9. mN em on D 3 «N 8 Nm «m «.550 ecu c9693 Emacs. mm cc 3 3 S 8 S 5 S on S 8 mm vm one..." .9528 new Eugen. S 0N S N S a. S «N l. l. I. .I S m S a 9.2m Evans. E0838? 9.3... N 3 m I l. l. r. l. N N 3.5.: 552353368520 on mm mm as .l l. r. l. on .N 338.2. S 9 S m. S o S m. S o... S o... S 8. S o... S 8 S we no.5: . was.» .9th on 3 3m 2. l. .l r. .l 3 Nm o_mEo... S 8 S 2. S 3 S .N . .l l r. l. S E S 3 2m: .3: m. mm mm. Rm om mm. 8. in 83 new momma .0 Lone”— E» NA 2.. N v 2.. NA 2.. N v 8» NA 2.. N v 8.. NA 8.. N v 2.. NA 8.. N v 8.8.23. 23 .o 3.3.3882. eem 55:3 2.32.: 8.3 _so» 28 .o cos-Lao e:- ecoaaem >_.Eeu an 325530 295m egos-em N. v sink 147 .882. 2.5 99.8.8 8.2.8 .58 a“. . i .8. .8: .8. : .38. 3... .8. .5. .8: .8. .8: am .2528 8...... .o .2. a. 828.8 N83 .8. 3.8 .8 «.8 N8 «.3 EN 3.8 33 .2 .8888 se. n.a.. 32.5.. .o S .88. c... .3. 6.... 3.... a... .3. .8. 8.3. .3. am 3 3.3 .8 .3... 3.2 .3 3.2 .8 3. 3.3 s. .8 .2. 28 .o 2.6: .88: .88: .88: 85.: .88: .833: .88: .88: .88: 8.8: cm fi 3.8 8.8 88. 838 238 23a 88. 88. 338 8 SN .2 2.8... 3.9.8.5.. 8.8.9. .3. .3: 8.... .3: .3... .8... ..... .8... .3: 6.9. am No 33 No 3 3. 8 .3 8 8 8 .2 39. S n. S .o S N. S 8 S N. S No L988. 8...... 8.....an S m. S a. S 8 S N.” S m. S o. .8. .03.... 8.3.3.0 8.. m. o. NN m. .I .I I. .I w .. 3.230332020303093 8 m3 8. m. I l I I m . 8.8... .962 S .3 S .3 S No S .. S 8. S 8. S 8. S 8. S 8 S 8 8E9... 8.8 .38... n. m a. 8N . m 3N 3N .. .N 8.0.an 8: . m o m m t . m N . .350 . 8 m m 8 N. o o m m. .33. $8.30 .3 3N N3 .8 NN o. .3 8 03 .3. 83m. .8320 S .N S 8 S 8 S 8 S 33 S 8 S 8 S 2 S .m S NN 8583.9. 32.3380 S 8 S No S .o S 8 S 8 S N3 S 83 S 8 S 8 S 33 82.8 9:8 .82 .3 .96. 32.838 8 m. .N O3 3 3 .. 8 m. 8 8.2qu 62 .N m. NN m. m. m. 8 83 38 8 88¢ 8 o. m. . . . . 3 . o N. m 35.38... S 8 8 S N.” S 8 S o. S m. S m S o. S o. S «N 05.1.8 .coE..o_..Ew a... NA a... N v 2.. NA 2. N v 3.. NA 8.. N v 8. NA 8» N v 2» NA 8. N v 23.15123 3 3.3.8023. 5m .3336 2.32... as. .58 €2.00. 5.3 03a... 148 Contrary to the patterns found among spouses and daughters, sons who provide long-term care are less likely than those new to the caregiving role to be college educated and to hold a professional occupation. In contrast to daughter caregivers, sons who provide long-term care are more likely than those new to the caregiving role to be older, employed part-time, retired, or unemployed, and thus considerably poorer. About six-tenths (59%) of short-term caregiving sons are employed full-time as compared to only 33 percent of long-term caregiving sons. The gap in household income, $3,700, between long-term and short-term caregivers in the son group is also the largest among all subgroups. Compared to long-term caregiving daughters, sons who provide long-term care are in a somewhat disadvantaged socioeconomic status. Therefore, it is not surprising that long-term caregiving sons provide substantially more hours of care than do sons in the short—term caregiving role (9.3 vs. 5.9) and their hours of care approach the level provided by their daughter counterparts (9.5). In addition, the number of hours of help long-term caregiving sons receive add up to only half (48.2%) of their contribution, which is much lower than the proportion for their short-term counterparts (75.8%) and even for daughters in the long-term caregiving role (56.4%). Therefore, the low socioeconomic status of sons who assume long-term caregiving role may be one explanation for their relatively active involvement in hands-on tasks. Power/resources, Time Availability, and Gender Ideology Gender differences in power/resources, time availability, and gender ideology have been suggested in some caregiving research (see Finley 1989 and Montgomery 1992) as plausible reasons for the bivariate gender differences of task participation 149 observed. The strong and prevailing effects of gender, supported by both the Wald statistics and significance tests of logit coefficients, in the female-oriented task domains after all other known effects are controlled, tend to refute such claims. The range of task participation primary caregivers perform is primarily dependent on who performs the tasks and the type of tasks that are performed rather than the power/resources, time, or gender belief caregivers have. Power/resources, time constraints, and gender beliefs indeed independently affect the range of tasks caregivers perform, but more in the housework domain than in the personal care domain, which, as compared to the former, is intensive and time- demanding and congruent with women’s nurturing role. Constraints of a combined workload, be it work schedule conflicts or competing family responsibilities, affect how caregivers assist with housework. Caregivers who possess social resources experience more power in family negotiation and, consequently, participate in fewer hands-on housework tasks than those without such power/resources. The attitudes caregivers learn through the socialization process about women’s and men’s place in the household and in the market place influence the way they participate in tasks demanded of certain gender roles. The overall lack of support for these hypotheses in the personal care domain is probably because, unlike housework that is an extension of routine household maintenance in a caregiving setting, caring for a relative in need of help with personal care takes place only at a time of family crisis.” They are, 25It is true that aging and dependence of elder relatives are "normal" life events, not a "crisis.” Nevertheless, the need for help with personal care by care recipients often suggests the substantive deteriorating health, thus calling for a great deal of caregivers’ obligation to provide care to the ill parents. 150 therefore, more likely to be performed out of sense of obligation than are housework tasks and thus not particularly subject to structural factors such as employment or time or gender beliefs. Because the sense of obligation of adult children is not as strong as that of spouses, adult children may tend to negotiate their task participation on the basis of their power/income in the family and to relegate tasks to other formal or informal helpers. This explains why the effect of household income in the personal care domain is significant among child caregivers. Nevertheless, the limited support for power/resources and gender role beliefs in domains such as personal care or household administration, should be read with caution since the operationalization of the variables is problematic due to data limitations. Occupation is measured by the occupation caregivers held at the time of interview or before retirement. Moreover, occupations classified within broad categories are subject to possible errors in the classification of the level of authority associated with an occupational position and in the assumption of a carry-over effect for pre-retirement occupations. Power and resources are measured by the income of the households in which primary caregivers reside; this operationalization assumes that primary caregivers, regardless of whether or not they earn income, have access to the income of the family members with whom they share a household. This broad assumption may have contributed to the ambiguous effect of household income found in this study. In addition, power is situated in every interaction in which it occurs (F ishman 1978). The absolute measure of household income, or even earnings if available, ignores the contextual meaning of power in a relative sense, that is, the power/resources of a primary caregiver vis-a-vis other family members who could be 151 potential primary caregivers. On the one hand, power may be best conceptualized, for marital dyads, as the ability of caregivers to mobilize resources and to purchase service and thus the more the absolute earnings of primary caregivers, the more likely they are to relegate tasks to paid helpers. On the other hand, for child caregivers, power might be conceptualized as the ability to negotiate within a family dynamic; that is, the majority of child caregivers negotiate with earnings not so much to purchase paid help as to bargain with other family members over who does what.26 The relative power of primary caregivers thus may be a more meaningful measure for the testing of the power/resources hypotheses, especially among child caregivers. Education as a proxy measure of gender ideology, as Ross (1987) points out, may be questionable because education may have an effect over and above gender-role attitudes. In addition, as mentioned above, education may not be a proper proxy for gender role ideology of elderly spouse caregivers, who tend to adhere to traditional gender ideology regardless of their educational level. A direct measure of gender-role ideology is undoubtedly preferable in future studies. It should also be noted that whereas this study showed the independent, direct effects of structural factors or personal belief systems on task participation, the effects of these factors could be gendered. For example, in domains traditionally defined as women’s responsibility, men who hold an egalitarian ideology should take on more tasks while women who voice an egalitarian viewpoint should assume fewer. In addition, the conception of time is often qualitatively different for men and women. 26In a national survey (Stone, Cafferata, and Sangl 1987), less than 16 percent of primary child caregivers provided care with any kind of formal or paid help. 152 Whereas men construct their time in a person-based and work-oriented fashion, women’s conception of personal/family time and work time are intertwined (Hessing 1994; Seymour 1992; Chambers 1986). As a result, men deal with their multiple responsibilities by reducing their volume of care whereas women deal with them by maintaining rigid schedules, negotiating care tasks around their work schedule, and giving up their own free time (Cantor 1983; Horowitz, Sherman, and Durmaskin 1983; Lang and Brody 1983). Whereas men’s power is direct and effective, women’s power/resources need to be mediated through egalitarian gender beliefs to be effective (Ross 1987; Hardesty and Bokemeier 1989). Therefore, a model in which power/resources, time, and gender role ideology are juxtaposed with gender without specification of interaction effects may be problematic. Although this study tested the interaction effect of role ideology by gender and found it insignificant, the small number of sons and poor measurement of gender role ideology prevent this study from providing a reliable result with regard to the differential effects of gender beliefs by gender. Future research with a large number of son caregivers and a better measurement of gender role ideology should be able to provide more evidence of the gendered effect. Finally, why gender differences, and thus inequality, persist, why the gendered nature of power/resources, time, and role ideology in fact perpetuate gender inequality in caregiving, and why it is problematic to conceptualize gender as a sociodemographic background variable that affects group placement in caregiving will be discussed in greater detail in Chapter 6. 153 The Multi-dimensionality of Caregiving Tasks Family caregiving, as defined in this study, is physical labor performed in a finite set of tasks for which assistance is required by impaired care recipients. Caregiving tasks are grouped into four task domains according to the degree of intimacy, time constraints, labor-intensity and, particularly, gender role expectations of these tasks. As expected, the effects of gender and family relations appear to vary according to the type of tasks caregivers perform. Thus far, the data show that all task domains are likely to produce different gender- and relationship-based patterns of task participation. For adult children, the divergence of caregiving task involvement between sons and daughters is likely to be observed only within domains that are extensions of women’s nurturing and reproductive roles. In contrast, spouses have less discretion about what tasks to assume, with the exception of mobility tasks, such as lifting and turning in bed or moving in and out of bed, in which the labor is probably too physically demanding for older spouse caregivers (who averaged 67 years of age in this sample) to perform. In addition, although both are repetitive, tedious, and are done on a routine basis, personal care and housework differ in the degree of control over time and physical place as well as in the demand for intimate contact. The normative expectations of marriage and prescription of gender roles are thus more likely to operate in the personal care domain than in the housework domain. Gender and relational differences in the task participation index are thus, as expected, more pronounced in the personal care domain than they are in the housework domain. That is, female caregivers (as compared to male caregivers) and spouse caregivers (as 154 compared to child caregivers) are more likely to take on responsibility in the personal care domain than they are in the housework domain. Cultural Taboo The cultural taboos thesis holds that personal care tasks involve intimate contact between caregivers and care recipients, and, thus, raise taboo thresholds for adult children as they provide care. The taboo hypothesis is confirmed, first, by spouses’ relative reluctance to yield to other caregivers tasks that involve bodily contact with their ill partners, and second, by adult children’ tendency to withdraw from participation in personal care tasks more than from housework tasks. Nevertheless, the validity of the taboo hypothesis is largely discounted by the differential pattern of caregiving observed between husbands and wives. Although spouses are indeed more likely than adult children to provide help with personal care tasks, gender remains the organizing principle in spouses’ assistance in tasks infused with intimate content. Although the intimate nature of personal care tasks discourages spouse caregivers from relinquishing responsibilities, the dual nature of personal care tasks --intimacy and labor intensity-- cannot be ignored. The intense labor and time that are required in assisting with these tasks, the equation of personal care tasks with nurturing, a feminine quality, may explain why husbands are more than twice as likely as wives to turn to other caregivers for help with tasks such as bathing or dressing. Taboos may indeed be a concern for children caring for parents of the opposite gender; the data show that daughters who care for fathers are far less likely than those who care for mothers to take on tasks such as dressing, grooming, and bathing that entail bodily contact. Nevertheless, whereas taboo is one of the driving force of adult 155 children’s participation in personal care tasks, the gender of caregivers is also an important organizing principle of such caregiving. When caring for parents of the opposite gender, sons are much more inclined than daughters to avoid participating in personal care tasks, thereby raising questions such as why the taboo is a concern for sons more than it is for daughters and the extent to which the taboo hypothesis can be used to account for sons’ less committed involvement in these tasks as Montgomery and Kamo (1989) suggested. Other Factors Finally, the age and health condition of caregivers and care recipients prove to be relevant to the range of tasks caregivers perform, especially when assistance is required in the labor-intensive, time-consuming or physically-demanding domains. The older and the poorer the health of caregivers, the smaller the range of tasks with which they can assist. The increase in the functional dependency level of care recipients, and thus the demands of care, constrains caregivers’ ability to provide all care themselves and, thus, increases the need to call in other helpers to take on some of the tasks for which assistance is required. The role that diagnosis plays in caregivers’ task participation is inconclusive. Caregiving for Alzheimer’s patients has been described as a stressful, fatiguing, and time-demanding experience (Novak and Guest 1989); and their caregivers are often required to provide round—the-clock care in order to maintain control over every segment of patients’ behaviors. In the full- sample analysis, caregivers caring for an elderly relative diagnosed with Alzheimer’s are indeed more likely to provide intensive and time-demanding personal care tasks than those caring for relatives with other diagnoses. In the adult-children-only model, 156 however, the effect is observed not in the personal care domain but in the housework domain «and in a reverse direction. Adult children caring for Alzheimer’s parents are less likely than those caring for parents with other diagnoses to take part in labor- intensive housework tasks. The inconclusive finding may be related to the way diagnosis is related to the services provided by formal organizations such as hospital discharge plans and community out-reach services. Research with a special focus on diagnosis and caregiving labor should take into account the complex relationship of diagnosis with the availability of formal organizations and with the type of services they offer to caregivers. Summary In this chapter, multivariate logistic regression models were employed to explore the source of variation in primary caregivers’ range of task participation. Consistent with the patterns found in the preliminary data analysis in Chapter 3, the intersection of gender and family relationships indeed influences the way caregiving labor is carried out, especially in the labor-intensive and time-demanding task domains. Governed by the normative expectations of marriage, spouses show greater commitment to caregiving even after a long-period of providing care and after the deterioration of care recipients’ health and have less discretion about what tasks to take on than do adult children. Nevertheless, the normative expectations of marriage are not the only the organizing principle of spousal care; gender continues to influence care patterns, as seen in the greater likelihood of husbands than wives to relegate to other helpers personal care tasks that are most labor-intensive and yet infused with intimate contents. As to adult children, sons’ task participation is 157 prescribed by stereotypic gender roles whereas daughters tend to adopt the roles of both care providers and care managers, working in areas congruent with their reproductive roles while extending their care to tasks in male-oriented domains. Bodily contact and thus intimate contents inherent in the personal care domain indeed impose constraints on adult children’s task assistance. Nevertheless, the taboo hypothesis may not be sufficient to explain why there is gender gap between caregiving sons and daughters even after parents’ gender is taken into account, and why taboos appear to be more a consideration for sons than for daughters. The prevalence of gender and relational effects, to some extent, throw doubt on the plausibility of explanatory frameworks such as time availability, power/resources, and gender ideology that are used to explain gender differences found in the bivariate analyses of caregiving research, although these commonly accepted hypotheses receive their own independent support in domains such as housework. Why does the gender effect prevail despite the fact that all known confounding factors are controlled? What mechanisms produce differences in the way men and women provide care to their relatives? Can gender be treated as a sociodemographic variable independent of the effect of power/resources, time, and gender ideology? Are their effects gendered? All these questions will be addressed in Chapter 6. CHAPTER 5 THE DIVISION OF CAREGme LABOR Research on gender and caregiving labor has predominantly focused on primary caregivers’ level of involvement. Rarely has it centered on how caregiving labor is divided between primary and secondary caregivers. The range of tasks primary caregivers perform and the division of labor between caregivers are two related but conceptually different phenomena. Two primary caregivers may take on the same number and types of tasks, but how these tasks are divided between primary caregivers and other caregivers may be very different from one caregiver to the other. The handful of studies that have data on the support primary caregivers receive tend to focus on spouse caregivers’ care experience and to report either that spouses receive the least support from secondary caregivers (Tennstedt, McKinlay, and Sullivan 1989; Miller and McFall 1991; Horowitz 1985b) or that husbands receive more outside help than do wives (Allen 1994; Stoller and Cutler 1992; Zarit, Todd and Zarit 1986; Noelker and Wallace 1985). Nevertheless, no research has systematically explored (1) how a division of labor varies according to the gender of adult children and to the type of tasks required for assistance, and (2) if there is also a differential effect of gender by family relationship, such as the confounding relationship found in the index of task participation. Two indices of the division of labor, outlined in Chapter 2, are constructed to capture the complexity of the caregiving labor provided to the elderly. The task 158 159 exclusivity index measures the extent to which task activities are performed by primary caregivers without the involvement of other caregivers. The relative contribution index gauges the total contribution of primary caregivers relative to the efforts of all other caregivers. Multivariate logistic regression technique is employed to explore the extent to which the division of labor is based on gender and family relation. This multivariate analytical approach also tests hypotheses about power/resources, time, and gender ideology, under the assumption that these effects operate the same for men and women. Nevertheless, as mentioned in Chapter 2, additional testing of the differential effect of gender role ideology by gender will be presented in a separate analysis to redress the deficiency of the model that specified no such interaction term. The findings are presented in two parts. The full-sample and adult-children—only analyses of the index of task exclusivity for total task activities and for four task domains are first presented. Then, the factors associated with primary caregivers’ proportionate volume of care are identified and presented. Task Exclusivity and the Incorporation of Helpers in Task Assistance Caregiving research (Enright 1991; Jutras and Veilleux 1991) that examines primary caregivers’ hours of care or level of task involvement generally concludes that husbands, like wives and daughters, are likely to abandon gender roles in their task assistance, whereas sons’ task involvement is prescribed by stereotypic notions of gender roles, a conclusion largely supported in this study. The other dimension of caregiving work, the division of labor, however, has been largely ignored. Although research based on proportionate volume of care has suggested some relationship- and 160 gender-based divisions of labor (Tennstedt, McKinlay, and Sullivan 1989; Miller and McFall 1991; Allen 1994; Stoller and Cutler 1992; Zarit, Todd and Zarit 1986; Noelker and Wallace 1985; Antonucci and Akiyama 1987; Pruchno 1990), still little is known about how tasks are allocated between primary and secondary caregivers or the degree to which tasks are shared according to the gender and family relationship of primary caregivers and to the type of tasks that they may assume. The index of task exclusivity, constructed for this study to measure the degree to which tasks required for assistance are performed exclusively by primary caregivers, is first presented to illuminate the complexity of the multi-dimensionality of caregiving behaviors. Three collapsed categories of the task exclusivity index are as follows: (1) exclusive care by primary caregivers, i.e., primary caregivers perform all tasks without any outside help; (2) partial task sharing, i.e., primary caregivers are involved in at least one task in a given task domain without others’ assistance; and (3) total dependence of primary caregivers on secondary caregivers, which means that primary caregivers do not perform any task alone or that they turn over all the tasks in a given task domain to secondary caregivers. In the analysis, polytomous logistic regression models are used since the outcome has more than two categories. For a dependent variable with three categories, two independent logits can be modelled. With total dependence as the contrast category, one logit is formed from the odds of exclusive care versus total dependence, and the other from the odds of partial task sharing versus total dependence. Values of odds ratios exceeding 1.00 indicate an independent variable raises the odds of primary caregivers being exclusive care providers or partial task 161 sharers (compared to total dependence), while values less than 1.00 indicate a decreased odds. The odds ratio for the third logit, exclusive care versus partial task sharing, can be obtained by exponentiating the difference between the logit coefficients estimated for the first two logits, or, simply equaling the ratio of the two odds.27 The discussion of odds ratios, significance test, and Wald statistic proceeds in the same manner as the discussion in Chapter 4. The full-sample and adult- children—only analyses are presented to identify the factors associated with primary caregivers’ task exclusivity patterns, with a special focus on the complex relationships among gender, family relationship, and task exclusivity. Full-Sample Analysis of Task Exclusivity Using a group of variables common to all caregiver groups, multivariate polytomous logistic regression analyses for all tasks combined and for four task domains are presented in Table 5.1. As in the models for the task participation index, both gender and duration of care were allowed to interact with family relation. The other possible interaction term of gender by duration of care is not considered because of the trivial effect it had in the preliminary data analysis. That is, the relationship between gender and task exclusivity is not likely to change when primary caregivers are long-term or short-term care providers. To simplify data presentation, only odds ratios are reported. Since interaction terms are specified in the model, estimated parameters of the dummy variables no longer represent the main effect for the total 27Use data in Table 5.1 as examples. With regard to the gender effect among spouse caregivers in the total task activities, the odds ratio for exclusive care versus partial task sharing is exp(ln[B,])-ln[B2]), that is, exp(ln[2.43]-ln[1.3])=1.80 (or 2.42/1.30=1.80). 162 §§8§.0§§0.3.oa§fiw ”30995808235000.9503". 00208389308020.2833“; .35.§§§.§0uvno.§un . .§B§.§.§u. .83.... .2625: 598.052.039.333... .. .802 00.v0...0.v0....00.v0... 00.0 00.0 00.0 00.0 00.. .. . 00.0.0 00.0 0000 000. 0000 0.000203 _ 00.000 0.000. 0.0000 0.0... 00.0000 0.0. 0859.... 0.0 .0. .. 00.0 00.0 .. .00 00.. ...0 0.0 .. 00.0 0.30.000... 0.800.... 2.. 000 . 0.0 t. 00.0 00.0 t. 00.0 00.0 .. 00.0 .. 0.0 s. 00.0 0.00.00 .0000... . .: 00.0 a. 0... .. 00.. .0. .3 000 00.. s. 00.0 t. .0. t. 000 0.8.0 000.000 0002060520200. 00.. . 00. E 00.. z. 00.. E 00.. 00.. 00.0 .. 00.. .. 00.. .7 000 0... .00 00.0 .: 00.0 .00 00.0 0... 00.0 _ 8000.8 .0 0033.00.08 .005 8.. 00. 00.0 .. .00 c .00 00.. 00.0 00.0 00.0 .23 3000080 _ .. 000 00.. .00 00.0 3. 00.0 . .00 t. 00.0 , .00 .: .00 00.0.0.0... 00.. .0. 00.. 0... 00.. 0... 00.. 0... . .0. 000008802 .00 0.. .00 00.0 00.0 00.. 00.. 0... 00.0 00000808050 00.. 00.0 00.0 000 00.0 00.. 00.. 00.. 00. 0.. 00.. 00.0 0... 00.. 00.. 00.. 0... 00.. 00.. .0. 00.. 00.0 .. 00.0 .. 00.0 00.. 00.. .0. 00.000. 0.000000: 289302053233. 0... 0.. 0.0 00.0 00.. .. 00.0 . .00 .0. .0. 0.00.0 0000.00 x 0000.00 ...0 00.0 00.. .00 .0. 00.0 00.0 .0. 00. 0000.00 x - , .. 00.0 . .. 00.0 .. .00 .00 .. 00.0 00.0 .. 00.0 00.0 00.0 000000... 0000.0 . 00.0 .. 00.0 0.0 .. 00.0 .. 00.0 .0. 00. E 00.0 .. 00.0 00.0.0.0 0000 000.00 .0-th .00. has... 200 .0 02.050 00.. . 00. 00.. 00.0 00. 0.0 0.. . 0.0 00.0 .000 0.0.00. 002.08. 0000.0 00.. 0... 00.. 0.0 00.. . 0.. E 00.0 000 .00 .000 0.0.00. 002.00.... 000.00 0.0 00.0 00.0 .00 00.. 0.0 .00 000 00.. .000 0.00. 003.08. 000.00 00.0 00.0 .00 00.0 00.. 00. .. .00 00.0 00.. .000 0.00. 009.00.... 000.00 .030 .00 000000. 0000.00 0.0.00 000 . 00.0 00.. . 00.. t. 00.0 . 0... :. 00.0 00.. .. 00.0 0000000 000.00 00.0 00.0 0.0 .. 00.. .. 00.0 00.. .. ...0 000 .0. 00.0.0.0 .000 000.00 .200. .00 0.00.00. 2 8200 00 .000 .mulo .80.. .08 30.. man .80. mud .30. fascia .2. .2. .2. .2. .2. 28 000m 220 .005. 28 0000. 200200000. 28 .00.“. 2000 3:0... 28 .3 £000 3...... :8 000m . luggllll 0§<20530= 0.03030: 28.0002... 0000.. 30.. I I §§.£§0§§..§80§0§¢g333 .0030. 163 sample; rather, they represent the effects for a certain subgroup, depending on which subgroup is coded the dummy variable. Odds ratios for all relevant contrasts are presented in the upper portion of Table 5.1. (See Table 4.1 for they way they are derived from the parameters estimated in the model.) Gender Stereotyping for Spouses and Adult Children The data in Chapter 4 on task participation showed that in contrast to sons’ task involvement, which is prescribed by stereotypic notions of gender roles, husbands are likely to abandon their gender roles when they provide care. Thus, like wife and daughter caregivers, they are involved in a wide range of caregiving tasks in both male- and female-oriented task domains. Nevertheless, when caregivers’ efforts are viewed in terms of networks of support, husbands’ commitment to caregiving becomes less impressive; gender is a significant predictor of spouses’ task exclusivity patterns in the total task activities and hands-on tasks such as personal care and housework. For example, in the personal care domain characterized by intimacy and labor intensity, the odds of wives versus husbands being exclusive care providers (as opposed to total dependence) are 2.66. The odds of having other caregivers share at least one of the tasks (as opposed to total dependence) are 1.78 times as high for wives as for husbands. Similarly, in the housework domain, the odds that wives will be exclusive care providers or partially share tasks (as opposed to total dependence) are, respectively, 2.99 and 1.58 times higher than for husbands. Wives also have higher odds than husbands of being exclusive care providers in all care activities combined (odds ratio=2.34). 164 With regard to the masculine mobility tasks, on the other hand, elderly husbands use other caregivers’ help no less often than do elderly wives. In the administrative domain, that is congruent with male gender roles, the positive relationship between gender and task exclusivity, though not fully supported in both sets of logits, is somewhat unexpected (odds ratio=2.04, p< .05 for partial task sharing vs. total dependence). Wives are still more likely than husbands to participate without help in tasks involving decisions and control, responsibilities traditionally assumed by men.28 As for adult children, although daughters and sons incorporate helpers in total activities in a similar way, gender is a significant predictive factor in two female- oriented domains. In the housework domain, daughters have higher odds of being exclusive providers or partial task sharers than are sons (odds ratio=2.80 and 1.82, respectively). In the personal care domain, daughters are also more likely than sons to be exclusive care providers as opposed to totally dependent on others (odds ratio=4.77). In the administrative and mobility domains that are extensions of male gender roles, on the other hand, the gender differences between adult children, seen in the preliminary data analysis in Chapter 3 (see Table 3.11), are not supported in the multivariate analysis. Sons are not any more likely than daughters to participate in tasks alone (odds ratios=0.72 and 0.88 for the administration domain, and 0.43 for the mobility domain, and p> .0 .29 28It is noted that whereas money management is a task largely assumed by men in this society, organizing transportation service is somewhat gender-neutral. 29The non-significant odds ratios are likely due to the small subsample of sons. 165 In all, when other caregivers’ presence is taken into account, the data show that wives (as compared to husbands) and daughters (as compared to sons) are more likely to assist with tasks traditionally defined as women’s domain by themselves. Yet, husbands and sons are not any more likely to take on tasks traditionally defined as men’s domain by themselves. Adult children’s caregiving behaviors in terms of both task performance and the configuration of support they draw upon are likely to follow stereotypic ideas about gender roles. Whereas the task involvement of husbands and wives remains constrained by the normative expectations of the marriage relationship, the allocation of tasks is very much gendered for spouse caregivers. Family Relationship and Use of Suppgrt Tennstedt, McKinlay and Sullivan (1989), in their study of the role of secondary caregivers, conclude that secondary caregivers are unlikely to be involved in care when the primary caregiver is a spouse. It is unknown whether an analysis based on the degree of task sharing, rather than the quantity of care received by primary caregivers, will produce the same conclusion. Contradictory to the patterns of task exclusivity in the preliminary findings (see Table 3.11), the data in Table 5.1, based on multivariate analysis, show that most of the odds ratios associated with the relational effect are statistically insignificant. Husbands do not differ very much from sons, and wives from daughters, in their tendency to use the assistance of others, especially when there are others in their helping networks.30 3OFor total tasks combined and the housework domain, for example, the odds ratios for the logit of partial task sharing versus total dependence are less than 1.0 (although p > .05), suggesting that when there are others in support networks, spouses are not any less, if not more, likely than adult children to ask for help with the task in which they assist. 166 The only exception to the lack of relational effect occurs when short-term caregivers assist with personal care tasks. Husbands new to the caregiving role have estimated odds of assisting with personal care tasks without help that are 6.87 times the odds for sons. Similarly, short-term wife caregivers have much higher odds of being exclusive care providers or partial task sharers in the personal care domain than are daughters (odds ratios=3.83 and 1.78, respectively). As their caregiving career lengthens, however, such relational effects dwindle. When they perform the long-term caregiving role, husbands and wives are not statistically more likely than adult children to be exclusive care providers in tasks that involve intimate content.’1 The differential effect of family relation by duration of care in the personal care domain is confirmed by the significant statistical test (odds ratios for two sets of logits=0.47 and 0.39). In other words, compelled by the normative expectations of marriage, spouses are more likely than adult children to be exclusive providers in tasks that require assistance. Nevertheless, when their task assistance is considered within the context of networks of support, spouses, especially long-term spouse caregivers, are just as willing as adult children to accept help with tasks, including assistance with intimate personal care tasks. Duration of e T k Sh in In Chapter 4 on task performance, the data showed that duration of care increases the likelihood that child caregivers will take part in tasks that are labor- 3‘Again, the lack of the support for the relational effects among long-term male caregivers, such as in the personal care domain (odds ratio=3.21 for exclusive care vs. total dependence, p > .05), may be a result of the small number of sons. 167 intensive and time-demanding. Spouses’ range of task participation, in contrast, has no association with the length of time they have performed the caregiving role. When the analysis moves from caregivers’ range of task participation to task sharing, duration of care becomes a significant predictor for both spouses and adult children in all task domains except personal care. The data in Table 5.1 show that, regardless of gender and family relationship, long-term caregivers, although widely involved in tasks required for assistance, are less likely than short-term caregivers to perform tasks without the assistance of others. In the total task activities, for example, the odds that long-term spouse caregivers will be exclusive care providers or partial task sharers are, respectively, 0.38 and 0.39 times, the same odds for spouses who are new to the caregiving role; the odds ratios for adult children in the long-term caregiving role versus those new to the caregiving role are 0.35 and 0.29, respectively. Most importantly, the negative relationship between duration of care and task exclusivity is not task-specific. Long- term caregivers have an increased need for help ranging from hands-on task domains to intermittent task domains. The only exception to the significant effect of duration of care occurs in child caregivers’ assistance with personal care tasks (odds ratios=1.08 and 1.61, p> .05). As noted above, long-term child caregivers are likely to share every task in a given task domain with, or even completely turn the chores over to, secondary caregivers. Nevertheless, in the most time-demanding task domain of personal care, they are not necessarily totally dependent on others for help, a finding that echoes the conclusion in Chapter 4 about the active involvement of long-term child caregivers. 168 In all, although adult children who provide long-term care tend to embrace a wide range of tasks, they are unlikely to perform caregiving chores alone. Whereas the length of time spouses have performed the caregiving role has little to do with what tasks they will take on, spouses, like adult children, tend to share tasks with other helpers, including personal care tasks that involve intimate contact with care recipients, as their caregiving career lengthens. Power Resources. Time Availability and Gender Ideology Power/resources, time availability, and gender ideology have been suggested as alternative explanations for the gender differences found in univariate analyses of caregiving behaviors. According to Finley (1989) and Montgomery (1992), proponents of these perspectives (Berardo, Shehan and Leslie 1987; Coverman 1985; Ross 1987) postulate that tasks, especially hands-on, women’s tasks, are likely to be performed exclusively by primary caregivers who have few power/resources to bring to family negotiation and to the purchase of service, hold gender stereotypic ideology (particularly women), and have few time constraints imposed by work or their own family responsibilities. Because women are less likely than men to be in the labor force and to possess social resources, and have been socialized to nurturing roles, they experience fewer competing demands and role conflicts, possess less power in family negotiation and purchasing formal service, and show more family concerns than do men. They, thus, are likely to adhere to traditional role expectations and to be engaged in time-consuming and labor-intensive nurturing tasks without help. If these hypotheses hold, first, the gender effect on task exclusivity should attenuate to a great extent. Second, primary caregivers who have few power/resources, hold gender 169 stereotypic ideology, and, especially, have free time are likely to perform tasks alone, particularly when the tasks involved are labor-intensive and time-consuming. Similar to the findings on range of task participation, the significant gender effect for both spouses and adult children, to a great extent, discounts these hypotheses as alternative explanations for gender differences in caregiving. Nevertheless, the data also show that power/resources, time availability, and gender ideology independently exert influences on how tasks required for assistance are allocated within total task activities and hands-on task domains. Time constraints due to full—time employment decrease the odds of caregivers being exclusive care providers or partial task sharers in total task activities, the housework domain, and, particularly, the personal care domain (two sets of odds ratios=0.26 and 0.57, the lowest among all task domains). Caregivers’ adjusted household income affects the odds of task sharing in the housework and personal care domains. The greater the household income of primary caregivers, the lower the odds that they take on a caregiving task without the assistance of other caregivers. Nevertheless, another measure of power/resources, occupation, receives no support in any of the task domains or in total task activities. With regard to gender role ideology, consistent with findings in caregivers’ range of task participation, the effect of educational level is found only in the housework domain that is an extension of routine household maintenance in a caregiving setting. Each additional year of education is estimated to decrease the odds of being an exclusive care provider or partial task sharer by a factor of 0.09 (=1.0-O.9l). An additional testing of the interaction effect of education by gender serves to redress the deficiency of the model 170 that considers the effect of gender ideology to be the same for men and women. As in the analysis for task participation, all tests, including within total task activities and all task domains, indicate (data not shown) that the interaction effect is not significant for either the total sample or samples of spouses and adult children. As speculated in Chapter 4, the lack of support for the interaction effect, as seen in domestic labor research (Baxter 1992; Ross 1987; Seccome 1986; Huber and Spitze 1983), may be due to, first, the problematic measure of education as a proxy for gender role ideology, especially for elderly spouse caregivers, who are likely to adhere to traditional gender-role beliefs regardless of their educational level, and, second, the small number of sons in this study. Other Predictors Factors associated with the health conditions of caregivers and the caregiving situation have been suggested in the caregiving literature (Enright 1991; Noelker and Wallace 1985; Stommel and Kingry 1991; Dwyer and Seccombe 1991) as important factors that influence the kind of care caregivers provide. The data in table 5.1 also confirm their significance in how tasks are allocated between primary and secondary caregivers. The increase in the number of additional adults residing in the care recipients’ households is likely to decrease the odds that caregivers perform tasks alone (odds ratios less than 1, p< .05, for most logits). Because caring for Alzheimer’ patients demands round-the—clock care, their caregivers are more likely than those of other care recipients to call for help and to share tasks with others, as evidenced by the decreased odds of tasks being performed exclusively by primary caregivers in all task 171 domains except the most labor-intensive personal care domain. Demands of care stand out as a primary determinant of primary caregivers’ use of help for all task domains, even more than it does in the case of task participation. But the direction of the effect of demand of care should be noted. In all task domains except administration, the odds ratios are less than 1.00 for the logits of exclusive care versus total dependence and exclusive care versus partial task sharing, but greater than 1.00 for the logit of partial task sharing versus total dependence.32 The findings suggest that primary caregivers are unlikely to be sole care providers as the need level of the elderly increases. The increase in the demands of care, however, does not necessarily compel primary caregivers to abandon tasks completely; they still manage to assist with one or more tasks, ranging from hands-on tasks to intermittent tasks, without outside help, reflecting their role as primary caregivers in the family. Finally, in contrast to the findings in the analysis of task performance, older caregivers have higher odds of performing housework or administrative tasks alone without help (odds ratios=1.04 and 1.03, respectively) than do younger caregivers. The positive effect of age in the task exclusivity analysis does not support the notion that age is a proxy for health. Two speculations can be made about the deviation of age effect. First, the older the caregivers, the more likely they are to adhere to traditional gender role prescription and thus assist with these tasks without help. This speculation, reflecting the assertion of Albrecht, Bahr, and Chadwick (1979) about the 32The odds ratio of a primary caregivers being an exclusive care provider versus a partial task sharer is 0.83 (0.89/ 1.07)) for total task activities, 0.77 for the personal care domain, 0.91 for the housework domain, and 0.95 for the administration domain, all statistically significant at p< .05. 172 "subculture of age cohorts," needs more data to confirm it because, if they are correct, the effect, like the education effect, should be gendered as well. Second, since the support networks older caregivers derive from kinship or work activities tend to be small, they have more difficulty in finding informal help than do younger caregivers. Adult-Children-Only Analysis of Task Exclusivity In Chapter 4, several factors associated with the context of adult children, such as competing family responsibilities, propinquity, and, to a lesser extent, the gender of care recipients, were seen to have a significant impact on task performance. In this section, the larger set of variables common to adult children’s caregiving context is again employed to see if results based on the full-sample model change when additional variables are introduced. The data in Table 5.2 show that, overall, variables that are statistically significant in the full-sample model remain significant predictors of task exclusivity in the adult-children-only model. The only exceptions are diagnosis and caregivers’ self- rated health; they have no effect whatsoever on the degree of tasks exclusively performed by adult children. Dependent children and propinquity, two additional variables that measure adult children’ time constraints and capacity to respond to caregiving demands, appear to be more relevant to their use of helpers than to their task participation. Having dependent children decreases the odds of adult children providing care by themselves in almost all task domains except the time- or physically-demanding domains of personal care and immobility. Living with the impaired parent, i.e., propinquity, allows adult 173 02.60.33 50.508“. ”00.0.0 3.00.2 0.005.008. 03.00.... .0 .803uo 003.0003". ”0.000005 .2383: 0. ”200.00 30200.0( ”0300000002. 300.80... .0 a .28 .0 BEES 6.00.. .0. "V o .0009. N A . .23 .0 02.050 ”0.09. 0. 69‘ 50285.1.— .._ooau......_0oz .0000» 0. .96. $08803. ”m. .000: 022.00 0=no .m. .30: 023.00". ”023.5 93> .0000 02...... 000.30. 02.". “£000.00... 69.20.00 .00 ..o 0.0.. tunno 6...... =2". ..0!0>o_nEw ”cacao 5000.30.20 u. 02.3300 ”Bu—.00 0. “2:80. 0.9.3.6.. 00.3.90 .ofiEuo 6.080.". 000000 .0223. 00 0.02.00 00.000 .. ”0.02 00,v0..0.v0:_.00.v0:. $6 8.0 mm... mod 8.. 3:339... IN on» 35 man 93 E089... .0 02900 .anN 3.3 QVKQQ 8.23 5me can! voozzoxfi 00.. 00.0 0.0 . 00.0 0.0 0.0 00.0 00.0 00. 00.0.0 0.00.... 00.. .00 0.. 0.. .... 00. 0.0 00. 00.. .0003 80008000000 9.00.0.8... 28 no. 0.... who and 3.0 n... 3.0 .... .6... 000E505... H0.000005 mm... 0.6 .. .md 00... a on... 3.0 : and .: n90 :. 0nd 28.0.. 302.60 :. 00.0 .: a... : 8.. mad . 8.0 No. .2 and no. .. .od 23.0 000.000 . 3.0 3. Rd 8.0 : .0... t. on... 00.. who 2.. 00.0 : 006 0.8.0 02.050 02.0.3.6 33320.0 00.. 00.. .0 no. No. .0 3.. 5.. on... No. 0 3.. on mud n... 00.. no.0 cod 0... sad no. um. 5.52... 3.09.00 .0 3.3.0.2..ch .050 .00 .0. 00.0 a 00.0 a. 00.0 00.0 00.. 00.0 00.0 .20.. 000000000 000 .. 000 .. 00.0 00.0 . 00.0 0... 00.0 000 .. 00.0 00.0.0.0 .00000000 . 00.0 00.0 00.. E 00.0 00.... .. 00.0 .. 00.0 .. 00.0 .. 00.0 0.000.020 . .00 . ...0 00.. 00.. .. 00.0 00.0 t. 000 00.0 . 000 005.2050 .00 0... 00.. .0. . 00.0 0.. . 0.0 00. 0.0 02.000000 00008.90 00.. .0. 00.0 00.0 00.0 00.0 00.. 8.. 00.. 00.80. 0.00030: ...0 030.003: 20...? 82:03.. 00.0 0.. 00.0 . .00 .. .00 0... .. 00.0 0... 00.0 0.0.2 .9, 0.050....000000 000200.00 2050 :0 .moe .mmo .50.. mule .80. moo .80.. ion .03. ..20.2020 “\— .W> .fix— 6) .M> 000... 0.0m 00000 .0000“. 20o .3. 000.6 .2200. 20o .06 2000 .0000“. 20o 0.5 00000 .005. 000.0 .3. 5:30.55 :o...m..£5=i< 30:00:01 foiomaot EMU .2320... mama... .80... 0.0.000 .00. .5 .530... 0.25.90. .00. .2020200 0.20 0. 00.0.00 2.00 0000 .0 0.00 0;. 0.0 0.00. 174 children to respond to caregiving demands and thus increases the odds that they will provide assistance without help in all but the least demanding domain of administrative tasks. Finally, two characteristics of care recipients, gender and marital status, do not seem to be associated with either adult children’s task participation or their use of help as has been suggested in the caregiving literature (Dwyer and Coward 1991). The only exceptions are the significant effect of parents’ gender on task performance in the personal care domain, as discussed in Chapter 4, and the significant effect of parents’ marital status (married vs. not married) on task exclusivity in the housework domain (odds ratio=0.49 for the logit of partial task sharing versus total dependence). All in all, the data regarding task exclusivity show that when caregivers’ task performance is placed in the context of helping networks, spouses are as likely as adult children to share tasks with others. This finding holds for tasks that involve intimate contacts among spouses who have performed the caregiving role for a long period of time. Most importantly, husbands are more likely than wives, just as sons are more likely than daughters, to receive assistance from, or share tasks with, either paid or unpaid helpers in domains that are labor-intensive and congruent with women’s traditional gender roles. Sons’ task sharing with secondary caregivers in the male-oriented domains, however, occurs no less often than it does for of daughters, and husbands, in fact, rely on outside help with administrative tasks more often than do wives. With regard to the relative importance of these variables, the Wald statistics suggest that gender and family relation, though remaining significant predictors of 175 patterns of task exclusivity, are no longer the most effective predictors of the task exclusivity index as they are of the task participation index. Rather, factors associated with caregiving situations, such as the demands of care and help from people who live with care recipients become the foremost variables in the logit models of task exclusivity for all task domains and for total tasks combined. In addition, time constraints imposed by paid employment become pivotal, although they are not as important as factors such as demands of care or available helpers in caregivers’ consideration of joint assistance in time-consuming housework and personal care tasks. Adult children’s geographic proximity, not employment, is the most important time factor affecting their degree of task sharing. Finally, education, the proxy measure of gender beliefs, is, as expected, relatively important for adult children in the housework domain that is an extension of routine housework.33 Relative Contribution in the Network of Support Although the task exclusivity index takes into account the support of secondary caregivers, it only measures the presence or absence of any assistance, rather than how much other helpers provide. The index of relative contribution constructed for this study provides information about the proportion of care primary caregivers provide in relation to other caregivers. Three polytomous categories of the relative contribution index are as follows: (1) primary caregivers perform all care activities, or exclusive care; (2) primary caregivers perform more than half of the caregiving activities, or a complementary pattern of care in which primary caregivers are more 33Its relative importance is only second to the effect of propinquity. 176 involved than secondary caregivers; and (3) primary caregivers perform less than half of caregiving activities, or a supplementary pattern of care in which the total amount of help from secondary caregivers exceeds that from primary caregivers. As in previous analyses, the presentation focuses on two odds ratios: exclusive care versus supplementary care, and complementary care versus supplementary care. The third odds ratio, complementary care versus supplementary care, can be recovered by taking the ratios of the two previous odds ratios. The full-sample and adult-children- only analyses are presented in Table 5.3 and Table 5.4, respectively. Full-Sample Analysis of Relative Contribution In the few studies with data on the amount of help received by primary caregivers, the importance of gender, family relationship of caregiver to care recipient, and the type of tasks for which assistance was required have been suggested, though not fully examined, as significant predictors of primary caregivers’ relative contribution effort (Tennstedt, McKinlay and Sullivan 1989). The amount of care secondary caregivers provide in labor-intensive tasks such as housework and, particularly, personal care tends to be low, especially when the primary caregiver is a spouse (as compared to a non-spouse) or a wife (as compared to a husband) (Allen 1994; Tennstedt, McKinlay and Sullivan 1989; Horowitz and Dobrof 1982). With few exceptions (Stommel et al. 1995; Tennstedt, McKinlay and Sullivan 1989), no research has systematically studied patterns of relative contribution, or proportionate involvement, of primary caregivers. Little is known about how the relative contribution of primary caregivers varies according to the gender and family relation of primary caregivers and for what kinds of tasks such gender or relational 177 differences occur. The analysis strategy for the relative contribution index is the same as for the task exclusivity index. Family relation is allowed to interact with gender and duration of care in the model. Another possible interaction term, gender and duration of care, had no effect on primary caregivers’ relative contribution in the preliminary data analysis and thus is not considered in the final model specification.34 Odds ratio for all relevant contrasts of differences are presented in the upper portion of Table 5.3. (See Table 4.1 for they way they are derived from the parameters estimated in the model.) WEE Thus far, the data have suggested that gender is an organizing principle for primary caregivers’ use of other helpers, particularly when the tasks involved are labor-intensive and time-demanding. This gender-based division of labor also seems to influence primary caregivers’ proportionate volume of care in task domains that are labor-intensive. The data in Table 5.3 show that daughters have higher odds of providing exclusive care versus supplementary care in personal care and housework domains than do sons (odds ratio=4.85 and 2.83). In male—oriented task domains, daughters’ proportionate volume of care, contrary to the preliminary finding in Chapter 3, is not any less, if not more, than that of sons. The odds of providing complementary care versus supplementary care in the administrative domain are 2.38 times as high for 34That is, the effect of gender on the relative contribution index in terms of direction and incremental change is not likely to be different between long-term and short-term care providers. 178 §%8§.8§. “gagsuggugggtfgfiguésaz u§8xou...ooau!§8:.§si>1§m.§5§8§E.§I¢u.§§m§§o§8§”§§ .35.§§§.§~58.§un . 6.838328 ”Saab-c.8388... .8918. 8.2.3.128... 38.358583 .682 8.v..._.8.va:..8.v..£ .88 88 8.. 8.. 8.. £338.. 8.. 8.8 88 88. 38 688...... .. . 3.8.. .88. 8.88 888. 888. 8..-m 885.8 . 88 88 8... 88 8.. . 88 a... 8.8 a... :8 $82... ._ 88 t. 8.8 . 88 t. 88 8.. .. 88 8... 2.8 88 .. 88 28.9. .. .._... . 88 :.. 38 . 8.. . 3.. 8.. . 88 a. 8... t. 88 .. 8.. ..: 88 28.88... ..._ 8.. 8.. .8. .. 8.. .8. . 8.. 88 . 88 8.. .8. .. 8.. 88 88 88 88 .. .88 88 88 88 88 5... .2828! 833.9820 bio 88 88 . 8.. 8.. . 88 88 . 88 . .88 . .88 .88 .96.. 38888 .. 38 t. 88 88 88 .t 88 t. 88 a. .88 t. 8.8 .: «v8 2.. 88 .8358... . .88 8.. 88 88 8.. 8.... B. 8... 8.. v... 88888 82 88 88 8.. 8... B8 88 :... 8... .8. 88 88888 .25 a... .8. .88 88 2.8 88 u... 88 8... 8.. .83 28 2.. .88 88 8.. 8.. 88 88 8... 8.. .886 :8 . 8 lb 0 . 88 88 8.. 88 88 . 88 88 8.. . 88 8.. 8.8... 388a: . §o§§3§o .. :8 88 8.. 3.. 88 28 t 88 t .88 :8 B8 280 .8 8858 x 88.8. 88 8.. 88 8.. 88 8.. 8.. 88 88 N... 88.3. .28. 8.. _ 88 8... 88 88 :8 88 .. :8 B8 38 888..» 88.8 8.. 88 88 88 88 88 t 88 .8. 88 88 829.5 .58 88... p.828 .9 has... 23 .8 882 .. 88 8.. : 88 . #8 8.. 88 8... t 88 s... .. 8... 83.95. .58. 88. 898 8... . 8.. . 88 .. 88 . 88 .. 88 s. 8.8 t. 8.8 .. 98 t. 88 .88 8.8.8. 52.85 88.8 8.. 88 .. 88 88 8.. 88 8.. 88 8.. 8+ .80 8.2. 5.2.88. 88.5 8.. 88 .. .3 88 . .88 . 8.8 .. 8... z. 8...“. . 88 .. 8.8 .88 8.2. 5.2-89.» 8:95 , .38 .9 88...... 88.3. ...:... . 8.. 88 8.. v... 8.. .. B8 . 88 :. 88 8.. .. 88 883% 88.8 8.. 3.8 . 88 .... B. . 88 8.. .. 8.. 8.. 3.8 8.2.5 .38 88.8 .28... .9 8.2.8... .. . 2.0 diam 200 imam 2.6 imam 2.0 imam Eco .agm $23039. d) g d) .0) 6) 28 1:30 28 .3“ 200 .350 28 .35 200 .550 8.30 ..uxm 2.0 .1230 200 ..uxm 28 1:30 28 ..oxm 88.88.... =8§==8< £2388: E8888: 28 385.. 3.8.. .88.. §§»£.§8§£§.§.§88§¢8¢§3§ofi 88.3.... 179 daughters as for sons. The same applies to the comparison of wives and husbands. The odds of wives being exclusive providers and/or complementary care providers (as opposed to supplementary care providers) are much higher than those of husbands in total tasks combined, the housework domain, and, particularly, personal care domain. Husbands’ relative involvement in administration and immobility tasks, however, are not any higher than that of wives. Gender differences in the relative contribution of spouse caregivers to a great extent cast doubts on the commitment of husbands to their caregiving role. Even in the personal care domain, infused with a great degree of intimacy, not only are wives more likely than husbands to fully participate and to assist without other’s help, but their proportionate volume of care is also likely to be more than that of husbands (odds ratios for the two logits=3.56 and 2.06, respectively). Gender as an organizing principle of the division of labor in hands-on tasks is as evident for child caregivers as it is for spouse caregivers, as confirmed by the insignificant interaction effect of gender by family relation. Famil R 1 i n Commitmen o are ivin Findings from the task exclusivity analysis have suggested that, except when spouses new to the caregiving role assist with intimate personal care tasks, they are as likely as adult children to share tasks with others. When the analysis moves from task exclusivity to relative contribution, strong relational differences in primary caregivers’ proportionate care can be found in all task domains except immobility assistance, which may be too physically demanding for elderly spouses to perform. In all task domains but mobility, the odds of providing exclusive care or complementary care are considerably higher for wives than for daughters and for husbands than for sons. The 180 relational effect, in both direction and incremental change, is not likely to be different between long-term and short-term caregivers, as confirmed by the insignificant effect of the interaction term, family relation by duration of care, in most of the task domains. Whereas spouses have shown substantial commitment to their caregiving role in terms of proportionate volume of care, the differential effect of family relation by duration of care in the personal care domain suggests that as long-term child caregivers increase their assistance in the personal care domain, the gap between spouses and adult children in the odds of providing exclusive or complementary care becomes less pronounced (or even insignificant)” These findings are consistent with those in the analysis of task participation, which showed that adult children who assume the primary caregiving role for a long period of time became increasingly involved in caregiving tasks, thus narrowing the gap between them and long-term spouse caregivers’ task involvement. DLnation of Care and Support Received Duration of care, as discussed in Chapter 4 and above in the section of task exclusivity, emerged as an important contingency and predictor of the task 35Take female caregivers’ experience in the personal care domain as an example. Whereas long-term caregiving daughters are less likely than their short-term counterparts to be exclusive care providers (23% vs. 31%, respectively, as exclusive care providers), the chance of their providing supplementary care also decreases (24% vs. 33% for their short-term counterparts). The majority of long-term daughter caregivers become complementary care providers (53% vs. 36% for their short-term counterparts). In contrast, wives who have performed the caregiving role for a long period of time not only have a small chance of being exclusive care providers (41% vs. 66 % for their short-term counterparts) but also an increasing chance of accepting a secondary caregiving role (13 % as supplementary care providers vs. 5 % for their short-term counterparts). ./ . “__— -I-L /— a. 181 participation and task sharing of primary caregivers, including both spouses and adult children, when tasks are labor-intensive and time-demanding. Long-term caregivers are less likely than short-term caregivers to be exclusive care providers, or, to put it another way, more likely to share tasks with other helpers. In the analysis of relative contribution of both spouses and adult children, the effect of duration of care on relative contribution, however, is no longer statistically significant in most of task domains. The burden of long-term caregivers, including both spouses and adult children, is not likely to be relieved by the extra help they receive from secondary caregivers. The only exception to the insignificant effect of duration of care occur in the personal care domain. Here, the effect of duration of care is in the opposite direction for spouses and adult children, as noted by the statistical significance of the interaction term (odds ratios=0.31 and 0.28). Not only are long-term child caregivers more likely than short-term child caregivers to become exclusive providers of personal care (as seen in Table 5.1), but they are also more likely to have a greater proportionate volume of care when compared to their respective short-term child caregivers (odds ratio=2.43). In contrast, spouses who provide long-term care are less likely to remain exclusive providers of personal care tasks (seen in Table 5.1) and to provide a lesser proportionate volume of care than are short-term spouse caregivers.36 All in all, the data point to: first, the unusual commitment of adult children who provide long-term care (compared to short-term child caregivers) to help 36The odds ratios of the logits of exclusive care versus supplementary care and exclusive care versus complementary care within the personal care domain are 0.47 and 0.57, respectively, both significant at alpha < .05. . /_ -~. .- 182 with the most labor intensive, time-demanding personal care tasks; and, second, the willingness of spouses to relinquish their caregiving responsibility in personal care tasks characterized by both intimacy and labor intensity when their caregiving career is extended. Power/resources. Time. (Lender Ideology am 0mg Predicmrs In addition to gender, family relation, and duration of care, many characteristics of caregivers and the caregiving situation have substantial influence on the relative contribution of primary caregivers. The data in Table 5.3 show that, overall, factors associated with caregivers’ characteristics and the caregiving situation exert an independent influence on their relative contribution in the same way as they do on task exclusivity. In all but the intermittent administration domain, time constraints imposed by employment decrease the odds that primary caregivers will be exclusive or complementary care providers. As the household income of caregivers increases, so does their family negotiation and purchasing power, as reflected in the low odds of their being exclusive care and/or complementary care providers in total task activities, housework, and immobility tasks, although not in the most labor-intensive task domain of personal care. Education, as a proxy measure of gender beliefs, appears to receive stronger support in the relative contribution index than in the indices of task exclusivity or task participation. As the educational level of primary caregivers increases, their relative contribution decreases in the housework domain as well as in the personal care domain and in total tasks combined, but increases in the administration domain. Again, the lack of specification of the interaction effect of 183 education by gender is redressed by an additional testing of the model that includes such an interaction term. According to the gender role ideology hypothesis, the effect of education, if used as a proxy for gender beliefs, is likely to be gendered. The less educated the female caregivers, the greater their proportionate volume of care in housework tasks but the greater their relative contribution in the administrative tasks. The opposite is true for male caregivers. All the tests, including within total task activities and all task domains, however, indicate (data not shown) that the interaction effect is not significant for either the total sample or samples of spouse caregivers and child caregivers. As speculated above and in Chapter 4, the lack of support for the interaction effect, as seen in domestic labor research, may be due to the limitations of the data in terms of the problematic measure of education as a proxy for gender role ideology and the small number of sons in the sample. The presence of other adults in care recipients’ households, again, offers relief for primary caregivers in all task domains except personal care. With regard to the demands of care, the above discussion on task exclusivity showed that as the number of care recipients’ functional dependencies increases, primary caregivers are less likely to be sole care providers. Nevertheless, the decline in the care recipients’ functional level may require primary caregivers to call in help, but not to withdraw from their role as primary caregivers; the data in Table 5.3 show that an increase in the demands of care also increases the odds that primary caregivers are complementary care providers as opposed to supplementary care providers, an effect that is observable in all task domains and that, again, confirms the commitment of primary caregivers to their caregiving role not only in total task activities but also for 184 all task domains. In addition, a diagnosis of Alzheimer’s disease seems to affect the odds of performing housework alone, but shows no far-reaching effects in the other task domains. Finally, as in the analysis of task exclusivity, the notion of age as a proxy for health is not supported; the positive effect of age is found in the housework and administration domains. The notion of a subculture of age cohort (Albrecht, Bahr, and Chadwick 1979) is speculated to explain the deviation effect of age, but is put into question by the negative effect of age in the personal care domain (odds ratio=.98). The shrinkage of support networks as individuals age may be one explanation as to why older caregivers are more likely than younger caregivers to provide a greater proportionate volume of care in the personal domain. Adult-Children-Only Analysis of Relative Contribution The discussion above on adult children’s task exclusivity pattern showed that variables that are significant in the full-sample model tend to be significant in the adult-children-only model, although their relative importance may be reduced.‘ The same can be observed in the analysis of relative contribution. The data in Table 5.4 show that variables found significant in the full-sample model are likely to maintain their significance in the adult-children model. The only exceptions are caregivers’ household income and, to some extent, education and additional helpers; their effects are no longer significant in some task domains and their relative importance is greatly reduced after additional contextual variables are introduced. The presence of dependent children and, particularly, a caregiver’s propinquity to the care recipient, which measure the time constraints and flexibility of adult children, are statistically significant in almost every task domain. Adult children’s 185 888.38 ”8.52.. ”3...... .28... 8858.... .81.... .o .8888 8.2525... 8.888... .2828 c. ”28.2. 388.83. ”8.88888 .8882... 8 .28... 88.8 .23. 8 av .8 .28.. a A . .28... 882.5 .28.. 8. 8.... 838...... .88... 8.8.. .28.. s .26. .8883 8. .85 828...... 8a.. 8. .85 8235.. 8236 88> ...:... 2.1. 88.88. can. 838.8... .2288 .8 .8 2.... .8... .2... 3.... 8.2.2.5. .38... £93329... .. 82.8.80 28...... 5 i=8... 8.8832 .2828 8.3.3. 58.8 8.822». .888 .86.... 8 822...... 8.80 .. 8.82 8.v.....8.v..:..8.v..:. 88 .u... 88 .88 8.. 8.882 8.. 8.. .8 8.. 8.. 2882.. .8 8288 8...: 88.8 8.8.. 8.8.. 883 8...... 88...... 2. .8 88 .88 .8... 88 88 88 88 :.. 3...... .28... 8.. 88 8.. .8. ...8 8.8 .88 8.. 88 .8... . 8.5.2888 25.38.. 28 .. 88 .8... 88 88 .8... . 88 88 B8 .8... . 88 ”2.88.. 8.8.8.8 8..... . 3.8 88 . 8.8 v... 8..... 8.. 8..... 88 . 88 28.2. .88... 8.. 2 36 8.. o 8.. had . mad 8. :.w .8 had 86 .2 36 0.8 .0 0:558 . 3d and 8.0 36 no... Ed 2. mVN n... 8.. Ned 2N0 .0 =2.-r50 gm m$802.0 5.. 5.. 8.. . 8.. mad «8.. 8... 8...... mad 5.. m... and 5.. x... 8... 3... 5.. No... .. ... o...— 5.3... 8:828 .o 8832-... .056 5.8 cm... 8.. 8... 1.... . 3... an... 8... S... 8... .26.. 382.825 8.. 2.... . an... . av... . av... : no... . 38 on... 2. 8..... : 8.... 880...... 8.9.8.00 :. 8.8 E 8.. . :8 8.. t. 3.. 88. z 88 z 88 a. 88 i 88 838.8... .m... : aw... 2.... a... z. 3.8 ..: a... 5... t. and .3 «n... ..: «a... 22.529... 5.. no... 8 ... mm. on. av... m ... Ed and 3... gauge"... .3882... 3.8 8... 8.. . 8... 8... 8... no... 8.. am... 3.. 2:8... 223...... ._ 320.0% 25.? .02....8. 8.. .88 . 9.8 8.. . 8.8 8+ 8.. .. 8.... . $8 . 28 .922 .8 28.8... .8880 88.823 2.0 imam 2.0 imam 2.0 imam 2.0 fl§m 2.0 immm 7280......— ..> .9. .2. .9. .2. 2.0 .330 2.0 .20 2.0 .5200 2.0 ..uxm 2.0 .530 2.0 .32 2.0 .550 2.0 ..uxm 2.0 i200 2.0 .05 23.3.5... §3£§< 223...... 22:02.... 2.0 .252...H 2...... .86... 5.28 38h an .208... sou-55:00 2520¢ .2228 3.50 3 3...! 03.". .000 3 80m 03... v6 .3... 186 proportion of care activities is likely to increase if they live with parents and decrease if they have their family responsibility for child rearing. The lack of support for the effects of the gender and marital status of impaired parents in the relative contribution index is consistent with the results found in the analyses of task exclusivity and, to a lesser extent, task performance. Contrary to the findings of other studies (Dwyer and Coward 1991; Horowitz 1985b), the spouse of the impaired parent for whom adult children care does not necessarily relieve their burden put on primary child caregivers. In addition, the gender of parents may affect the task performance of adult children when tasks require bodily contacts with their parents, but it becomes irrelevant to their incorporation of help or their proportionate volume of care. The insignificant findings, contrary to the pattern found in the preliminary data analysis in Chapter 3 (see Table 3.10 and Table 3.11), however, should be read with caution. This may be a case of insufficient statistical power due to the small number of sons. All in all, the data on primary caregivers’ relative contribution, or the amount of care they receive from secondary caregivers, have suggested a gender- and relationship-based division of labor. Nevertheless, gender and relational effects in the indices of division of labor, though prevalent, no longer seem to have the greatest impact among variables as they do on the index of task participation. Using the Wald statistic as an indicator of relative importance, this study shows that, for all groups of caregivers, demands of care, employment conflicts, and, to a lesser extent, help from people living with the care recipients become the most important predictors of the division of labor patterns in domains that demand time, physical presence, and intense 187 labor. In the administration domain, the primary determinant of caregivers’ relative contribution is the availability of helpers; and, as expected, in the immobility domain, the demands of care are the most important predictor. For adult children, the relative importance of demands of care and full-time employment is largely replaced by propinquity. The relative contribution of adult children is mainly determined by where they live, though their gender, length of time performing the caregiving role, work schedule, and the functional dependency level of their parents remain important determinants. Concluding Remarks and Discussion Burden of Spouse Caregivers: Marital Norm or Lack of Resources This study has shown the wide range of tasks in which spouses participate and the intensity of their involvement in care relative to other caregivers’ efforts. Resistance to outside help and the norms of the marital bond have been proposed as explanations for spouses’ extensive involvement in caregiving (Tennstedt, McKinlay, and Sullivan 1989). The assertion that spouses act as gatekeepers, resisting institutional and informal assistance of others, however, is questionable, given the overall lack of relational differences in the task exclusivity index. Spouses, at least long-term spouse caregivers, are just as willing as children to share intimate personal tasks or routine housework with other caregivers. Dwyer and Seccombe (1991), as well as others (Stoller 1992), have suggested that, given the differences in the nature of caregiving relationship between marital dyads and child-parent dyads, caregiving for family members must be understood in light of marital norms and expectations. Indeed, this study confirms that spouses’ task 188 participation is likely to be influenced by the normative expectations of the marriage relationship. The disproportionate volume of care of spouses, however, should not be read as mere evidence of the effects of the marital bond that discourage spouses from seeking outside help. The availability of support networks, secondary caregivers’ perception of spousal care, and the resources with which the elderly can negotiate help may all help to explain why spouses, who are not necessarily different from adult children in their use of outside help for task assistance, simply receive less help than adult children from other caregivers. Secondary caregivers’ perception that spousal care is guaranteed care because of the marital obligation may also be related to the sporadic care secondary caregivers provide. When asked whether they felt abandoned by family, spouses in this study felt more abandoned than did adult children (mean on a 5-point Likert scale=2.68 for spouses and 2.16 for adult children, p < .05). Although their responses may be subjective rather than objective, the findings do highlight spouses’ need for outside help to relieve their caregiving burdens, rather than their resistance to outside help based on notions of obligation derived from the marital bond. According to a national survey (Stone, Cafferata, and Sangl 1987), the majority of spouses (60%) who assume the primary caregiving role tend to be sole providers whereas most adult children (53%) rely on unpaid help from family or friends.37 Noelker and Wallace (1985) 37According to the reconstructed figures obtained from the national survey (Stone, Cafferata, and Sangl 1987, Table 2), among those who assume primary caregiving responsibilities, 60% of spouses are sole providers, 28% have one or more unpaid helpers, and 12% use a combination of paid and unpaid assistance. In contrast, 31% of adult children are sole providers, 53% receive informal help, and 16% receive a combination of informal and formal help. There are slight gender differences in the use of help by primary caregivers. Husbands are somewhat more likely than wives to 189 also found that caregivers, including spouse caregivers, often state the need for additional help, particularly with household tasks and personal care. Nevertheless, even if paid or unpaid helpers are available, the support networks for spouse caregivers can be fragile (Wilcox and Taber 1991), as this study has also revealed. As they negotiate for help, elderly spouses’ use of formal services may be restricted by their limited resources, including financial resources, inadequate insurance coverage and/or reimbursement levels for formal assistance (Miller 1991).38 The informal networks of elderly spouses may also not be as resourceful as those of adult children, who are middle aged and have a wide range of social networks derived from kinship and work activities. Male Caregivers as Primary Caregivers Although sons have been underrepresented among caregivers, Kaye and Applegate (1990) suggest that with the shrinkage of the American family size and women’s entry into the labor market, there is a possible shift toward sons as primary caregivers in home care. Some researchers consider this new trend a sign of an emerging valuable resource of care for aging relatives, one that might reverse women’s disadvantageous status in caregiving (Kaye and Applegate 1990; Horowitz 1992; Montgomery and Kamo 1989). In their view, caring will no longer be devalued use a combination of informal and formal help (16.2% and 9.3% for husbands and wives, respectively) whereas sons are more likely than daughters to have family or friends share caregiving responsibilities (60.5% for sons versus 51.6% for daughters). 38As expected, elderly spouse caregivers are usually worse off than child caregivers in terms of financial resources. From the perspective of care recipients, however, surviving widows (who make up the majority of parent care recipients) are worse off than the elderly with spouses present, when it comes to finance and insurance. 190 when men are integrated into women’s caring work (see Fisher and Tronto 1990), or when women are released from the caregiving role and, therefore, able to remain in the labor force and enhance their status in the public sector and, in turn, in the household. Sons do indeed care for their parents. Nevertheless, when sons participate as caregivers, they are less likely than daughters to continue to help throughout the length of their parents’ dependency period (Montgomery and Kamo 1989) and to be primary caregivers providing daily, routine care. Even when sons occupy the primary caregiving role, as this study shows, their experience of caregiving is still very different from that of daughters. The types of tasks primary caregiving sons perform are subject to stereotypic gender roles and they are less likely than daughters to engage in hands-on tasks without the involvement of others. Daughters, however, seem to have less discretion than sons about which tasks to take on and how much care to provide, as is evidenced by the lack of gender differences in the administration and immobility domains. . Even among spouse caregivers, the division of caregiving work is still partially based on gender. Although husbands’ task participation appears to be constrained by the norm of reciprocity in marital relationships, how tasks are allocated and provided remains very much gendered. The confounding effect of gender by family relation in the provision of care suggested by Dwyer and Seccombe (1991) may be overrated. In addition, gender differences between spouse caregivers in the task exclusivity index suggest that husbands’ provision of care is often part of a joint effort. The tendency of husbands to receive help with personal care and housework tasks confirms the notion of male helplessness. As suggested by Hooyman (1989, p.9) and 191 others (Stoller and Cutler 1992; Stoller 1992; Guberman 1988), "since their caregiving tends to be viewed as an expected expression of compassion compared to an expected duty for women," husbands are more likely than wives to gain attention and sympathy for their stressful role as primary caregivers, thereby receiving more support from both formal and informal sources, particularly when tasks considered part of women’s domain are involved. In contrast, as suggested by Stoller and Cutler (1992), women are often considered less needy than men and their efforts often go unnoticed because of their expected nurturing role and extensive experience in managing ”women’s" tasks. Therefore, wives’ low level of task sharing and high level of relative contribution should not be read simply as resistance to seeking outside help. Rather, this finding should be viewed in light of the social construction of these roles as nurturing and their responsibility for domestic tasks. Duration of Care The length of time primary caregivers perform the caregiving role emerged in this study as a significant determinant of how care is provided. But, its effect varies depending on primary caregivers’ gender, family relation, and the type of tasks involved. Compared to short-term child caregivers, adult children who provide long-term care show unusual commitment, especially in personal care tasks that demand time, labor, and physical presence. Although long-term caregiving sons still lag behind daughters who provide long-term care in terms of task participation, task exclusivity, and relative contribution, their active involvement after a long period of caregiving contradicts the general perception about sons who provide care (Stoller 1990; 192 Montgomery and Kamo 1989). This particular finding may be due to the primary caregiving role of sons in this sample and to the relatively low socioeconomic status of sons who provide long-term care, the lowest among all caregiver groups in this sample. It appears that sons who provide care over long periods may have less power and resources relative to their family and, thus, become the designated primary caregivers for the long-term care of the elderly.39 Their socially disadvantaged status may have reduced their ability to resist the responsibilities of taking on intensive, time-demanding tasks, though these sons are still far from meeting the level of care provided by long-term daughter caregivers. Interestingly, and contrary to the general belief about daughter caregivers’ socially disadvantaged status, daughters who provide long-term care in this study are somewhat better off, at least better off than long-term son caregivers. They tend to be professionals, college educated, and somewhat financially better off than long-term son caregivers and short-term daughter caregivers, though also largely retired or employed part-time, which probably is an outcome of their adjustment to the long-term caregiving role.40 In addition, the overall significant effect of duration of care in the task 39Aside from the differences in socioeconomic status, long-term son and daughter caregivers also differ in their experience of competitive family responsibility. For example, the data in Table 4.7 showed that long-term son caregivers are far more likely than their daughter counterparts to have never married (40% vs. 15%). 40It is true that even women who are not caregivers often retire in their middle to late fifties; thus, one would not expect high employment levels among either short— term or long-term female caregivers. However, Franklin, Ames and King (1994) used a subset of the data of this study (i.e., CGZ, the homecare study of new caregivers) in their analysis of female employment adaptation and found that: (1) one-sixth (16.5%) of female caregivers quit work after they assume the caregiving role, and (2) among those who stay employed, nearly one-seventh (13.4%) left the workplace to provide care after three-months of caregiving. 193 exclusivity index for both spouses and adult children implies that, with the extension of primary caregivers’ caregiving careers, the size of primary caregivers’ helping networks is likely to increase, although it is unknown how large the networks become because data about other caregivers involved in task assistance, including both number and sources of help, were not collected in this study. The possible change in the size of the support networks contradicts the findings of other research that suggests the stability of support networks (Stoller and Pugliesi 1991; Miller and McFall 1991). The lack of support for the duration of care effect in the relative contribution index, however, also suggests that expansion of helping networks does not necessarily relieve the burden of primary caregivers or shift responsibility for care to secondary caregivers in task domains ranging from hands-on housework tasks to tasks related to male gender roles.41 This research, by focusing on duration of care, complements a handful of caregiving studies conducted from a time perspective (Montgomery and Kamo 1989; Stoller and Pugliesi 1991; Miller and McFall 1991; Stoller 1990; Litwak 1985). The examination of time in this study, however, has drawbacks. First, the cross-sectional analysis of the time factor does not accurately describe how task participation or the division of labor vary over the course of a caregiver’s career. Second, the variability of responses in the time factor is lost as the duration of care is recoded into a dichotomy, with the cut-off point at two years of performing the caregiving role. For “The ”duration of care" effect, based on cross-sectional data, may simply be a differential survival effect. That is, caregivers who rely more on others may be able to hold on to the caregiving role longer; thus no "expansion” of networks may actually occur. 194 example, the effect of duration of care, if any, is likely to level off after a relatively long period of providing care. More longitudinal research or cross-sectional research with a better measurement of time is needed to confirm the findings of this study. Task Domains in Networks of Social Support This study uses a task-centered definition of caregiving and focuses only on the physical, rather than emotional, aspect of family caregiving provided in up to twenty caregiving tasks, fifteen of which are grouped into four task domains. The topologies are constructed on the basis of the gendered task demands and the level of emotional, intimate, physical, and time commitments required for task accomplishments. The data show that the relationship differences in caregiving labor often vary with the levels of intimacy and physical and time commitment required by tasks. For example, elderly spouses’ task participation and proportional contribution is not as noticeable in the immobility domain as in other task domains because of the physical level involved in such tasks as lifting or moving care recipients in and out of the bed or house. In addition, the commitment of spouses, particularly those new to the caregiving role, to caregiving in personal care tasks is more pronounced than their commitment to other task domains. As expected, tasks such as bathing, dressing, and toileting require high levels of physical, time, emotional, and intimate commitments that individuals in the marital dyad are better equipped to meet than are those parent- child dyad. In comparison to children, spouses’ greater task participation and relative contribution in administrative tasks also illustrates the characteristics of administrative domain -— commitment and trust, aside from control—over-- when it comes to financial matters . 195 From the point of view of substitution, one might ask whether tasks are distinctive or whether the source of help is interchangeable? Do tasks need to be performed by a certain group of people? In other words, is there a preferred source of help for certain tasks? The overall lack of relational effects in the task exclusivity index suggests that when spouses, who usually are preferred by care recipients as a source of help, are the primary caregivers, their assistance can largely be substituted by other helpers, including their assistance with intimate personal care tasks. In addition, the spouse of an impaired elderly is not necessarily a source of support when adult children assume the primary caregiving role. Many tasks, including intimate tasks, need not to be performed by the spouse of care recipients. Spouses’ vulnerability as primary caregivers in terms of the disproportionate amount of care they provide may not be a result of the fact that they are the preferred source of care. Rather, suggested in caregiving literature, this fact may reflect the non-availability of outside help and their lack of resources (Wilcox and Taber 1991; Miller 1991; Hooyman and Lustbader 1986). The same can be said about gender-based task participation or the division of labor. The fact that women are unlikely to share housework and, particularly, personal care tasks with others, is not necessarily because they are the preferred source of help; as many as two-thirds of parents cared for by a daughter and nearly four-tenths of husbands cared for by a wife receive some type of assistance in personal care and housework from a source other than their female primary caregivers 196 (see Table 3.11)."2 Women are unlikely to share tasks because their socially constructed nurturing roles have placed them in a structurally vulnerable position, thereby requiring them to provide care in areas in which compassion, personal attending yet monotonous routines are emphasized. Finally, this study has highlighted unique patterns of assistance in personal care tasks that not only are different from those in male-oriented task domains but also are often different, in direction, significance, and magnitude, from those in the housework domain that call for hands-on assistance and female gender roles.“3 Several explanations have been proposed for the uniqueness of the personal care domain. Unlike housework or administrative tasks that are an extension of routine household maintenance in a caregiving setting, caring for a relative in need of help with personal care takes place when the severity of the health condition of the relative requires it. Such tasks are, therefore, likely to be performed out of sense of obligation. Second, unlike hands-on housework tasks, personal care tasks demand more labor-intensity, time commitment, and, most of all, intimate contacts between caregivers and care recipients. Thus, personal care is likely to be influenced by the “Since the gender of other helpers is unknown, it is possible that the secondary caregivers who functionally substitute or supplement the assistance of female primary caregivers are also women. “For example, in task exclusivity, adult children’s duration of care effect is found in all task domains except personal care. The effect of power/resources, time availability, or gender ideology in the analysis of task participation is found in the housework domain but not in the personal care domain that is more congruent with female gender roles. Spouses and adult children differ in the degree of task exclusivity in the personal care domain but not in any other domains. In addition, gender and relational effects, if significant, appear to be stronger in the personal care domain than in the housework or male-oriented task domains. 197 normative expectations of marriage. Third, since the nurturing aspect of personal care tasks are congruent with women’s socially constructed gender roles, the prescription of gender roles are more likely to operate in the personal care domain than in the housework domain. Cultural taboo This study has suggested that the cultural taboo hypothesis proposed by Montgomery and Kamo (1989) and Montgomery (1992) to explain sons’ minimal involvement in personal care should be examined, first, in the context of the caregiving experience of spouses versus that of adult children and, second, with the within-group and between-group comparison of children caring for parents of the opposite gender. The taboo hypothesis is confirmed, first, by spouses’ greater involvement in intimate personal care tasks than in any other type of task and by the reluctance of spouses, at least among those new to the caregiving role, to seek outside help with personal care tasks.“4 Second, the taboo appears to be a concern for children caring for parents of the opposite gender, at least for daughters where sufficient number of cases allows the comparison. Daughters caring for a father are less likely than those caring for a mother to be involved in personal care tasks, although the gender of parents is by no means associated with adult children’s use of help or their volume of “According to the taboo hypothesis, since personal care tasks involve bodily contact between caregivers and care recipients, husbands and wives are likely to feel uneasy about relegating tasks involving intimate contact with their spouses to other helpers unless there is a great need to call in help. Thus, spouse caregivers are more likely to be sole care providers in the personal care domain than in any other task domains. 198 care."5 Personal care tasks, however, are not just tasks involving bodily contact, thereby possibly raising taboo thresholds for adult children who provide care. Caregivers often see personal care tasks requiring round-the-clock attention as tedious, repetitive, and time consuming and tasks that involve a loss of control over time and place (Montgomery, Gonyea, and Hooyman 1985). Therefore, it is not surprising that spouses are likely to seek outside help with personal care tasks as their caregiving career is extended. In addition, the fact that personal care tasks which involve nurturing and attending to others’ needs fit well into women’s socially constructed caretaker role explains why gender remains a major organizing principle of spouses’ and adult children’s care providing behaviors, and why bodily contact between children and parents is considered taboo for sons but less so for daughters. Power/resources, Time, and Gender Ideology Contrary to the findings of Finley (1989), power/resources, time, and gender role attitudes in this study independently exert influence, though in various degree, on task participation and the division of labor. But, their effects by no means explain 45The effect of the interaction term, gender of parents and children, on the division of labor indices was not tested for all child caregivers because the zero frequencies in response cells cause computation problem of parameter estimates (Hosmer and Lemeshow 1984; SAS Manual 6.01). However, this study managed to run an analysis with daughter sample only (N=295 for daughter-mother dyads and N=74 for daughter-father dyads) and found that the effect of parents’ gender on daughters’ use of help and proportionate volume of care in the personal care domain is not significant. The analysis of daughter-father dyads (N=74) and son-mother dyads (N=28) also showed that there is no group difference in the patterns of task exclusivity or relative contribution in the personal care domain. The lack of support for the effect of the gender of parents on adult children’s division of labor, however, should be read with caution. This may be a case of insufficient statistical power due to small number of sons or of daughters caring for fathers. 199 away the effect of gender in caregiving. Although time availability is not a crucial factor to primary caregivers’ decisions about which tasks to assume, it is relatively important in the decision to share tasks and to the volume of care provided."6 Different sources of time constraints, however, carry different weights in adult children’s use of help and proportionate volume of care. Propinquity, or the time flexibility to respond to parents’ functional needs, is the most important predictor of children’s assistance in hands—on housework, but, somewhat surprisingly, not as crucial in their assistance with personal care tasks that demand a high level of time commitment and physical presence."7 Competing family responsibilities impose significant constraints on adult children’s ability to provide care in hands-on tasks, but, overall, it is not as important as the other two time factors, propinquity and work schedule conflicts. In contrast to the strong effects of time constraints on the two division of labor indices, the effect of caregivers’ adjusted household income varies somewhat from domain to domain. The only consistent finding is associated with the housework domain: with an increase in power/resources derived from income, caregivers are less likely to take part in housework tasks and more likely to recruit helpers, and thus to provide a less proportionate volume of care. The limited support for income and “As mentioned in Chapter 4, gender and relationship are the foremost predictors of caregivers’ task participation. The time factors are significant predictors of primary caregivers’ participation in total task activities and housework tasks, but only propinquity is relatively important among variables when it comes to child caregivers’ participation in housework tasks. 47It is possible that housework can be skipped, but personal care needs immediate attention and has to be fulfilled. 200 occupation, however, should be read with caution because of the problematic operationalization of the variables resulting from the limitation of data. As mentioned in Chapter 4, the measurement of occupation in this study is subject to possible errors due to a misclassification in the level of authority associated with occupational positions and in the assumption of a carry-over effect from pre-retirement occupations. The assumption that primary caregivers have access to the income of family members with whom they share a household also raises issues of intergenerational financial flows and the gendered access and control of household finance. Most of all, the absolute measure of primary caregivers’ power/resources ignores the contextual meaning of power in a relative sense, that is, the power/resources of a primary caregiver vis-a-vis other family members who could be potential primary caregivers, especially when power is conceptualized as negotiation in a family dynamic as to who does what and how much. Education, as the proxy measure of gender egalitarian ideology, exerts an influence primarily on housework tasks that are an extension of routine domestic responsibilities in a caregiving setting.“8 The overall lack of support in the personal care domain (that is congruent with women’s socially constructed nurturing role), as speculated in Chapter 4, is probably because caring for a relative in need of help with personal care takes place only when the health of the relative is severely deteriorated and thus is unlikely to be subjected to gender role beliefs. The indirect measure of gender-role attitudes, however, also raises doubts about some of the conclusions made “The relative importance of education in the task exclusivity index is only second to the effect of propinquity and duration of care. 201 in this study because it is possible that education carries an effect over and above gender-role beliefs (Ross 1987). The major problem with the testing of these hypotheses, as mentioned in Chapter 4, is the lack of specification of their gendered effects in the model. For example, if the gender ideology hypothesis holds, men who hold egalitarian gender ideology would rely less on other helpers to assist them with personal care or housework tasks while women who voice an egalitarian viewpoint would be likely to relegate tasks traditionally considered part of their domain to other helpers; the converse would hold true when household administration tasks congruent with men’s gender roles are involved. In addition, the conception of time and power/resources, as suggested in the literature, are gendered and, as a result, utilization of time or exercise of power/resources are often qualitatively different for men than they are for women. The assumption that men’s work has primacy while personal/family time and work time are intertwined in women’s conception of time can lead to different patterns of use of outside help and a rigid division of caregiving labor along gender- defined lines. Men deal with their multiple responsibilities by reducing their volume of care whereas women deal with them by maintaining rigid schedules, negotiating care tasks around their work schedule, and giving up their own free time (Cantor 1983; Horowitz 1983; Lang and Brody 1983). It is thus not surprising that, when asked about the caregiving impact on their schedule on a 5-point Likert scale, women in this study report more time impacts than do men (mean on a 5-point Likert scale=3.59 for women and 3.23 for men, p< .05). In addition, cultural rules proscribing women’s exercise of power has inhibited them from using their resources 202 as effectively as men, even if their levels of power/resources are equivalent (Blumstein and Schwartz 1991; Gillespie 1971). Therefore, power/resources, time, and role ideology should not be juxtaposed with gender in a model that specifies no interaction effects with gender. It is through these gendered mechanisms that gender differences in caregiving are produced and reproduced. This study tested the interaction effect of gender by gender-role attitudes, but the insignificant finding should be read with reservation. As speculated above, the limitation of data, in terms of insufficient primary caregiving sons and an indirect measure of gender role ideology may be responsible for the lack of support of the interaction effect. C aregiving Situation and Other Characteristics of Caregivers Other predictors of caregivers’ behaviors primarily revolve around the advanced age of caregivers, the level of functional impairment of care recipients, and available helpers in the care recipients’ households. Caregivers’ health characteristics and the caregiving situation impose constraints on caregivers’ abilities to take on caregiving tasks, and, most of all, require them to call for help, as the magnitude of the Wald statistics in the division of labor indices suggests. Whereas the task performance of primary caregivers is primarily constrained by the norms of reciprocity in marital relationships and/or stereotypic gender roles, the allocation of tasks between primary and secondary caregivers, though still based on the gender or family relationship of primary caregivers, is primarily driven by the characteristics of the caregiving environment. Finally, this study concludes that age is more than an indicator of health condition. As caregivers age, they are less likely to take on all types of caregiving 203 tasks; but the division of labor analyses showed that older caregivers are much more likely than younger caregivers to provide assistance without help in housework or administrative tasks. Two explanations have been proposed for the positive effect of age on the division of labor: age as a subculture of beliefs about gender roles and equality and age as a reflection of social networks. More studies are needed to examine, first, the relationship between age and gender beliefs and if such relationship is gendered, and, second, the relationship between age and size of social networks derived either from kinship or work activities. Summary This chapter has assessed factors associated with the division and allocation of labor between primary and secondary caregivers. Caregivers’ gender and their relationship to the care recipient were shown to have a major effect on this division of labor. The outside help spouse caregivers receive does not relieve spouses’ caregiving workload as they still provide a greater proportionate volume of care than do child caregivers. In addition, when primary caregivers’ involvement is placed in the context of support networks, the division of labor is as gendered for spouse caregivers as it is for adult children. Researchers of caregiving have tentatively proposed several explanations for the relationship-based division of labor. This study has rejected using resistance of outside help and marital bonds as explanations for spouses’ disproportionate volume of care. More research is needed to examine how secondary caregivers’ perception of spousal care and lack of resources are linked with spouses’ fragile support networks and their vulnerability as primary caregivers. 204 This study is primarily concerned with the gendered division of caregiving labor. It has confirmed the notion of male helplessness, the gendered practice of the taboo proscription, and the inequitable gendered division of labor when tasks are labor-intensive and time-demanding. The persistence of the gender effect raises questions about why women, including both wives and daughters, persistently provide more care than do men when tasks are labor intensive, and about the mechanisms that contribute to such gender differences. In Chapter 6, some speculations from a feminist stance will be proposed to answer these questions and group-specific logistic regression analyses are presented to examine what factors affect the way men and women provide care in total task activities and in two hands-on domains, where gender- and relationship-based divisions of labor are largely found. CHAPTER 6 A FEMINIST INTERPRETATION The purpose of this study was to explore how (I) the primary activities of caregiving in terms of task involvement and the division of labor differ on the basis of the gender of primary caregivers; and (2) gender intersects with family relationship to influence caregivers’ behaviors. The data showed that when tasks are labor-intensive, time consuming, and congruent with women’s nurturing role, the division of labor is distinctively gendered regardless of caregivers’ family relationship to care recipients. When tasks are defined as part of men’s gender roles, in contrast, male caregivers’ endeavors are not any more than those of female caregivers. In this chapter, I draw upon feminist theory to explain these persistent gender differences in caregiving. In so doing, I identify problems associated with the model specification of this study, a gender difference approach that is frequently used in other research (Finley 1989). A theoretical framework of gender stratification is then proposed for future research and a preliminary data analysis is presented to outline the mechanisms that affect how men and women provide care. From a Gender Differences Approach This study has demonstrated that men and women provide care in a gender- specific way only when tasks are labor-intensive, time consuming, and undervalued. Men and women may be biologically or psychologically different but these differences cannot explain why gender has become associated with an unequal ranking and 205 206 rewarding of tasks and thus a division of labor disadvantageous to women (Chafetz 1988). Gender differences or divisions of labor do not, conceptually and logically, imply inequality. Empirically, however, divisions of labor have been translated into hierarchies in which female traits are devalued relative to male ones, women’s contributions at work and in the household are considered more valuable than those of men, and tasks that are repetitive and time consuming are usually associated with women.49 As put by Thorne (1990), when the subject is gender, there is no escaping the theme of difference. Comparisons of men and women in a gender difference approach, however, ”deflect focus from gender as a social relation and obscure the processes that amplify or mute its significance" (Rhode 1990, p.6). One of the problems with the gender difference approach, then, is its simplistic dualism that conceptualizes gender as a relatively static and fixed binary variable (Thorne 1990; Rhode 1990; Stacey and Thorne 1985). Another problem with it is that it ignores the social context in which one gender becomes subordinate to the other. Looking through the lens of context allows examination of the social relations and institutions in which multiple differences are constructed and given meaning. By shifting the level of analysis from sex categories to the social relations and social organization of gender, a researcher may learn not only why gender differences persist but also the mechanisms involved in producing gender differences in caregiving. To explain the process by which structural arrangements and normative ”I acknowledge the danger of treating ”women" as a unified category. Not all women are equal. The subordination of women differs by class, race, ethnicity, nationality, and many other hierarchial social markers (Lerner 1990). 207 expectations legitimate the value and social arrangements of caregiving work based on the sex category (Ferree 1990), one must understand the social relations that construct differences -- and diminish or undermine them (Thorne 1990). That is, one must understand how the interplay of the social relations of production and reproduction and the intersection of capitalism and patriarchy in a sex/gender system has led to "structural, relational, and symbolic differentiations between women and men" at both macro and micro levels and thus a division of labor disadvantageous to women (Acker 1989, p.238). A Feminist Theory of Gender Stratification In the pages that follow, I outline a feminist theory of gender stratification to highlight the mechanisms underlying the relationship between women and caregiving. I begin by discussing the concept of the sex/gender system and then consider production and reproduction and the intersection of capitalism and patriarchy. Following this examination, I show how caring is a manifestation of the capitalism/patriarchy nexus in everyday life. Sex/Gender System The "sex/gender system” is a concept introduced by Gayle Rubin (1975) to explain gender inequality and women’s subordination. According to Rubin (1975, p. 165), the sex/gender system is fundamental to the organization of social life and consists of "a set of arrangements by which the biological raw material of human sex and procreation is shaped by human, social intervention and satisfied in a conventional manner. " Like the mode of production, the sex/gender system is socially constructed and changes historically, but it always involves three components: the 208 social creation of two dichotomous genders from biological sex, a particular gender division of labor, and the social regulation of women’s sexuality. Although not static across time, all sex/gender systems thus far have been organized in ways oppressive to women (Thorne 1982; Vance 1980); they have become systems of male-dominance in which men control women’s productive and reproductive labor through a broad array of social institutions, including the family, the market, and the state bureaucracy (Chafetz 1990). In other words, in a sex/gender system, biological sex is given social significance and forms the foundation of a stratification system. One advantage of this approach, as explained by Frank (1985), is that it delineates the systematic links between the creation of two genders, the gender division of labor, and gender relations. Rather than seeing gender as fixed, unchanging positions of single individuals, the sex/gender system approach focuses attention on the power relationship between men and women and how the interrelation of the genders are constructed and changed in a social process (Frank 1985). An analysis of the interplay between production and reproduction and the intersection of capitalism and patriarchy illustrates the primary basis of the power differential between men and women and the power imbalance inherent in gender relations. The Production and Reproduction of Labor The production of the means of subsistence (i.e., food, clothing, and shelter) and the production of human beings themselves (i.e. childbearing and rearing) are two activities essential for the survival of all societies. Because of women’s biological connection with the production of life, a division of labor by gender historically has 209 assigned men responsibility for the production for subsistence and women responsibility for both subsistence production and domestic work.’0 Although a division of labor between men and women is not inherently disadvantageous to women, gender divisions of labor have become a relationship of dominance and exploitation (Mics 1986). Men’s activities are usually rewarded with money, power, and prestige, while those of women are more or less invisible (Osmond and Thorne 1993; Hartmann 1981). Male prerogative in the gender division of labor can be located in the historical transition from subsistence agriculture to an economic system based on private property (Engels 1948). Before the rise of private property, men’s productive work and women’s reproductive work were accorded equal social significance. With the development of private property (i.e., land and animals), however, men became concerned with inheritance. To ensure that their property was passed down only to the children they sired, men began to control women’s sexuality and generative capacities (Engels 1948; O’Brien 1981). The consequence of the emergency of private property and men’s control of women’s sexual activity was the transformation of relations between men and women within the household. As production based on exchange expanded and came to overshadow subsistence production, women’s work came to be seen as private and for family use (Sacks 1975). In other words, women worked within and for a family and whatever production they carried out in the public sphere became invisible. ”Women’s role in subsistence production varies from society to society. 210 The public/private dichotomy that equates men with the public sphere and women with the private family was magnified after the rise of industrial capitalism, a new mode of production that emerged at the end of the 18th century, separated the work place from the household, and changed the social relations of production (Calasanti and Bailey 1991). As economic production shifted from family to factory, men left the household for productive, wage work while women remained bound to their reproductive function, staying at home to rear children and maintain the household.51 Even in contemporary society, where women’s increased participation in the public domain has blurred the division between public and private realms, many normative expectations of women in the family and the work place still start with the notion of private women and public men (Ferree 1990). The consequence of the public/private distinction is the reinforcement of the ideological definition of women’s work as unpaid reproductive work. For feminists, this work involves three activities: biological reproduction, reproduction of the labor force, and reproduction of the social system (Edholm, Harris, and Young 1977; Harris and Young 1981). More specifically, first, women bear and rear children who inherit men’s property and social position. Second, women feed, clothe, nourish, and sustain people on a daily basis, thereby reproducing the labor force. This reproduction of labor force, commonly termed domestic work, is attuned to the human organism, that is, to human’s physiological needs for food, cleanliness and protection, and thus is work that maintains laborers. Third, women teach children societal values and 51Women’s work, however, was shaped by their class position and women of the lower classes almost always worked outside the home. The ideology of the "housewife" -- a bourgeoisie phenomenon» shrouded their work from view, however. 211 norms, consequently creating individuals to fit into the social structure of society and thus ensuring its continuation (Mackintosh 1981). Social reproduction ensures the reproduction of adequately socialized labor and the preservation of the prevailing structures of the social system (Harris and Young 1981). The Intersection of Capitalism and Patriarchy To understand men’s control of the public sphere and women’s reproductive work, feminists argue, one must understand the intersection of capitalism and patriarchy. Patriarchy is not merely an ideological form of women’s oppression, as assumed by Mitchell (1977). Patriarchy must be understood as ”material practices embodied within patterns of behaviour and social institutions” (Harris and Young, 1981, p.125) Patriarchy, as defined by Hartmann (1981, p.14), is ”a set of social relations between men, which have a material base, and which, though hierarchical, establish or create interdependence and solidarity among men that enable them to dominate women".52 Men’s dominance assumes very concrete forms; the material base of patriarchy rests on men’s control over women’s sexuality and labor and the exclusion of women from access to economically productive resources (Hartmann 1981). 52Whereas men of different classes and races have different hierarchical places in the patriarchy, they are dependent on each other to maintain a relationship of dominance over women. In other words, according to Hartmann (1981b), men share solidarity and vested interests in the status quo, regardless of their rank in the patriarchy. For example, evidence suggests that "when patriarchy was first institutionalized in state societies, the ascending rulers literally made men the heads of their families (enforcing their control over their wives and children) in exchange for the men’s ceding some of their tribal resources to the new rulers" (Hartmann 1981 p.14). 212 Women’s subordination in a capitalist society lies in the strong partnership between patriarchy and capital. Capitalism’s imperative for profit requires that the labor force be produced and reproduced at minimal cost. Although capitalism does not need to be inherently male-structured, it benefits from men’s control of women and the ideological definition of women’s work as unpaid reproductive work in many ways. First, women give birth, thereby producing a wage labor force for the capitalist system. Second, by the day-to-day maintenance of the labor force, i.e., caring for adult paid workers and rearing children, women reproduce the working class on a daily basis at low cost. Unpaid housewives attend to the instrumental and psychological needs of men and their families, thereby producing and maintaining healthier workers than do wage-working wives (Hartmann 1981). By employing a man with a home-working wife, the capitalist receives two workers for one wage, thereby lowering their cost of production (Glazer 1984). Third, when women socialize the next generation into appropriate behaviors and beliefs, into readiness for school, and into integral components of capitalist society, they produce subordinate capitalist workers and thus preserve the prevailing structures of class and power (Glazer 1984). Finally, since wages are capitalism’ major cost, the placement of women in the labor force, as both low-paid wage workers and a reserve army of labor, further benefits the accumulation of capital.”3 While capitalism benefits from men’s control of women and the ideological definition of women’s work as reproductive work, men also benefit and secure S'3’Women constitute a reserve army of labor in the sense that capital draws them into or expels them from the labor force according to the exigencies of the market. 213 privilege from this arrangement. First, women’s work as unpaid reproductive labor benefits men directly; it releases men from domestic responsibility, thereby allowing them to invest more time than they might otherwise be able to work-related activities. Second, the gendered location of women in wage relations perpetuates men’s material advantage over them and justifies women’s home responsibilities. Women’s domestic responsibility and double work day, in turn, reinforce their inferior labor market position and assure women’s economic dependence on men in both the labor market and the family, thus legitimizing men’s control in the public sphere. Men’s occupancy of the elite positions in the economy and administrative-bureaucratic political system ensures that the distribution of resources and opportunities gives preference to their "own kind” and thus sustains men’s privileged status and the private nature of women’s reproductive work. Last, but not least, capitalism’s imperative for profit, sustained by patriarchy, has reproduced a social system in which values, attitudes, and norms reinforce the power imbalance in gender relations. Gender ideology systematically distorts women’s reproductive activities by trivializing some aspects of women’s reproductive work, such as housework, and idealizing other reproductive activities, e.g. , mothering and caring. As a result, motherhood and nurturing continue to be central to the definition of female gender roles, family is perceived as women’s territory, and domesticity and care activities are deemed to be women’s "natural" inclination, despite women’s entry into the labor force and into professional jobs (Fisher and Tronto 1992; Holstein-Beck 1995). 214 Caring as a Manifestation of the Capitalism/Patriarchy Nexus in Everyday Life The ideology of caring and the gendered division of caregiving work is a manifestation of the intersection of capitalism and patriarchy. Caring, as defined by Graham (1983, p.13), is a concept ”encompassing a range of human experiences which have to do with feeling concern for, and taking charge of, the well-being of others. " The dual nature of caring, as labor as well as love, however, has been ignored throughout history because of the cultural rhetoric that caring is ”natural" for women (Finch and Groves 1983; Graham 1983). The qualities demanded of caring -- sensitivity, empathy, and compassion-- have been defined as female characteristics while the labor-intensive and highly stressful nature of the caring role has remained invisible. The material aspect of caring, which is repetitive, time consuming, and burdensome, is obscured by the glorification of caring as a labor of love and a bond between the carer and the individual who is cared for. Caregiving, like housework and other unpaid reproductive labor, is accomplished under the guise of family responsibility and the ideology that caring is women’s nature (Humphries and Rubery 1984; Fisher and Tronto 1990). "’Caring’ becomes the category through which one sex is differentiated from the other" (Graham 1983, p.18). Consequently, it is women, not men, who are expected to nurture and attend to others’ instrumental and psychological needs when the need for care arises. And, it is women, not men, who are expected to take charge of nurturing and housework tasks that are intensive and time-consuming. Women’s nurturing inclination, however, is less a natural expression of a woman’s personality (Chodorow 1978) or moral feeling (Gilligan 1982) than a reflection of her place in the broader 215 social system (Graham 1983). Explaining gender differences from the perspective of women’s psychic structure obscures the material aspects of caregiving (Graham 1983) and perpetuates the stereotype of caregiving as women’s nature (Kerber 1986; Greeno and Maccoby 1986; Graham 1983). The idea of caring as women’s nature is a social construct that has to be situated in the intersection of patriarchy and capitalism. Within this nexus, women are confined to the private sphere and caring is defined as women’s work, women’s access to resources and independence are constrained, and they are obliged to assume caregiving responsibility in areas that are labor intensive and time demanding (Fisher and Tronto 1990; Aronson 1992). The disadvantageous position of women in both the household and the market, in turn, perpetuates their role as caregivers and legitimates a division of labor in which work activities of men and women in both the private and public domains are gender segregated and rewarded with different value on the basis of sex categories. A Proposed Model for Future Research To identify differences between men and women, in caregiving research, variables such as power/resources, time, and gender role attitudes are often juxtaposed with gender in the model specification (Finley 1989). In this gender- difference approach, that was also adopted in this study, gender is treated as a static, binary variable that can be separated from the social context that gives rise to the unequal power relations between men and women and the gendered division of caregiving labor. Power/resources, time, and role ideology, however, are gendered, and it is through these mechanisms that differences between men and women in 216 caregiving are produced and reproduced. Individuals’ power/resources, time constraints, and belief systems have to be placed in the sex/gender system so as that we can understand their effects on how men and women provide care. Take time for example. The development of industrial capitalism transformed time patterns into a modern differentiation between work time and non-work time (Thompson 1967 ; Pasero 1994).“ As the public sphere was separated from the private sphere and work time competed with non-work time, new disciplines shaped by the needs of industrial capitalism were implemented. Work time took precedence over domestic and family time, which came to be seen as residual (Straw and Elliott 1986; Seron and Ferris 1995). Time, however, is not a gender-neutral concept. Women’s perception of time takes place within an ideological framework that is subject to the interrelated demands of capitalism and patriarchal relations. Because the demands of the public sphere take precedence over those of the private sphere, and women are assigned to the private sphere, they are subjected much more strongly to the tensions of divergent demands and are more likely to adjust their time around them than are men (Hessing 1994; Cyba 1992). Time for women is not only consumed but is also organized as a resource to cope with the combined demands of paid work and unpaid housework (Hessing 1994; Seymour 1992). Therefore, even when women have little discretionary “Before industrial capitalism, the demarcation between "work” time and "social life" time was less distinct. Social intercourse and labor were intermingled in the tasks people performed. There was no great sense of time conflict between productive labor and domestic chores, nor between productive labor and ”passing the time of day" (Thompson 1967, p.60). 217 time (Pasero 1994), they exercise their time management skills, often at the expense of their leisure and personal time, to meet the temporal demands of the household (Hessing 1994; Seymour 1992; Chambers 1986). Personal/family time and work time are intertwined in women’s conception of time and women’s time is viewed by women themselves and by other family members as a common household resource. Women thus are likely to engage in various time management strategies to meet the combined demands of work and caregiving (Hessing 1994; Franklin, Ames, and King 1994). Employed women’s intentional and creative "time management," such as the routinization of daily tasks and synchronization of household and workplace events, often enables them to accommodate the constraints of a combined workload (Pasero 1994; Chamber 1986), but often at the expense of their subjective well-being (Kramer and Kipnis 1995; Barber 1988; Morris et al. 1991). Men’s conception of time, in contrast, usually starts at work. Their time, especially their work time, is valued more than is women’s time, and thus men’s competing responsibilities for work and family are more likely to be acknowledged. As a result, men often use paid work to legitimize their non-engagement in domestic responsibility and to relegate tasks in the private sphere to another person, either a spouse or a hired helper. Therefore, although employed men, especially those in professional occupations, usually have more control over work time (Seron and Ferris 1995), paid employment releases men from domestic responsibilities more than it does women (Seymoure 1992). Caregivers’ responses to the constraints of the combined demands of caregiving and their own family responsibility (e. g., home maintenance and child rearing), 218 however, may be different from their day-to-day responses to conflicts in their schedules. As noted, women tend to give priority to their domestic commitments rather than to their work and personal time. When the competition is between the domestic commitments of their own family and taking care of an ill parent, however, women’s ability to provide care may be limited. In contrast, men, who are not imposed upon by family responsibility as much as are women, may vary little in their response to the competing demands of caregiving and their own family/domestic responsibility, especially if a spouse is available to relieve their double burden (Horowitz 1985a). The same might be argued about the conceptions and exercise of power/resources. Power is manifest in every interaction. Control of resources provides an individual with the potential to exercise authority, that is, with the ability to bargain, to alter the behavior of another, and to impose his or her definition of what is possible, what is right, and what is real (Fishman 1978). But what people do in specific interactions also reflects historical and social structural forces beyond the boundaries of their encounters (F ishman 1978). Although people with greater resources generally have more power and thus exert more control over situations and other persons’ behaviors or opinions than do those with fewer resources, the ability to exercise power is very gender-specific. Empirical studies show that wives need to earn more income than husbands in order to acquire equivalent amounts of power in the family (Blumstein and Schwartz 1991). Moreover, even when women have resources, they do not use them as effectively as do men to advance their position in the family and society (Blumberg and Coleman 1989). 219 The difference in the way men and women exercise power may reflect gendered conceptions of power and power styles. Power, as pointed out by Johnson (1976) and Miller and Cummins (1992), is a gendered concept. Women tend to define power in terms of empowerment, autonomy, and personal authority whereas men tend to identify it with domination and ”control over” others (Miller and Cummins 1992). That is, men tend to define power in terms of their ability to make decisions about others’ lives or to influence others to see things their way. Women, in contrast, tend to experience power in a personal and indirect fashion -- as a feeling of sharing, personal growth, and a choice of their own. These gendered conceptions of power are manifest in different styles of power, according to Johnson (1976): (1) indirect (e.g., manipulation) vis. direct power, (2) personal (e.g., affection, sexuality) vis. concrete (e. g., money, knowledge) power, and (3) helplessness vis. competence. The style of power women tend to use --indirect, personal and helpless-- usually is less effective than men’s style of power, i.e., direct and concrete. As explained by Miller and Cummins (1992), gender differences in conceptions and styles of power may be due to women’s lack of access to structural power in the economy and polity, a deficit sustained by the sex/gender system. Because women do not historically have access to economic and political power, their principal sources of power tend to lie in indirect, helpless, and personal modes of influence that are oriented toward personal autonomy or empowerment. Moreover, since the male- dominated hierarchy of the public sphere tends to diminish a woman’s leverage in exercising the power she possesses (Blumberg and Coleman 1989; Ferree 1991), women, even if they have access to power/resources, may not effectively convert 220 their resources into power in a hierarchical, structural and interpersonal way, that is, to control or shape the wants or interests of those who have less power. In the case of caregiving, gendered conceptions of power can be pivotal in how men and women provide care and, especially, organize help. When men tend to think of power as control over resources and people, even if they play the primary caregiving role, they are still likely to position themselves as managers or overseers in a caregiving situation, giving commands and organizing help rather than assuming the hands-on provision of care. Men with few power/resources will be less able to influence a caregiving situation than men with more power/resources. In contrast, in a caregiving situation that is characterized by the ”natural” relations between women and care, female caregivers may have difficulty exercising the power/resources they gain from income or work. To alter the traditional division of labor, as Hardesty and Bokemeier (1989) and others (Blumstein and Schwartz 1991; Rodman 1972; Gillespie 1971) have suggested, women’s exercise of power/resources and time need to be mediated through egalitarian gender role attitudes. As explained by Ross (1987), women, even if holding gender egalitarian attitudes, face more ideological and structural barriers than men when they try to enhance their social status. At the societal level, ideology continues to define women’s work as unpaid reproductive work, despite the increase in women’s entry into the labor market, and women’s access to structural power has been blocked by men’s control of the economy and polity. In the absence of structural power or change in the values of the men around them, women’s egalitarian gender ideology is unlikely to effectively alter the gendered division of labor. As Chafetz 221 (1990) explains, because gender ideologies provide fewer restrictions and more "real" and positive rewards for men than they do for women, traditional household patterns are likely to alter only when men favor egalitarian roles and are willing to envisage change in the gendered division of labor within the family. This explains why, in domestic labor research, it is usually husbands’ egalitarian attitudes, not wives’, that are significant in altering traditional patterns of work in the division of labor (Baxter 1992; Ross 1987; Seccome 1986; Huber and Spitze 1983)?5 Although women’s enactment of gender egalitarian beliefs hinges on the structural power they possess, it is unknown if individual-level gender role attitudes, in the case of caregiving, will prevail in the face of the overshadowing societal ideology that emphasizes women’s nurturance role in caregiving. Although gender ideology at both levels exerts influences on gender relations, the macro level exerts, in general, more influence than does the micro level (Coleman 1988). One may voice an egalitarian viewpoint "but actually be pulled more strongly by the more general sexist learning of the macro level” (Coleman 1988, p.143). This especially may be the case for women in caring. The dominant societal ideology, sustained by male dominance in the political economy, has normatively assigned tasks to women and men in caregiving. A woman caregiver who voices an egalitarian viewpoint about women’s place at work or in the household may actually be pulled more strongly by the macro-level ideology that calls for women’s sense of responsibility than by her own beliefs. In other words, as argued by Blumberg and Coleman (1989), societal 5"In these studies, only the direct effect of gender-role attitudes was specified in the model. 222 ideology on gender acts as a "discount" factor at the macro level, diminishing a woman’s ability to enforce her beliefs or to wield her resources in family negotiation. Future research should explore how the gender ideology that prevails at the societal level mediates the effect of individuals’ gender-role attitudes. The feminist perspective provides a theoretical grounding for understanding the gendered division of caregiving labor among adult children (Finley 1989). When using this perspective to account for spouses’ caregiving experience, however, two additional factors related to the caregiving context must be taken into consideration: (1) marital norms and the expectations of spouse caregivers, and (2) elements related to stage in the life cycle such as age and health condition. The labor spouses provide in a caregiving setting is very different from that invested in household tasks within the average family. The obligatory and reciprocal components of the marital relationship emphasize the obligation of husbands and wives to fulfill each other’s instrumental and emotional needs when a spouse becomes ill. The relationship between the caregiver and care receiver is also fundamentally different for spouses than it is for adult children. The caring-as-feeling that sustains the emotional and symbolic bonds between the caregiver and care receiver is more likely to come into play among spouses than between parents and adult children, thus enabling spouses to move away from stereotypic gender roles. Nevertheless, the social organization of caregiving is a complex process in which the transaction of goods and services takes place within a helping network involving both informal and formal helpers. The amount of help primary caregivers receive can be as gendered for spouses as it is for adult children; spouses’ way of 223 seeking and receiving help can be equally, if not more significantly, shaped by the societal definitions of caring as women’s work than that of their children. As Blumberg and Coleman (1989) suggest, the prevailing societal ideology that equates caring with femininity may be so strong a macro—level ”discount" factor for an older cohort that it prevails over their power/resources or personal gender beliefs as they organize help in the care of their spouse. Despite their personal beliefs or structural power, elderly wives are likely to be governed by the caring as nurturing feeling and, therefore, unlikely to relegate tasks to others, particularly if these tasks are associated with traditional female gender roles. Further, as noted in previous chapters, due to the prevailing conceptions of male helplessness and women’s long-lasting caretaker role, that are sustained by the public/private dichotomy, wives’ long-lasting caretaker role is likely to be taken as an expected duty whereas husbands are likely to gain sympathy from informal helpers and thus be encouraged to seek outside help with their new caretaker role. Finally, to explain the role power/resources or time play in spouse caregivers’ behaviors, a feminist framework should take into consideration stage in the life cycle. The care provided by spouse caregivers who reach a late stage in the life cycle, for example, is less likely to be affected by structural arrangements, such as power/resources and time or gender-role beliefs, than by situational factors such as care recipients’ and caregivers’ physical health or the size of social networks, which usually decline as caregivers age (Miller and McFall 1991). This decline, however, may occur at a different pace for women than for men. Moreover, because expectations about women’s and men’s helplessness differ, men may receive more 224 assistance from those who remain in their networks than do women. A Tentative Testing The gendered concepts included in the framework presented above provide a starting point for the development of a model that explores the gendered division of labor. They suggest that individuals’ gender-role attitudes, time, and power/resources need to be placed in a sex/gender system to understand their association with male and female caregivers’ task involvement and the division of caregiving labor. In addition, any conceptual model adopted by future research should take into account the interactional relationship of gender-role attitudes with power/resources. This study did not attempt to propose, thus test, a conceptual framework based on some of the discussion above for the following reasons. First, statistical techniques that can be used to test a causal model involving intervening relationships are limited by the nature of outcome variables, that is, polytomous variables. Second, an intervening model is further limited by the small number of sons (N=58). Thus, only some preliminary analyses of the group-specific logistic regression are presented to examine if each group has a different combination of predictors that exert influence on their caregiving behaviors, and what role power/resources, time availability, and gender beliefs play for each group. The son group is excluded due to the small number of cases. In addition, because the analysis is preliminary, a causal model is not tested. Rather, a model that does not consider the interaction of the concepts discussed above is adopted. Based on the discussion above, some hypotheses can be proposed, although not all of them can be tested due to the exclusion of the son group from analysis, the 225 problematic measure of power/resources, and the separate subgroup logistic regression analyses. H1: The power/resources women hold have little impact on their level of task involvement and, to a lesser extent, on how tasks are shared with secondary caregivers. In contrast, the more power/resources men have, the less likely they are involved in caregiving, especially in tasks that are labor-intensive and time- consuming, and the more likely they will be to share tasks with secondary caregivers. H2: Women’s provision of care is affected by time constraints imposed by their own family responsibility more than they are by constraints from their work schedule, particularly when tasks that demand immediate physical presence are involved. Men’s caregiving behaviors, on the other hand, are affected in a much more significant way by time constraints from work schedule conflicts whereas competing family responsibilities exert little influence on how they provide care. H3: Egalitarian gender-role attitudes have little direct impact on how women, particularly wives, are involved in caregiving. For male caregivers, particularly sons, egalitarian gender-role attitudes are positively related to their task involvement and allocation of tasks in domains traditionally defined as women’s work. H4: Although factors related to structural arrangements have an impact on the care provided by spouses, especially husband caregivers, their effect may not be as pronounced as the effect of situational factors, such as care recipients’ and caregivers’ own physical health, that are related to their stage in the life cycle. Group-specific logistic regression analyses of three indices for total task activities and two hands-on task domains are presented in Tables 6.1 through 6.3. Findings that are relevant to the discussion above are briefly summarized. The data on significance tests and the Wald statistic show that, for both husbands and wives, advanced age and caregiving situations such as a frail spouse’s dependency level and the presence of other adults in the household, are the primary determinants of caregiving behaviors. In addition, work schedule conflicts increase wives’ need to call for help and decrease their proportionate volume of care, but only when the tasks involved call for an extremely high level of time commitment, that is, personal care 83...; .35.... as... 3...... €2.59... .81.... a 39.8... 52525.. .3883 £82.... s .339. 3:232... .28.. c. 63 98.8.8... .38.... ”5.8.. 3.8.. c. ”.26. .8883 .5... c. 6.... ”£385.. .88.... ”5.8: .23.. 5 .26. .2883 a... .85 5.2.5 8a.. .n. 8...... 535.". ”can... 9.3.. ”:2... «are 5582 2.7. $358.... .359... .8 a 2... 5...... .2... 3... ”gm ”one... 6.223%... a. 333.80 358 s .28... 2283.. .9328 8.3.... 632.2 2. 062.3 8.30 .202 3.x. . ..o. v a .. ”.8. x. :. 226 8.. 8.. 8.. 8.. 8.. 8.. 8... 8.. 8... 8... 8... 8.. £382.. 8' 8.. ...N 8v «an 8.. 2.. 8.. ..fi 8.. 8n n... 588.. .o 8.80 8.8. 3.9.“ 5...... 3.8. vudfi 8.93 2.8. 2...... 3.8.. 8.....» B.8~ 8.8.. 2.5. .859... . a... 8... ..:. 8.8a 3...... we... .. 8... 8.. .850 8.8.3.835 «3&8: 2.6 9... 3... 3.. «a... 3.... : 8... Ed ...:. 8... n... .. R... 2... 89.8.0 5.. .. 8a 2... no... 8... 8... .2. 8... ..:. 8... E... .: ...:. 28.2 .22....2 8.. 8.. 8... 8.. 8.. 8.. . 8... . 3... . 3... B... .: 8... :. ..:. .26. 689.88 . 8a . 8a a... .... : 8.4 : 3.4 «a... «a... .: 8... .: ....n 8... 8.. 28... =2.23 2885.35 9.3930 : «a... . 3... 8... 8... . 8... : 8... : 8... ..: 8... .: 8... : B... . 8... .: a... on... 3.. a... 8... : «a... x... 5.. 2.... 8... . :... : 8.. 8... :. 8... 5.8.. . .3950 3 83.8255 .23 . 5... . 2.... «a... ...:. 8.. 8.. 8... «a... 8... 8... 8... 8... .8... 38:83.5 «2. I... : .2. 8.9.6 88.. ..: on... 8.. .: :..N 33232.. . 8... . 3.... . 8... a... B... 8... 2... .2. 2.... ..:. «v... 8... .85835 8... 8... 8... 8... . 8... . ...:. :... 8.. 8... 8... 8.. 8... 2.8... 29.33.. ...:. .323335Euofiaouo a < u < m < $2222.... ...x. .82.. .23: 829.6 2.3:... 8..... 32%... 2.33.. ...:. 82.98 2.3:... ...:. 5.9.380... use 3209...“... 3:55... .68. .30.. 8.3.... 5.5.... .3 £33.25: .2... .200 .233... .00—8.5“; ..uah .83. c. .300... 05 .0 0:02 .3 Icon .05. 8.3... .2 5 95.1—oasis 9.32928 3 003.1 .35 p6 Oink 227 Table 0.2 M Sets at Odds Ratio Related to Caregivere' Task Exclusivity Pattern in ‘I’otal Taalt Activltlee and Two Female-Oriented Taelt Domaina, by Femilly Relation Wife Husband Daughter ModelA Modela Excl. Care Portia! Share Excl. Care Partial Share Excl. Cane Partial Shana Excl. Care Partial Shara ve. va. va. va. Predictor: Total Dep. Total $9. TotalDep. Total Dep. TOTAL TASK ACTIVITIES ‘Reeuucealnnnfldeology Adi.houeeholdincome 1.00 ' 1.00 1.04 ' 1.02 1.00 1.00 1.00 1.00 5W1 0.03 1.02 0.15 " 0.40 0.54 0.00 0.50 0.70 Pmpnquity 3 57 " 217 " Young children 0 30 ‘ 0 00 Baum level 0.98 1.01 0.90 1.01 0.90 0.07 ‘ 0.00 0.07 ' Other W of Caregiver Health 0.04 1.20 0.55 0.90 1.40 1.00 1.32 1.02 Age 1.00 1.02 1.03 1.05 1.07 "' 1.03 " 1.05 ' 1.02 WWt Dintion otcere 0.52 0.54 0.35 ' 0.20 " 0.27 " 0.20 "' 0.20 " 025 '" Dependenq level 0.89 " 1.09 " 0.07 " 1.00 0.02 1.05 0.00 " 1.04 Additional helpers 0.42 "' 0.01 0.20 " 0.44 ‘ 0.42 " 0.72 ' 0.31 "' 0.07 ' Diagnose 0.39 " 0.40 ' 0.45 0.50 0.04 1.00 0.74 120 Can Recipient's Characteristics Gender 1.14 1.00 Marital stem: 1.71 0.04 Likelihood Ratio 057.70 420.01 040.00 010.00 Degree of Freedom 1020 510 004 050 Probabllity 1.00 1.00 1.00 1.00 PERSONAL CAR; DOMAIN y Adi. household income 1.00 0.07 1.01 0.00 0.00 0.00 ‘ 1.00 0.00 ‘ Ermioyment 0.21 " 0.05 0.27 ’ 0.10 0.34 "‘ 0.75 0.34 "' 0.73 Propinquity 2.00 " 2.10 ' Young children 0.04 1.10 Educational level 0.97 1.00 0.03 1.01 1.00 0.04 1.02 0.05 Other cum 0! Caregiver Health 0.71 0.71 0.04 0.07 1.05 1.37 0.00 1.32 Age 0.05 ' 0.95 ' 0.97 1.00 1.01 1.02 1.00 1.01 mm Duration ot core 0.02 0.04 0.52 0.40 0.00 1.20 0.00 1.14 Dependency level 0.70 "' 1.04 0.00 "' 1.04 0.02 "‘ 1.03 0.00 "‘ 1.01 Additional helpers 0.08 0.04 0.40 0.43 0.04 ' 0.07 0.50 " 0.07 Diagnosis 0.51 ’ 0.05 0.03 1.30 1.10 2.50 " 0.04 2.30 " Can Reclplent‘b Characteristics Gender 1.04 1.00 Marital statue 0.00 0.00 Likelihood Ratio 052.30 320.05 052.10 033.77 Deane of Froedorn 750 340 072 074 Probability 1.00 0.70 0.70 0.00 Housgwonxgomm Wm)! Adj. household income 0.90 1.00 0.00 0.00 0.00 0.00 0.00 0.00 Ermloyment 0.53 0.50 0.32 ' 0.00 0.50 ' 1.37 0.55 ' 1.20 Propinmity 0.03 "' 3.17 "' Young children 0.00 0.00 Educational level 0.90 0.99 0.93 0.92 0.00 ' 0.03 0.07 ° 0.02 Other Characteristics of Caregiver Health 0.40 "' 0.75 0.50 " 0.52 " 0.00 1.02 0.07 0.05 Age 1.00 1.05 ' 1.00 1.00 1.00 "‘ 1.03 " 1.04 " 1.02 Car-giving Environment Duration otcane 0.02 0.30 " 0.00 ' 0.40 0.37 " 0.40 " 0.32 "‘ 0.30 " Dependency level 0.91 " 1.00 0.00 " 0.03 0.00 1.01 0.04 ' 0.00 Additionel helpers 0.30 ”‘ 0.70 0.32 " 0.45 ° 0.71 ' 0.00 0.52 "‘ 0.04 Diagnosis 0.40 ' 0.75 0.09 1.21 0.74 1.00 0.00 0.05 Care Recipient‘b Char-eroded“ Gender 1.04 1.15 Mental etatm 0.07 0.43 ' leelihood Ratio 741.72 400.25 051.77 000.10 Degree of Freedom 004 470 014 000 Probability 1 .oo 0.47 0.20 0.50 Note: ”‘ p< .001; “ p < .01; ' p<.05; for coding scheme. see Table 6.1 228 Table 6.3 Two Sets ot Odds Ratio Related to Careglvers' Relatlve Contrlbutlon Pattern ln Total Taslr Actlvltlee and Two Fernale-Orlented Task Domalns. by Famllly Relatlon WIte usband Daughter ModeIA Hedda Excl. Care Comp. Care Excl. Care Comp. Care Excl. Care Comp. Care Excl. Care Comp. Care vs. vs. vs. vs. Predlctors Supp. Cara Supp. Care Supp. Care Supp. Care I TOTAL TASK acnvmes Reecurcea/Tlme/ldeology Adj. household income 0.96 ‘ 0.96 1.00 0.97 1.00 0.99 1.00 1.00 Employment 0.79 0.45 0.16 ' 0.54 0.30 " 0.36 "‘ 0.24 " 0.32 "‘ Propinquity 13.52 "‘ 10.46 "' Young children 0.25 ' 0.56 Educational level 0.90 0.92 0.63 0.92 1.04 0.93 1.02 0.93 Other Characterisflcs o! Caregiver Health 0.39 " 0.45 ‘ 0.35 " 0.56 1.62 1.23 1.37 1.05 Age 0.96 0.97 0.94 0.95 1.05 ' 1.00 1.02 0.96 CareglvlngEnvlronrnent Duration of care 0.39 0.42 0.60 0.60 0.61 1.04 0.79 0.91 Dependencylevel , 1.00 1.23 "‘ 0.64 " 1.04 0.90 ' 1.02 0.64 "' 0.96 Additional helpers 0.37 " 0.60 0.14 "' 0.24 ”' 0.69 1.33 0.43 ' 0.96 Diagnosis 2.69 4.57 ' 0.77 1.06 0.51 0.74 0.29 ° 0.43 ' Car-Recipient’s Charactenstlcs Gender 0 70 1 00 Mattel status 1 06 0 54 Likelihood Ratio 725.62 375.66 624.64 559.33 IDegree at Freedom 1026 516 664 use Probability 1 .00 1 .00 1.00 1.00 I PERSONAL CARE DOMAIN Recount-meadow Adi. household income 1.00 0.99 1.01 0.99 0.99 0.96 ' 0.99 0.96 Employment 0.06 "' 0.20 ' 0.19 ' 0.25 ' 0.33 "‘ 0.69 0.32 "‘ 0.66 Propinqurty 3.56 "‘ 3.24 "‘ Young children 0.59 0.49 Educational level 0.92 0.96 0.66 0.93 0.96 0.90 0.99 0.92 Other Characteristics of Caregiver Health 0.61 0.69 0.63 0.91 0.95 1.03 0.66 0.99 Age 0.69 "' 0.69 " 0.97 1.00 1.00 1.00 0.96 0.96 Caregiving Environment Duration otcere 0.36 ° 0.52 0.61 0.66 ' 1.19 2.53 " 1.05 2.36 “ Dependenqlevel 0.77 "' 1.02 0.63 " 1.06 0.90 " 1.19 "‘ 0.66 "' 1.16 "' Additional helpers 0.56 0 76 O 39 0.46 0 62 1 60 " 0.74 1.45 Diagnosis 0.62 1.43 1.21 1.69 0.76 0.95 0.61 0.64 Care Recipient‘s Characteristics Gender 1.61 0.72 Marital 61811.16 0.54 0.67 Llltellhood Ratio 572.32 314.45 629.66 ”1.57 Degree 01 Freedom 756 340 672 664 Probability 1 .00 0.64 0.66 0.96 r Housewomr oougam Reaourcea/Tlrnefldeology Adi. household income 0.95 ' 0.96 0.96 0.99 0.96 0.99 0.99 1.00 Employment 0.24 0.27 0.10 " 0.27 ' 0.29 " 0.46 ° 021 “‘ 0.33 "‘ Propmquity 21.32 "‘ 10.60 "' Young children 0.34 " 0.46 ' Educational level 0.95 0.97 0.64 0.64 0.62 ‘ 1.00 0.91 ' 0.99 Other Characteristics o! Caregiver Health 0.34 ”‘ 0.56 ‘ 0.46 ' 0.61 0.96 1.05 0.61 0.69 Age 0.96 1.00 0.69 °' 0.66 " 1.05 " 1.01 1.02 0.99 CereglvlngErrvlrunrnent Duration of care 0.36 0.32 0.67 0.95 0.55 1.01 0.49 0.69 Dependency level 0.95 1.06 0.67 ' 0.96 0.99 1.01 0.90 " 0.94 Additional helpera 0.26 "' 0.53 °' 0.12 "‘ 0.16 "' 1.01 1.56 " 0.65 1.13 Diagnosis 1.25 2.54 0.63 1.04 0.56 0.72 0.32 " 0.45 ' Care Recipient‘s Characteristics Gender 0.65 0.63 Marital status 0.56 0.49 leelihood Ratio 650.11 410.66 799.04 717.63 Degree of Freedom 664 476 614 606 Probability 1.00 0.99 0.64 0.99 Note: ”‘ p< .001; " p < 01; ' p<.05; forcodlng scheme. see Table 6.1 229 tasks. For husbands, in contrast, the effect of employment on task exclusivity and relative contribution can be overwhelmingly found not only in the personal care domains but also in total task activities and the housework domain. In other words, the difference in the employment effect for husbands and wives occurs primarily in total task activities and, especially, housework. Caregiving wives are more likely to plan and do housework so that it does not interfere with work whereas working full- time legitimates husbands’ use of outside help not only with personal care but also with housework. Although only a limited conclusion can be drawn due to the separate subgroup regression and the small percentages of husbands (16%) and wives (9%) employed full-time, work schedule conflicts appear to be more of a factor for husbands than for wives as they provide care. Time factors such as propinquity, work schedule conflict, and competing family responsibility also impose constraints on the ability of daughters, who are often caught in the middle, to provide care. Moreover, data based on the Wald statistic show that, whereas the caregiving situation is relatively important for spouse caregivers, propinquity is the primary determinant of daughters’ pattern of care provision. Contrary to the proposed hypothesis, daughters’ ability to help appears to be constrained more by conflicts imposed by work schedules than by competing family responsibilities. This significant finding contradicts the conclusion of some researchers who found that employment has no impact on daughters’ provision of care (Enright 1991; Stoller 1983), but it is consistent with the conclusion of other researchers (Brody and Schoonover 1986; Barnes, Given, and Given 1995). In the absence of comparison data for sons, it is unknown how the effect of work schedule 230 conflicts is gendered. Nevertheless, based on the findings for spouse caregivers, sons may also experience more work time constraints than do daughters who work full- time. The group-specific logistic regression analysis raises questions about the problematic measure of education as a proxy for gender-role attitudes. If education truly reflects an individual’s beliefs about gender roles, better educated men would rely less on other helpers to assist them in personal care or housework tasks. The relationship between education and the division of labor, however, is ambiguous. The same negative direction of educational effect for men and women, though insignificant for husbands, raises doubts about using education as a proxy measure for gender-role attitudes. In future research, a direct measure of gender-role attitudes is undoubtedly preferred. In addition, the results of the group-specific analyses are still unable to show if the effect of age cohort reflects a subculture of gender-role beliefs or the size of social networks. On the one hand, the positive relationship between age and task exclusivity in the housework domain for all caregiver groups, though insignificant for husband caregivers, is consistent with the assumption about age as an indictor of the size of social networks. On the other hand, the opposite direction of the effect of age on relative contribution in the housework domain for daughters and husbands is consistent with the assumption that age cohort reflects a subculture of gender-role beliefs. The pattern of age effects on the division of labor raises doubt about the use of age as a proxy measure of caregivers’ health, especially when young caregivers are sampled. 231 In sum, although limited interpretations can be drawn from the group-specific logistic regression of primary caregivers’ caregiving behaviors, the data show that first, for spouse caregivers, caregiving situations and factors associated with a late stage of the life cycle are the primary determinants of their care providing behavior. Second, for wife caregivers, providing help in labor-intensive personal care tasks is inhibited by full-time employment but helping with housework is arranged around a work schedule. For husband caregivers, however, time constraints from work inhibit their assistance in most areas. As mentioned earlier, because personal/family time and work time are intertwined in women’s conception of time, they are likely to employ various adaptations to meet the combined demands of work and caregiving.56 In contrast, paid employment releases men from domestic responsibilities more than it does women because men’s work demands are more likely to be acknowledged, and men are more likely to use paid work to legitimize their relegation of tasks to helpers. Third, contradictory to findings in other research (Stoller 1983), time, including work time, is of great concern to daughters when they provide care and incorporate help. Summary In this chapter, a feminist theory of gender stratification was proposed to understand the persistent gendered differences in caregiving. The gendered division of labor is rooted in the intersection of capitalism and patriarchy which shapes the context of the private and public and ideologically defines women’s place as the ”Franklin, Ames, and King (1994), using part of the same data set (CG2), found that female caregivers who were employed adapted their work schedules by taking leaves of absence and adopting other short-term adjustments such as taking sick or personal days, arriving late or leaving early, altering work hours, and missing work without pay. 232 home. The gendered division of caregiving labor reflects the public/private dichotomy that perpetuates women’s commitment to traditional gender roles. It was suggested that mechanisms which may be responsible for the variation in caregiving behaviors, such as time, power/resources, and gender-role attitudes, should be located within the context of a sex/gender system to understand their specific influence on how men and women provide care. It was also suggested that the unique patterns of spouses’ caregiving behaviors be taken into account in the development of any model because marital norms and expectations, age cohort, and stage in the life cycle may confound the effects that power/resources, time, and role ideology have on the way husbands and wives provide care. CHAPTER 7 CONCLUSION AND PUBLIC IMPLICATION The major purpose of the study was to identify how men and women differ in the care they provide to their frail elderly relatives in a home care setting and how that relationship is confounded by the family relationship of caregiver to care recipient. This research complemented past research on gender and caregiving in many ways. First, caregiving tasks were conceptualized as task domains that comprise several dimensions of demands and roles expected of caregivers. Four task domains, in contrast to the two (i.e., ADLs and IADLs) commonly used in gerontological research, were carefully constructed from twenty caregiving tasks. Second, primary caregivers’ provision of care was considered not only by itself but also in the context of networks of support. Primary caregivers’ level of task involvement and two division of labor indices, task exclusivity and relative contribution, were constructed for this study to capture the multi-dimensional facets of caregiving behaviors. Third, multivariate logits regression analysis of the care provided by primary caregivers took into account gender and factors with which gender might covary. The methodological approach taken in this study thus provided a better, systematic, empirical base for estimating gender differences in patterns of care provided by primary caregivers than does existing research. Several sociological perspectives have been proposed in the caregiving literature to explain gender differences in caregiving: power/resources, time availability, gender 233 234 role ideology, specialization of tasks, and cultural taboos. A review of critiques of these perspectives revealed that the institutionalized gender difference "has virtually taken on a life of its own, unresponsive to household variation in potential efficiency, male power, and ideology” (England and Farkas 1986, p.99). This study, with a refined methodological design and enhanced measurements of outcome variables, adopted a gender-difference approach to show the persistent gender differences in caregiving after all known factors with which gender covaries were controlled. A feminist theory of caregiving was introduced to explain how power/resources, time availability, the ideology of roles within the context of the sex/gender system are gendered and have produced and reproduced the gender structure of caregiving. Some preliminary findings about the factors associated with the caregiving behaviors of men and women were then presented to disclose why men and women provide care differently. This chapter presents a summary of the findings and proposes the policy implications of the research. Summary of Findings Background Characteristics of Primary Caregivers The study involved an analysis of survey data produced by six panel home-care studies of primary caregivers, including spouses and adult children, who provided care to relatives with Alzheimer’s disease, cancer, and a broad range of physical impairments. The analysis showed that the background characteristics of the elder and primary caregivers differ when the primary caregiver is a man or a woman. Daughters are more likely than sons to care for a parent who is severely impaired. Achieving the same level of educational attaintment as do sons, daughters are more 235 likely than sons to be confined within the private sphere as they provide care and to be caught in the middle of competing obligations to caregiving and their own family. Nevertheless, when duration of care is taken into account, the data showed that sons who provide long-term care are relatively socially disadvantaged. They are less likely to be college educated and to hold a professional occupation than are those new to the caregiving role; they are thus considerably poorer. In contrast, long-term daughter caregivers are somewhat better off, socially and financially, than are their short-term counterparts as well as long-term son caregivers. They tend to be professionals and college educated, although largely retired or employed part-time. As the social and economic status of long-term caregiving daughters approaches to that of long-term sons caregivers, the impairment level of the parent for whom they care narrows as well. In contrast, caregiving husbands and wives, who are constrained by the normative expectation of marriage, are unlikely to differ in the length of time they occupy the caregiving role or the impairment level of the spouse for whom they care. The primary difference lies in their socioeconomic status. Wife caregivers in the older age cohort tend to be less educated, out of the labor force, to have held gender- specific jobs, and to have lower household incomes than husband caregivers; in fact, elderly wife caregivers had the lowest income among all caregiver groups. Finally, designated primary caregivers in the family do not hold the same attitudes about their obligation to provide help to aging relatives. As expected, spouses show a stronger sense of role responsibility to care for their frail loved ones than do adult children who care for their parents. Husbands’ sense of role obligation 236 about their caregiving role is stronger than that of wives whereas there is no gender difference between caregiving daughters and sons. As their caregiving career lengthens, however, the sense of role responsibility plummets for all groups of caregivers, probably a result of the fact that they feel their obligation has been fulfilled by the care they have already provided. Task Participation and the Division of Labor The data showed that family relationship of primary caregivers to care recipients plays an important role in how care is provided. The range of task involvement by spouses is wider than that of adult children. When primary caregivers’ assistance is examined within the context of networks of support, however, spouses are not very different from adult children in their tendency to use outside help, especially when there are others in their helping networks. Nonetheless, the extra assistance spouses receive is not likely to relieve their caregiving burden. The amount of total care secondary caregivers provide is much lower when the primary caregiver is a spouse than an adult child. The strong normative expectation of marriage and sense of role responsibility emphasized for spouse caregivers may affect their level of task involvement in ways that adult children do not experience. Nevertheless, the availability of support networks, secondary caregivers’ perception of spousal care, and resources with which the elderly can negotiate help may also explain why spouses, who are not necessarily different from adult children in their use of outside help for task assistance, simply receive less help from other caregivers. Aside from family relationship, gender is a significant predictor of the care provided by primary caregivers, child caregivers in particular. Sons and daughters are 237 likely to differ in the level of participation in tasks that are an extension of women’s reproductive roles but not in tasks that are congruent with men’s gender roles. Husbands and wives, on the other hand, have less discretion about which tasks to assume, probably because spouses are likely to view caregiving of a frail spouse as a normative expectation of marriage. Nevertheless, care provision takes place within the context of social networks, in which goods and services provided by informal and formal helpers can be affected by the prevailing societal ideology that equates caring with femininity. This study showed that gender is the organizing principle of the division of caregiving labor, in terms of both task exclusivity and relative contribution, for both spouses and adult children, particularly when tasks are labor intensive, time demanding, and congruent with women’s gender role. In domains that are extensions of women’s reproductive role, female caregivers, including both wives and daughters, are more likely than male caregivers to provide care without help and they receive less help from secondary caregivers than are male caregivers. In domains that are extensions of men’s gender roles, primary male caregivers are not any more likely than primary female caregivers to take on the tasks alone or to provide a greater proportionate volume of care. This study showed that the length of time primary caregivers have performed the caregiving role affects how care is provided. Whereas the duration of care is irrelevant to spouse caregivers’ task participation, again reflecting the strong sense of role responsibility emphasized for spouse caregivers in the marriage contract, adult children are likely to be actively involved, particularly in labor-intensive and time- demanding domains, as their caregiving career lengthens. When the effect of duration 238 of care is examined in the context of networks of support, the data showed that as primary caregivers’ caregiving career is extended, both spouses and adult children are less likely to provide care all by themselves, suggesting that the size of primary caregivers’ helping networks is likely to increase. The task burden of long-term spouse and child caregivers, however, is not likely to be relieved by the extra help they receive. The data also showed that long-term child caregivers, including both daughters and sons, tend to assume intense and time-consuming personal care tasks but not necessarily intermittent tasks that involve money management. Moreover, they do not necessarily receive help with personal care tasks as they do for other task domains, and they thus provide a greater proportionate volume of care in personal care than do child caregivers new to the caregiving role. In contrast, when providing long-term care, spouses are likely to assume responsibility in administrative tasks but are not necessarily actively involved in the labor-intensive personal care tasks. In fact, long- term spouse caregivers are likely to share personal care tasks that are characterized by both intimacy and labor intensity with other helpers and thus provide a lesser proportionate volume of care than do those new to the caregiving role. These findings highlight the unusual commitment of adult children who provide long-term care (when compared to short-term child caregivers) to help with the most labor-intensive, time- demanding personal care tasks. The unusual commitment of long-term child caregivers, at least among sons, may be attributed to their socially disadvantaged status, which gives them little power and resources to negotiate help within the family or for paid help. The findings also highlighted the willingness of spouses to relinquish 239 their caregiving responsibility in intimate personal care tasks when their caregiving career is extended. This study confirmed the taboo hypothesis in many ways. Spouses are more likely than adult children to be involved in intimate personal tasks than in any other type of tasks and they, at least among those new to the caregiving role, are reluctant to seek outside help with these tasks. In addition, the gender of frail parents affects how adult children provide help with personal care tasks. Nevertheless, this study also showed that personal care tasks are more than just tasks involving bodily contact. Caregivers often perceive personal care as nurturing tasks that are tedious, time consuming, and often demanding of round-the-clock attention. Therefore, it is not surprising that spouses are likely to seek outside help with personal care tasks after their caregiving career is extended. In addition, the fact that personal care tasks involving nurturance and attendance to others’ needs coincide with women’s socially constructed caretaker role explains why gender remains a major organizing principle of spouses’ and adult children’s care provision in personal care. This study took a gender-difference approach to examine how men and women differ in the care they provide to their frail elderly relatives while all known variables are controlled. Without considering the differential effects of time, power/resources, and ideology by gender, this multivariate analytical approach tested their effects under the assumption that they Operate in the same way for men and women. The data showed that time constraints are indeed a concern for caregivers as they provide care, particularly when they decide whether or not to seek outside help. For adult children, the foremost time factor is that imposed by geographic distance rather than work 240 constraints or competing family responsibilities. In contrast to the strong and consistent effects of the time factor, the effect of caregivers’ adjusted household income varies somewhat from domain to domain. Education, as the proxy measure of gender egalitarian ideology, exerts influence primarily in the housework domain that is an extension of routine domestic responsibilities in a caregiving setting. The limited support for power/resources and gender ideology hypotheses may have been a result of the problematic operationalization of the variables due to the limitations of the data. Future research should operationalize the absolute as well as relative earning of primary caregivers and use a direct measure of primary caregivers’ gender-role attitudes. This study also showed the importance of structural limitations associated with the age of caregivers and the caregiving situation. These were especially significant in the decision of primary caregivers to call in help. Among all variables, demands of care and, to a lesser extent, additional adults residing in care recipients’ household, stand out as the foremost determinants of primary caregivers’ use of help and the amount of help they receive from other caregivers. Although the problematic measure of power/resources and gender ideology limit the scope of generalization, this study revealed a gender-based, aside from a relationship-based, division of labor. It is argued that women and caregiving must be understood within the context of the economic and political forces that shape the consciousness of women and men and rely on the material contribution of women’s caring role to the maintenance of capitalism and patriarchy. Power/resources, time, and gender ideology are products of the sex/gender systems and affect how men and 241 women provide care. A conceptual framework based on the gender stratification theory was proposed to examine how the effect of power/resources, time, and ideology about women and men are gendered and how these gendered mechanisms have produced and reproduced gender differences in caregiving. Policy Implications To relieve the burdens of family caregivers, several strategies have already been proposed or delivered. These include the use of tax breaks, cash incentives, the expanded use of home health care and respite services, and the growth of caregiver training programs and support groups (Day 1985; Abel 1989). For example, family caregivers may deduct work-related expenses for the care of a disabled elderly person who is living with them. A variety of monetary incentives including direct monthly payments for services, vouchers, and low-interest loans are available to family caregivers. Programs and services such as respite care, day care for the elderly, visiting nurse, in-home help, overnight and temporary care have been funded through Medicaid. A variety of educational and training programs have also been delivered to offer caregivers advice about disease processes, community resources, and legal and financial matters, and to teach caregivers basic survival skills such as cooking, shopping, and caring for an impaired person. Support groups that provide family caregivers emotional support and promote personal adjustment have also been proliferating. From a health policy perspective, this study highlights the need to frame a policy agenda that is responsive to gender as well as to relationship or cohort (Hooyman 1992; Osterbusch et al. 1987). It should take into account the special 242 situation of women, particularly that of elderly wives, as caregivers, broadening their options and addressing their needs. Intervention strategies need to be appropriately targeted in relation to the gender and family relationship of primary caregivers (Noelker and Wallace 1985). For example, some form of direct financial assistance, rather than cash incentives such as tax credits, would most appropriately benefit spouse caregivers, especially elderly wives who are often at risk of impoverishment when their husbands become ill. Female caregivers, especially elderly wife caregivers, may also be most helped by both respite and in-home services such as housework and, especially, personal care, that would periodically relieve them of the strains of physical care. Medicare regulations, however, specifically rule out personal care services in its coverage; and the eligibility criteria of Medicaid and the range of in-home services provided differ from state to state (Manheimer 1994; Maddox 1995; Crichton 1987).57 More government subsidies are necessary to ensure that formal help with housework and personal care tasks is offered widely, thereby providing true relief for primary 5"’Medicare is designed to meet the need of people over 65 who require acute care. Therefore, most nursing home care is not covered by Medicare; neither are home health service, unless there is a skilled nursing or rehabilitation need (Manheimer 1994). To remedy the limitation of coverage of medicare, state-run Medicaid provides home care services, including medically-related home health care and in-home chores and personal care, for the chronically ill and home-bound elderly, but primarily those who are poor. The in-home help includes homemaker services, such as cooking and housecleaning, and personal care services, such as assistance in hygiene, grooming, and dressing. Although all states have chosen to provide some homemaker or personal care services for the disabled elderly residing at home under some federally subsidized programs, the eligibility criteria and the range of services provided differ from state to state (Manheimer 1994). Whereas Medicaid is largely limited to serving the poor, in many cases, Medicaid eligibility was obtained after the costs of health care and long-term-care exhausted the elderly persons’ resources (Maddox 1995). 243 caregivers, particularly female caregivers, who are most burdened in these areas. In addition, training programs that focus on basic survival skills such as cooking, household cleaning, and other personal care tasks should be made an integral part of referral programs for male caregivers. Other non-gender responsive intervention programs may also indirectly help women provide care. For example, tax credits for child caregivers who live with a disabled parent should be increased to reflect the actual cost of care they provide due to coresidence, and the tax credit should not be limited only to caregivers who are employed as is currently the case (Osterbusch et al. 1987). Tax credits should also take into account the number of non-disabled adults coresiding with care recipients, who prove to be a source of relief for primary caregivers. Nevertheless, because caregivers generally think services that ease their caregiving burden and the support of others are more crucial than financial incentives (Abel 1989), government subsidized services and incentives that encourage a broad array of community-home- based services to augment secondary caregivers’ assistance should be the top priorities among all services delivered. Of all services that are currently delivered, restructuring the work environment, so that the impact of a combined workload of caregiving and paid employment and the interruption of work careers are minimized, may be a crucial solution to women’s disadvantage in caregiving. Alternative work arrangements, such as job sharing and flexible work hours, are essential for many caregivers who seek to combine paid employment and caregiving responsibilities (Glendinning 1992; Seccombe 1992). In addition, employee assistance programs that provide employees with elder care 244 information, referral programs, and community resources may also reduce job turnover and enhance the ability of employed caregivers to handle their multiple responsibilities. Most of all, because women’s economic dependency on men is the primary basis of the gendered division of caregiving labor, caregivers who interrupt their work lives to care for the elderly need guarantees that their jobs will await them when they return and that they can continue to accrue pension credits (Abel 1989). Aside from corporate policies on employment and elder care, a broad-scale reform of the nursing home industry is needed. Nursing home must provide humanizing and personal care to the elderly so that adult children may be relieved of the guilt that often comes when they make the decision to relinquish the caregiver role to strangers. Finally, a community care policy that is no longer rooted in traditional concepts about the family and women could help broaden women’s caregiving alternatives and destabilize the notion that caregiving is women’s role. In all, women’s subordination in caregiving will be fundamentally change only when both corporate and government policies relieve their burden as caregivers, provide them caregiving alternatives, and empower women in the caregiving process. BIBLIOGRAPHY BIBLIOGRAPHY Abel, Emily K. 1986. "Adult Daughters and Care for the Elderly." Feminist Studies 12(3): 479-497. Abel, Emily K. 1989. "Family Care of the Frail Elderly: Framing an Agenda for Change." Women ’s Studies Quarterly 1-2: 75-87. Abel, Emily K. 1990a. ”Informal Care for the Disabled Elderly: A Critique of Recent Literature." Research on Aging 12: 139-157. Abel, Emily K. 1990b. ”Family of the Frail Elderly." Pp. 65-91 in Emily K. Abel and Margaret K. Nelson (eds.), Circles of Care: Work and Identity in Women’s Lives. Albany: State University Abel, Emily K., and Margaret K. Nelson. 1990. ”Circle of Care: An Introductory Essay." Pp. 4-34 in Emily K. Abel and Margaret K. Nelson (eds.), Circles of Care: Work and Identity in Women’s Lives. Albany: State University Acker, Joan. 1989. "The Problem with Patriarchy." Sociology 23: 235-240. Ahn, Helen Noh, and Neil Gilbert. 1992. "Cultural Diversity and Sexual Abuse Prevention." Social Service Review 6(3): 410-27. Albert, Steven M. 1991. "Cognition of Caregiving Tasks: Multidimensional Scaling of the Caregiver Task Domain." The Gerontologist 31(6): 726-734. Albrecht, Stan L., Howard M. Bahr, and Bruce A. Chadwick. 1979. ”Changing Family and Sex Roles: An Assessment of Age Differences.” Journal of Marriage and the Family 41: 41-50 Allen, Susan M. 1994. ”Gender Differences in Spousal Caregiving and Unmet Need for Care. " Journal of Gerontology 49(4): 8187-8195. Andorka, Rudolf. 1987. "Time Budgets and Their Uses. " Annual Review of Sociology 13: 149-64. Antonucci, Toni C., and Hiroko Akiyama. 1987. ”An Examination of Sex Differences in Social Support Among Older Men and Women." Sex Roles 17: 737-749. 245 246 Arber, Sara, and Jay Ginn. 1991. ”The Invisibility of Age: Gender and Class in Later Life." Sociological Review 39: 260-291. Archbold, PG. 1983. ”An Impact of Parent-Caring on Women." Family Relations 32(January): 39-45. Aronson, Jane. 1992. ”Women’s Sense of Responsibility for the Care of Old People: ’But Who Else Is Going to Do It?” Gender and Society 6(1): 8-29. Barber, Clifton E. 1988. "Correlates of Subjective Burden among Adult Sons and Daughters Caring for Aged Parents." Journal of Aging Studies 2(2): 133-144. Barnes, Carla Leddy, Barbara A. Given, and Charles W. Given. 1995. ”Parent Caregivers: A Comparison of Employed and Not Employed Daughters.” Social Work 40(3): 375-381. Barnett, Rosalind C., and Grace K. Baruch. 1987. ”Determinants of Fathers’ Participation in Family Work. " Journal of Marriage and the Family 49: 29-40. Baxter, Janeen. 1992. "Power Attitudes and Time: The Domestic Division of Labour. " Journal of Comparative Family Studies 23(2): 165-182. Becker, Gary. 1981. A Treatise on the Family. Cambridge, MA: Harvard University Press. Benhabib, Seyla, and Drucilla Cornell. 1987. Feminism as Critique: On the Politics of Gender. Minneapolis, MN: University of Minnesota Press. Berardo, Donna H., Constance L. Shehan, and Gerald R. Leslie. 1987. "Jobs, Careers, and Spouses’ Time in Housework." Journal of Marriage and the Family 49: 381-390. Berk, SF. 1985. The Gender Factory: 77w apportionment of work in American households. New York: Plenum. Berk, Richard A., and Sarah Fenstermaker Berk. 1978. ”A Simultaneous Equation Model for the Division of Household Labor.” Sociological Methods and Research 6(4): 431-468. Binstock, R.H., and E. Shanas (Eds). 1985. Handbook of Aging and the Social Sciences (2nd ed. ). New York: Van Nostrand Reinhold. Blood, Robert O, and Donald M. Wolfe. 1960. Husbands and Wives. New York: Free Press. Blumberg, Rae Lesser. 1984. ”A General Theory of Gender Stratification." Pp. 23- 247 101 in Randall Collins (ed.), Sociological Theory. San Francisco: Jossey-Bass. Blumberg, Rae Lesser, and Marion Tolbert Coleman. 1989. ”A Theoretical Look at the Gender Balance of Power in the American Couple.” Journal of Family Issues 10(2): 225-250. Blumstein, Philip, and Pepper Schwartz. 1991. "Money and Ideology: Their Impact on Power and the Division of Household Labor." Pp. 261-288 in Rae Lesser Blumberg (ed.), Gender, Families, and Economy: The Triple Overlap. Newbury Park, CA: Sage. Boaz, Rachel F ., and Charlotte F. Muller. 1992. "Paid Work and Unpaid Help by Caregivers of the Disabled and Frail Elders." Medical Care 30: 149-158. Brody, Elaine. 1981. "Women in the Middle and Family Help to Older People." The Gerontologist 21: 471-79. Brody, Elaine. 1990. Women in the Middle: Their Parent-Care Years. New York: Springer. Brody, Elaine M., and Claire B. Schoonover. 1986. ”Patterns of Parent-Care When Adult Daughters Work and When They Do Not. " The Gerontologist 26(4): 372-381. Brody, Elaine, Kleban, M.H., Johnsen, P.T., Hoffman, C.B., and CB. Schoonover. 1987. ”Work Status and Parent Care: A Comparison of Four Groups of Women. " The Gerontologist 27: 201-208. Brody, Elaine, Johnsen, P.T., Fulcomer, M.C., and A.M. Lang 1983. "Women’s Changing Roles and Help to Elderly Parents: Attitudes of Three Generations of Women. " Journal of Gerontology 38: 597-607. Brubaker, Timothy H. 1985. "Responsibility for Household Tasks: A Look at Golden Anniversary Couples Aged 75 Years and Older." Pp. 27-36 in Warren Peterson and Jill Quadagno (eds), Social Bonds in Later Life. Newbury park, CA: Sage. Calasanti, Toni M., and Carol A. Bailey. 1991. ”Gender Inequality and the Division of Household Labor in the United States and Sweden: A Socialist-Feminist Approach." Social Problems 38(1): 816-833. Cantor, Majorie H. 1979. ”Neighbors and Friends: An Overlooked Resource in the Informal Support System.” Research on Aging 1: 434-63. Cantor, Majorie H. 1983. ”Strain among Caregivers: A Study of Experience in the US" The Gerontologist 23: 597-624. 248 Cantor, Majorie, and V. Little. 1985. ”Aging and Social Care." pp. 745-81 in R. Binstock and E. Shanas (eds), Handbook of Aging and the Social Sciences. New York: Van Nostrand Reinhold. Carmines, Edward G., and Richard A. Zeller. 1979. "Reliability and Validity Assessment.” Newbury park, CA: Sage. Chafetz, Janet Saltzman. 1988. "The Gender Division of Labor and the Reproduction of Female Disadvantage: Toward an Integrated Theory." Journal of Family Issues 9(1): 108-131. Chafetz, Janet Saltzman. 1990. Gender Equity: An Integrated Theory of Stability and Change. Newbury Park, CA: Sage. Chambers, Deborah A. 1986. "The Constraints of Work and Domestic Schedules on Women’s Leisure.” Leisure Studies 5: 309—325. Chang, Cyril F., and Shelley I. White-Means. 1991. ”The Men Who Care: An Analysis of Male Primary Caregivers Who Care for Frail Elderly at Home. " Journal of Applied Gerontology 10(3): 343-58. Cicirelli, V.G. 1983. A Comparison of Helping Behavior to Elderly Parents of Adult Children with Intact and Disrupted marriages. The Gerontologist 23: 619-625. Clark, Robert A., F. Ivan Nye, and Wiktor Gecas. 1978. "Husbands’ Work Involvement and marital Role Performance." Journal of Marriage and the Family 40: 9-21. Chodorow, N. 1978. The Reproduction of Mothering: Psychoanalysis and the Sociology of Gender. Berkeley: University of California. Coleman, Marion Tolbert. 1988. "The Division of Household Labor.: Suggestions for Future Empirical Consideration and Theoretical Development. " Journal of Family Issues 9(1): 132-148. Coverman, Shelley. 1983. ”Gender, Domestic Labor Time, and Wage Inequality." American Sociological Review 48(5). 623- 637. Coverman, Shelley. 1985. "Explaining Husbands Participation in Domestic Labor." The Sociological Quarterly 26: 81-97. Coverman, Shelley. 1989. "Women’s Work is Never Done: The Division of Domestic Labor." Pp. 356-368 in Jo Freeman (ed.), Women: A Feminist Perspective. Mountain View, CA: Mayfield. 249 Coverman, Shelley, and Joseph F. Sheley. 1986. "Change in Men’s Housework and Child-Care Time, 1965-1975. " Journal of Marriage and the Family 48: 413-422. Coward, Raymond T. 1987. ”Factors Associated with the Configuration of the Helping Networks of Noninstitutionalized Elders.” Journal of Gerontological Social Work 10: 113-132. Coward, Raymond T., Claydell Horne, and Jeffrey W. Dwyer. 1992. "Demographic Perspectives on Gender and Family Caregiving.” Pp. 18-33. in Jeffrey W. Dwyer and Raymond T. Coward (eds), Gender, Families, and Elder Care. Newbury Park, CA: Sage. Crichton, Jean. 1987. Age Care Sourcebook: A Resource Guide for the Aging and Their Families. New York: A Fireside Book. Cyba, Eva. 1992. "Women’s Attitudes Towards Leisure and Family." Society and Leisure 15(1): 79-94. Day, Alice T. 1985. ”Who Cares? Demographic Trends Challenge Family Care for the Elderly." Population Trends and Public Policy 9 (September): 17. DeMaris, Alfred. 1991. "A Framework for the Interpretation of First-Order Interaction in Logit Modeling.” Psychological Bulletin 110(3): 557-570. DeMaris, Alfred. 1992. Logit modeling: Practical Applications. Newbury Park, CA: Sage. Dressel, Paula L., and Ann Clark. 1990. ”A Critical Look at Family Care." Journal of Marriage and the Family 52: 769-782. Dwyer, Jeffrey W., and Raymond T. Coward. 1991. "A Multivariate Comparison of the Involvement of Adult Sons Versus Daughters in the Care of Impaired Parents.” Journal of Gerontology 46(5): S259-269. Dwyer, Jeffrey W., and Raymond T. Coward. 1992. ”Gender, Family, and Long-Term Care of the Elderly." Pp.3-17 in Jeffrey W. Dwyer and Raymond T. Coward (eds) Gender, Families, and Elder Care. Newbury Park, CA: Sage. Dwyer, Jeffrey W., and Karen Seccombe. 1991. ”Elder Care as Family Labor: The Influence of Gender and Family Position." Journal of Family Issues 12(2): 229-247. Edholm, F. Harris, 0., and K. Young. 1977. "Conceptualising Women.” Critique of Anthropology 3(9/10) 250 Elchardus, Mark, and Ignace Glorieux. 1994. "The Search for the Invisible 8 Hours: The Gendered use of Time in a Society with a High labour Force Participation of Women." Time and Society 3(1): 5-27. Engels, Friedrich. 1942. The Origin of the Family, Private Property, and the State. New York: International Publishers. (Original work published 1884.) England, P., and G. Farkas. 1986. Household, Employment, and Gender: A Social, Economic, and Demographic View. New York: Aldine. Enright, Robert B. 1991. "Time Spent Caregiving and Help Received by Spouses and Adult Children of Brain-Impaired Adults. " The Gerontologist 31(3): 375-383. Farkas, George. 1976. ”Education, Wage, Rates, and the Division of Labor Between Husband and Wife. " Journal of Marriage and the Family 38: 473-484. Ferree, Myra Marx. 1991. "The Gender Division of Labor in Two-Earner Marriages: Dimensions of Variability and Change.” Journal of Family Issues 12: 158-180. Ferree, Myra Marx. 1990. "Beyond Separate Spheres: Feminism and Family Research. " Journal of Marriage and the Family 52(4): 866-884. Finch, Janet. 1989. Family Obligations and Social Change. Southampton, England: Camelot. Finch, Janet, and Dulcie Grovers. 1983. "Introduction.” Pp. 1-10 in J. Finch and D. Grovers (eds), A Labor of Love: Women, Work, and Caring. London: Routledge and Kegan Paul. Finch, Janet, and Dulcie Groves. 1980. "Community Care and the Family: A Case for Equal Opportunities. " Journal of Social Policy 9(4): 487-511. Finley, Nancy J. 1989. "Theories of Family Labor as Applied to Gender Differences in Caregiving for Elderly Parents. " Journal of Marriage and the Family 51: 79-86 Finley, Nancy J., MD. Roberts, and BF. Banahan. 1988. ”Motivators and Inhibitors of Attitudes of Filial Obligation toward Aging Parents." The Gerontologist 28: 73-78. Fisher, Berenice, and Joan Tronto 1990. "Toward a Feminist Theory of Caring." Pp.35-62 in Emily K. Abel and Margaret K. Nelson (eds), Circles of Care: Work and Identity in Women ’s Lives. Albany: State University of New York Fishman, Pamela M. 1978. "Interaction: The Work Women Do." Social Problems 25 1 25(4): 397-406. Fitting, Melinda, Peter Rabins, M. Jane Lucas, and James Eastham. 1986. "Caregivers for Dementia Patients: A Comparison of Husbands and Wives. " The Gerontologist 26(3): 248-252. Frank, Karen A. 1985. "Social Construction of the Physical Environment: The Case of Gender." Sociological Focus 18(2): 143-157. Franklin, Susan T., Ames, Barbara D., and Sharon King. 1994. ”Acquiring the Family Eldercare Role. " Research on Aging (Special Issue: Work and Eldercare) 16(1): 27-42. Geerken, Michael, and Walter R. Gover. 1983. At Home and at Work: The Family ’5 Allocation of Labor. Newbury Park, CA: Sage. Gillespie, Dair L. 1971. ”Who Has the Power: The Marital Struggle." Journal of Marriage and the Family 33: 445-458. Gilligan, C. 1982. In a Different Voice: Psychological Theory and Women’s Development. Cambridge, MA: Harvard University Press. Given, Barbara A., Charles W. Given, Manfred Stommel, and Chen-Sheng Lin. 1994. ”Predictors of Use of Secondary Carers Used by the Elderly Following Hospital Discharge. " Journal of Aging and Health 6(3): 353-376. Glazer, Nona Y. 1990. "The Home as Workshop: Women as Amateur Nurses and Medical Care Providers. " Gender and Society 4(4): 479-499. Glendinning, Caroline. 1992. "Employment and ’Community Care’: Policies for the 19903.” Work, Employment, and Society 6(1): 103-111. Greeno, G., and Eleanor E. Maccoby. 1986. ”On In a Difference Voice: How Different is the ’Different Voice’?” Signs 11(2): 310-316. Graham, Hilary. 1983. ”Caring: A Labour of Love.” Pp.13-30 in Janet Finch and Dulcie Groves (eds), A Labor of Love: Women, Work and Caring. London: Routledge & Kegan Paul. Hardesty, Constance, and Janet Bokemeier. 1989. ”Finding Time and Making Do: Distribution of Household Labor in Nonmetropolitan Marriages. " Journal of Marriage and the Family 51: 253-267. Hardy, Melissa A. 1993. Regression with Dummy Variables. Newbury Park, CA: Sage. 252 Harris, Olivia, and Kate Young. 1981. "Engendered Structures: Some Problems in the Analysis of Reproduction." Pp. 109-147 in Joel S. Kahn and Josep R. Llobera (eds) The Anthropology of Pre-Capitalist Societies. Atlantic Highlands, New Jersey: Humanities Press. Hartmann, Heidi. 1981. "The Unhappy Marriage of Marxism and Feminism: Towards a More Progressive Union." Pp. 1-41, in Lida Sargent (ed.), Women and Revolution. Boston: South End Press. Hess, Beth B. 1985. "Aging Policies and Old Women: The Hidden Agenda." Pp.3l9-331 in Alice S. Rossi (ed.), Gender and the Life Course. New York: Aldine. Hess, Beth, and Beth Soldo. 1985. "Husband and Wife Networks." Pp. 67-92 in W.J. Sauer and R.T. Coward (eds) Social Support Networks and the Care of the Elderly. New York: Springer. Hessing, Melody. 1994. ”More Than Clockwork: Women’s Time Management in their Combined Workloads." Sociological Perspectives 37(4): 611-633. Holstein-Beck, Sabna. 1995. ”Consistency and Change in the Lifeworld of Young Women." in Johan Forras and Goras Bolin (eds), Youth Culture in late Modernity. Sage. Hooyman, Nancy. 1989. Women, Caregiving and Equity: A Feminist Perspective. Invited presentation at the annual program meeting of the Council on Social Work Education. Hooyman, Nancy. 1992. "Social Policy and Gender Inequalities in Caregiving." Pp. 181-201 in Jeffrey W. Dwyer and Raymond T. Coward (eds), Gender, Families, and Elder Care. Newbury Park, CA: Sage. Hooyman, Nancy, and W. Lustbader. 1986. Taking Care: Supporting Older People and their Families. New York: Free Press. Horowitz, Amy. 1985a. "Sons and Daughters as Caregivers to Older Parents: Differences in Role Performance and Consequences.” The Gerontologist 25(6): 612-617. Horowitz, Amy. 1985b. "Family Caregiving to the Frail Elderly." Annual Review of Gerontology and Geriatrics 5: 194-246. Horowitz, Amy. 1992. "Methodological Issues in the Study of Gender Within Family Caregiving Relationships." Pp. 132-150 in Jeffrey W. Dwyer and Raymond T. Coward (eds), Gender, Families, and Elder Care. Newbury Park, CA: Sage. 253 Horowitz, Amy, and Rose Dobrof. 1982. The Role of Families in Providing Long- term Care to the Frail and Chronically Ill Elderly Living in the Community. (Final Report, Health Care Financing Administration Grant No. 18-P-97541/20- 02). Unpublished Manuscript. Horowitz, Amy, Sherman, R.H., and SC. Durmaskin. 1983. Employment and Daughter Caregivers: A Working Partnership for Older People? " Paper presented at the 36th Annual Meeting of The Gerontologist Society of America, San Francisco, CA. Hosmer, David W., and Stanley lemeshow. 1989. Applied Logistic Regression. New York: John Wiley & Sons Huber, Joan, and Glenna Spitze. 1981. "Wives’ Employment, Household Behaviors, and Sex-Role Attitudes. " Social Forces 60: 150-169. Huber, Joan, and Glenna Spitze. 1983. Sex Stratification: Children, Housework, and Jobs. New York: Academic Press. Humphries, J ., and J. Rubery. 1984. "The Reconstruction of the Supply Side of the Labor Market: The Relative Autonomy of Social Reproduction.” Cambridge Journal of Economics 8: 331-346. Johnson, Paula. 1976. ”Women and Power: Toward a Theory of Effectiveness. " Journal of Social Issues 32: 99-110. Johnson, Colleen L. 1983. "Dyadic Family Relations and Social Support." The Gerontologist 23: 377-383. Jutras, Sylvie, and France Veilleux. 1991. "Gender Roles and Care Giving to the Elderly: An Empirical Study." Sex Roles 25(1-2): 1-18. Kamo, Yoshinori. 1988. ”Determinants of Household Division of Labor: Resources, Power, and Ideology." Journal of Family Issues 9(2): 177-200. Kaye, Lenard W., and Jeffrey S. Applegate. 1990. "Men as Elder Caregivers: Building a Research Agenda for the 1990s " Journal of Aging Studies 4(3): 289- 298. Kempen, G.I.J.M. , T.P.B.M. Suurmeijer. 1990. ”The Development of a Hierarchical Polychotomous ADL-[ADL Scale for Noninstitutionalized Elders. " The Gerontologist 30(4): 497-502. Kerber, Linda K. 1986. "On In a Difference Voice: Some Cautionary Words for Historian." Signs 11(2): 304-310. 254 Kessler, R., and J .D. McLeod. 1984. "Sex Differences in Vulnerability to Undersirable Life Events.” American Sociological Review 49(5): 620-31. Kinnear, D., and A. Graycar. 1984. ”Aging and Family Dependency.” Australian Journal of Social Issues 19: 13-25. Kramer, BL, and S. Kipnis. 1995. "Eldercare and Work-role Conflict: Toward an Understanding of Gender Differences in Caregiver Burden. " The Gerontologist 35(3): 340-348. Lang, A.M., and Brody, EM. 1983. "Characteristics of Middle-aged Daughters and help to their elderly Mothers. " Journal of Marriage and the Family 45: 193- 202. Lee, Gary R., Jeffrey W. Dwyer, and Raymond T. Coward. 1993. "Gender Differences in Parent Care: Demographic Factors and Same-Gender Preferences. " Journal of Gerontology 48: S9-16. Lemeshow, Stanley, and David W. Hosmer. 1984. ”Estimating Odds Ratios with Categorically Scaled Covariates in Multiple Logistic Regression Analysis." American Journal of Epidemiology 119(2): 147-151. Lerner, Gerda. 1990. "Reconceptualizing Differences Among Women.” Journal of Women ’s History 1(3): 106-122. Lingsom, Susan. 1989. "Filial Responsibility in the Welfare State." Journal of Applied Gerontology 8: 18-35. Litvin, Sandra J ., and Steven M. Albert, Elaine M. Brody, and Christine Hoffman. 1995 . "Marital Status, Competing Demands, and Role Priorities of Parent- Caring Daughters.” Journal of Applied Gerontology 14(4): 372-390. Litwak, Eugene and Stephen Kulis. 1987. "Technology, Proximity, and Measure of Kin Support. " Journal of Marriage and the Family 49: 649-661. Mackintosh, Garcia. 1981. "The Sexual Division of Labour and the Subordination of Women.” Pp. 1-15 in Kate Young, Carol Wolkowitz, and Roslyn McCullagh (eds) Of Marriage and the Market. London: CSE Books. Maddox, George L. 1995. The Encyclopedia of Aging. 2nd Edition. New York: Springer. Mancini, Jay, and Rosemary Blieszner 1989. "Aging Parents and Adult Children" Research Themes in lntergenerational Relations” Journal of Marriage and the Family 51: 257-290 255 Manheimer, Ronald J. 1994. Older Americans: Almanac: A Reference Work on Seniors in the Unties States. Detroit: Gale Research Inc. Matthews, Sarah H., and Tena T. Rosner 1988. "Shared Filial Responsibility: The Family as the Primary Caregiver. " Journal of Marriage and the Family 50: 185-195. McAuley, William J ., Mary D. Jacobs, and Carol S. Carr. 1984. ”Older Couples: Patterns of Assistance and Support. " Journal of Gerontological Social Work 6(4): 35-48. Mies, Maria. 1986. Patriarchy and Accumulation on a World Scale: Women in the International Division of Labour. Highlands, New Jersey: Zed Books. Miller, Baila. 1987. ”Gender and Control among Spouses of the Cognitively Impaired: A Research Note." The Gerontologist 27(4): 447-453. Miller, Baila. 1990. "Gender Differences in Spouse Management of the Caregiver Role.” pp. 92-104 in Emily K. Abel and Margaret K. Nelson (eds), Circles of Care: Work and Identity in Women ’3 Lives. Albany: State University. Miller, Baila. 1991. ”Gender Differences in Spouse Caregivers’ Strain: Socialization and Role Explanation?” Journal of Marriage and Family 52: 311-21. Miller, Baila, and Lynda Cafesso. 1992. "Gender Differences in Caregiving: Fact of Artifact?" The Gerontologist 32(4): 498-507. Miller, Baila, and Stephanie McFall. 1991. "Stability and Change in the Informal Task Support Network of Frail Older Persons. " The Gerontologist 31(6): 735-745. Miller, Cynthia L., and A. Gaye Cummins. 1992. "An Examination of Women’s Perspectives on Power. " Psychology of Women Quarterly 16: 415-428. Mitchell, Juliet. 1971. Woman’s Estate. New York: Pantheon Books. Model, Suzanne. 1981. "Housework by Husbands: Determinants and Implications." Journal of Family Issues 2(2): 225-237. Montgomery, Rhonda, J .V. 1992. ”Gender Differences in Patterns of Child-Parent Caregiving Relationships" Pp.65-83 in Jeffrey W. Dwyer and Raymond T. Coward (eds), Gender, Families, and Elder Care. Newbury Park, CA: Sage. Montgomery, Rhonda, and Mary McGlinn Datwyler. 1990. "Women and Men in the Caregiving Role." Generation 14(3): 34-38. 256 Montgomery, Rhonda, and Yoshinoro Kamo. 1989. ”Parent Care by Sons and Daughters.” pp. 213-230 in Jay A. Mancini (ed.), Aging Parents and Adult Children. Lexington Books. Morris, R.G., Woods, R.T., Davies, K.S., and L.W. Morris. 1991. ”Gender Differences in Carers of Dementia Sufferers. " British Journal of Psychiatry Supplement May (10): 69-74. N ickols, Sharon Y., and Edward J. Metzen. 1982. ”Impact of Wife’s Employment upon Husband’s Housework." Journal of Family Issues 3(2): 199-216. Noelker, L.S., and Poulshock, SW. 1982. The Effects on Families of Caring for Impaired Elderly in Residence. F inal Report Submitted to the Administration on Aging. The Margaret Blenkner Research Center for family Studies, The Benjamin Rose Institute, Cleveland, OH. Noelker, L.S., and R.W. Wallace. 1985. ”The Organization of Family Care for Impaired Elderly.” Journal of Family Issues 6: 23-44. Novak, Mark, and Carol Guest. 1989. ”Caregiver Response to Alzheimer’s Disease." International Journal of Aging and Human Development 28(1): 67-79. O’Brien, Mary. 1981. The Politics of Reproduction. Boston: Routledge & Kegan Paul. Osmond, Marie Withers, and Barrie Thorne. 1993. "Feminist Theories: The Social Construction of Gender in Families and Society.” Pp. 591-623 in Pauline G. Boss et. al (eds) Source Book of Family Theories and Methods: A Contextual Approach. New York: Plenum Press. Osterbusch, Suzanne E., Keigher, Sharon M., Miller, Baila, and Nathan L. Linsk. 1987. "Community Care Policies and Gender Justice." International Journal of Health Services 17(2) 217-232. Parsons, T. and R. Bales (eds) 1955. Family, Socialization, and Interaction Process. Glencoe, IL: Free Press. Pasero, Ursula. 1994. "Social Time Patterns, Contingency and Gender Relations." Time and Society 3(2): 179-191. Peduzzi, P.N., Hardy, R.J., and Holford, TR. 1980. ”A Stepwise Selection Procedure for Nonlinear Regression Models.” Biometrics 36: 511-516. Penning, Margaret J. 1990. "Receipt of Assistance by Elderly People: Hierarchical Selection and Task Specificity." The Gerontologist 30(2): 220-227 257 Pruchno, Rachel A. 1990. ”The Effects of Help Patterns on the Mental Health of Spouse Caregivers." Research on Aging 12: 57-71. Reece, D., Walz, T., and Hageboeck, H. 1983. lntergenerational Care Providers of Non-Institutionalized Frail Elderly: Characteristics and Consequences." Journal of Gerontological Social Work 5: 21-34. Rhode, Deborah L. 1990. "Theoretical Perspectives on Sexual Difference.” pp. 1-12 in Deborah L. Rhode (ed.), Theoretical Perspectives on Sexual Difi’erence. New Haven: Yale University Press. Rodman, Hyman. 1972. ”Marital Power and the Theory of Resources in Cultural Context. " Journal of Comparative Family Studies 3: 50-68. Rosenwaike, I. 1985. "A Demographic Portrait of the Olderest Old." Milbank Memorial Fund Quarterly 63: 187-205. Ross, Catherine E. 1987. ”The Division of Labor at Home. " Social Forces 65(3): 816-833. Rubin, Gayle. 1975. "The Traffic in Women: Notes on the ’Political Economy’ of Sex.” pp. 157-210 in Rayna R. Reiter (ed.), Toward an Anthropology of Women. New York: Monthly Review Press. Sacks, Karen. 1975. ”Engels Revisited: Women, the Organization of Production, and Private Property.” Pp. 211-234 in Rayna R. Reiter (ed.), Toward an Anthropology of Women. New York: Monthly Review Press. Scalon, W.J. 1988. A Perspective on Long-term Care for the Elderly. Health Care Financing Review (Annual Suppl): 7-15. Seccombe, Karen. 1986. "The Effects of Occupational Conditions upon the Division of Household Labor: An Application of Kohn’s Theory.” Journal of marriage and the family 48: 839-848. Seccombe, Karen. 1992. ”Employment, the Family, and Employer-Based Policies." Pp. 165-180. in Jeffrey W. Dwyer and Raymond T. Coward (eds), Gender, Families, and Elder Care. Newbury Park, CA: Sage. Seron, Carroll, and Kerry Ferris. 1995. "Negotiating Professionalism: The Gendered Social Capital of Flexible Time.” Work and Occupation 22(1): 22-47. Seymour, Julie. 1992. "’No Time to Call My Own’ Women’s Time as a Household Resource." Women’s Studies International Forum 15(2): 187-192. 258 Shanas, Ethel. 1979. ”Social Myth as Hypothesis: The Case of the Family Relations of Old People.” The Gerontologist 19: 3-9. Soldo, Beth J ., and Jana Myllyluoma. 1983. "Caregivers Who Live with Dependent Elderly. " The Gerontologist 23: 605-611. South, Scott J ., and Glenna Spitze. 1994. ”Housework in Marital and Nonmarital Households.” American Sociological Review 59: 327-347. Spector, W.D., S. Katz, J.B. Murphy, and LP. Fulton. 1987. ”The Hierarchical relationship between activities of daily living and instrumental activities of daily living. " Journal of Chronic Diseases 40: 481-489. Spitze, Glenna, and John Logan. 1990. "More Evidence on Women (and Men) in the Middle.” Research on Aging 12(2): 182-198. Stacey, Judith, and Barrie Thorne. 1985. "The Missing feminist revolution in Sociology.” Social Problems 32(4): 301-316. Stoller, Eleanor Palo. 1983. ”Parental Caregiving by Adult Children.” Journal of Marriage and the Family 45: 851-858. Stoller, Eleanor Palo. 1990. ”Males as Helpers: The Role of Sons, Relatives, and Friends." The Gerontologist 30(2): 228-235. Stoller, Eleanor Palo. 1992. "Gender Differences in the Experiences of Caregiving Spouses." Pp.49-64 in Jeffrey W. Dwyer and Raymond T. Coward (eds) Gender, Families, and Elder Care. Newbury Park, CA: Sage. Stoller, Eleanor Palo, and Stephen J. Cutler 1992. "The Impact of Gender on Configurations of Care Among Married Elderly Couples. " Research on Aging 14(3): 313-330. Stoller, Eleanor Palo, and Karen L. Pugliesi. 1991. ”Size and Effectiveness of Informal Helping Networks: A Panel Study of Older People in the Community. " Journal of Health and Social Behavior 32(2): 180-191. Stoller, Eleanor Palo, and Lorna L. Earl. 1983. "Help with Activities of Everyday Life: Sources of Support for the Noninstitutionalized Elderly." The Gerontologist 23(1): 64-70. Stommel, Manfred, Collins, Clare E., and Barbara A. Given. 1994. "The Cost of Family Contributions to the Care of Persons with Dementia." The Gerontologist 34(2): 199-205. 259 Stommel, Manfred, and Margaret Kingry. 1991. ”Support Patterns for Spouse- Caregivers of Cancer Patients: The Effect of the Presence of Minor Children. " Cancer Nursing 14: 200-205. Stommel, Manfred, Given, Barbara A., Given, Charles, W., and Clare E. Collins. 1995. "The Impact of the Frequency of Care Activities on the Division of Labor between Primary Caregivers and Other Care Providers." Research on Aging 17(4): 412-433. Stone, Robyn, Gail Lee Cafferata, and Judith Sangl. 1987. ”Caregivers of the Frail Elderly: A National Profile.” The Gerontologist 27(5): 616-626. Straw, Pat, and Brian Elliot. 1986. "Hidden Rhythms: Hidden Powers: Women and Time in Working-Class Culture." Life Stories 2: 34-47. Szinovacz, M. 1989. ”Retirement Couples and Household Work." Pp. 33-58, in S.J. Bahr and E. Peterson (eds), Aging and the Family. Lexington, MA: Lexington. Tennstedt, S., J. McKinlay, and L. Sullivan. 1989. ”Informal Care for Frail Elders: The Role of Secondary Caregivers.” The Gerontologist 29: 677-683. Thompson, ER 1967. "Time, Work-discipline and Industrial Capitalism.” Past and Present 38: 55-97. Thorne, Barrie. 1982. "Feminist Rethinking of the Family: An Overview." pp. 1-24 in Barrie Thorne and Marilyn Yalom (eds), Rethinking the Family: Some Feminist Questions. New York: Longman. Thorne, Barrie. 1990. ”Children and Gender: Constructions of Difference.” in pp. 100-113 in Deborah L. Rhode (ed.), Theoretical Perspectives on Sexual Difierence. New Haven: Yale University Press. U.S. Bureau of the Census. 1990 Census of Population: General Population Characteristics, Part 1, United States. U.S. Department of Commerce, Economics and Statistics Administration. U.S. Bureau of Census. 1993. Poverty in the United States: 1992. Current Population Reports, Series P60-185. Washington, DC: U.S. Government Printing Office. U.S. Senate, Special Committee on Aging. 1987. Developments in Aging, 1986. Washington, DC: Government Printing Office. Ungerson, Clare. 1983. ”Why Do Women Care?” Pp.31-49. in Janet Finch and Dulcie Groves (eds), A Labor of Love: Women, Work and Caring. London: Routledge & Kegan Paul. 260 Vance, Carole S. 1980. ”Gender Systems, Ideology, and Sex Research: An Anthropological Analysis." Feminist Studies 6(1): 129-141. Vanek, Joann. 1974. "Time Spent in Housework." Scientific American 231: 116-120. Walker, Alan. 1991. "The Relationship Between the Family and the State in the Care of Older People." Canadian Journal on Aging 10(2): 94-112. Walker, Alan. 1983. "Care for Elderly People: A Conflict between Women and the State." Pp. 106-128 in Janet Finch and Dulcie Groves (eds), A Labor of Love: Women, Work and Caring. London: Routledge & Kegan Paul. Warner, Rebecca L. 1986. "Alternative Strategies for Measuring Household Division of Labor. " Journal of Family Issues 7(2): 179-195. West, Candace, and Don H. Zimmerman. 1991. ”Doing Gender." pp. 13-37 in Judith Lorber and Susan A. Farrell (eds), The Social Construction of Gender. Newbury Park, CA: Sage. Wilcox, Julie A., and Merlin A. Taber. 1991. ”Informal Helpers of Elderly Home Care Clients." Health and Social Work 16(4): 258-265. Zarit, Steven H., Pamela A. Todd, and Judy M. Zarit. 1986. "Subjective Burden of Husbands and Wives as Caregivers: A Longitudinal Study." The Gerontologist 26(3): 260-266. nICHran STATE UNIV. LIBRRRIES ll1llWIWIIIHIIWHNI1|”HWIWIHHIWMHI 31293015706389