Gender in Cohabitation and Marriage The Influence of Gender Ideology on Housework Allocation Over the Life Course MICK CUNNINGHAM Western Washington University The analysis examines the direct and indirect influences of early gender socialization on the allocation of routine housework later in the life course. The study articulates hypotheses suggesting that the relationship between gender socialization early in adulthood and housework allocation later in adulthood is moderated by gender and union type and is mediated by subsequent contextual characteristics of the couple. The analysis draws on panel data from a sample of 586 young adults spanning 31 years. Findings indicate that married men's attitudes about gender early in adulthood are more influential for the later division of labor than are married women's attitudes, but gender differences in the influence of early attitudes on later housework patterns are not present among cohabitors. The influence of early gender socialization on later housework allocation is mediated by couple-level resources and time availability among cohabitors but not among married individuals. Keywords: gender; housework; life course; socialization Over the past several decades, family scholars have identified three central theoretical mechanisms that consistently explain variation in allocation of gendered work in families. These mechanisms include influences associated with gender socialization, the time couple members have available for family work, and the relative human capital resources of the couple members. Despite these significant theoretical and empirical advances, our understanding of the causal interrelationships among these factors remains limited, in part as a result of a dearth of longitudinal studies of this topic (Coltrane, 2000). The current study draws on a life course perspective and on longitudinal data in an effort to integrate our understanding of the processes through which family work is allocated. Author's Note: / would like to thank Arland Thornton, Terri Orbuch, and Debra Carrfor their helpful comments on previous drafts of this article. An earlier version of this article was presented at the annual meetings of the Population Association of America in Los Angeles in March 2000. JOURNAL OF FAMILY ISSUES, Vol. 26 No. 8, November 2005 1037-1061 DOI: 10.1177/0192513X04273592 © 2005 Sage Publications 1037 1038 JOURNAL OF FAMILY ISSUES / November 2005 Another important focus of research on the gendered division of family labor draws on a symbolic interactionist approach to argue that the gendered division of household labor serves as an important site for the routine production of gender within the context of romantic, heterosexual relationships (Berk, 1985; South & Spitze, 1994). According to West and Zimmerman's (1987) original formulation, the performance of housework constitutes a "situated doing, carried out in the virtual or real presence of others who are presumed to be oriented to its production" (p. 126). Furthermore, these authors argue that "for a woman to engage in it and a man not to engage in it is to draw on and exhibit the 'essential nature' of each" (p. 144). However, as Greenstein (1996) points out, it is more accurate to conceive of individuals' gender ideologies as a key variable that differentially shapes individuals' views of what it means to do gender in the context of a relationship than as a cultural constant that exerts a similar influence in all relationships. Furthermore, if the performance of housework is dependent on context (a situated doing), it seems likely that accounting for the type or context of a union may be particularly important for understanding how gender is displayed in particular relationships. A recent study by Brines and Joyner (1999) provides an important potential contribution to symbolic interactionist theories of gender by positing that the factors associated with union duration differ between cohabiting and married couples. They argue that "cohabitors . . . tend to embrace individualism, as well as ideals of personal autonomy and equity in each partner's contribution to the household" (p. 333). They find support for the idea that a specialized division of labor is positively associated with marital duration but that a division of paid labor characterized by equality is positively associated with cohabitation duration. I extend this line of reasoning by examining the extent to which the factors influencing the gendered division of family labor differ according to the context of a union—whether it is a cohabitation or marriage. In the current analysis, I formulate a dynamic theoretical model hypothesizing that gender socialization early in the life course exerts a direct influence on housework allocation. Furthermore, I argue that the influence of gender socialization operates indirectly by shaping individuals' subsequent levels of time availability, relative couple resources, and gender egalitarianism. In addition, I hypothesize that gendered power dynamics in couples will condition the process through which gender ideology influences the allocation of family work. This moderating hypothesis specifically implies that the influence of gender socialization on the allocation of routine housework will be roughly equivalent for women and men Cunningham / GENDER IN COHABITATION AND MARRIAGE 1039 in cohabiting couples, among whom gendered power differences may be viewed as less desirable. Conversely, I expect the influence of early gender socialization on later housework allocation to be stronger for men than for women in the context of marriage, where gendered family arrangements are more highly institutionalized and thus are more likely to be based on a model of specialization rather than one of equality (Nock, 1998). In other words, to the extent that gendered power relations are more deeply embedded in marriage than in cohabitation, I argue that men's attitudes about gender will trump women's attitudes with regard to the performance of routine housework in the context of marriage but that women's and men's gender ideologies will exert similar influences on housework allocation in the context of cohabitation. Drawing on data from a 31 -year panel study of parents and children, the analysis begins by identifying the extent to which attitudes about gender early in adulthood exert a lasting influence on housework allocation many years in the future. I improve upon previous research not only by controlling for potentially important parental characteristics but also by examining the influence of gender ideology measured early in adulthood on the allocation of routine housework many years later. I assess the direct effects of early measures of gender socialization on the gendered division of routine housework as well as the extent to which the influence of early attitudes operates by influencing later attitudes about gender, the relative human capital resources of a couple, and the time availability of couple members. I assess the extent to which the processes through which gender socialization is linked to housework allocation differ by union type by separately analyzing married and cohabiting couples. I examine gender differences in the influence of gender ideology on housework allocation for each type of union. This strategy allows me to test the hypothesis that men possess a greater ability than women to convert their attitudes into actual household practices within the context of marriage but not within the context of cohabitation. The analysis begins with the articulation of a number of causal hypotheses, then it describes the data and measures with which the hypotheses are tested, and finally it carries out a series of multivariate regression models designed to test these hypotheses. SOCIALIZATION AND HOUSEWORK ALLOCATION Previous research on housework allocation has drawn on a range of methods and indicators to assess gendered patterns of family work. The 1040 JOURNAL OF FAMILY ISSUES / November 2005 current analysis follows a number of previous studies that focus on the proportional allocation of stereotypically female housework between men and women (Atkinson & Huston, 1984; Barnett & Baruch, 1987; Deutsch, Lussier, & Servis, 1993; Goldscheider & Waite, 1991; Starrels, 1994). The kinds of tasks most frequently performed by women are of particular interest because previous research suggests that these tasks are the most time intensive, are perceived as the least enjoyable, and are the most likely to require relatively rigid time schedules for their completion (Barnett & Shen, 1997; Coleman, 1988; Greenstein, 1996). Most empirical studies of the division of household work in general and the allocation of stereotypically female tasks in particular have included at least one indicator of attitudes about appropriate family roles for women and men (Berk, 1985; Coltrane & Ishii-Kuntz, 1992; Greenstein, 1996; Kamo, 1988). These measures are commonly labeled socialization or gender ideology and are interpreted as the degree to which individuals support or reject a role-specialized model of the family in which men are responsible for the family's financial support and women are responsible for housework and child care. Although the size of the influence of attitudes about appropriate roles for women and men varies across studies, there is a growing consensus that egalitarian attitudes about gender are positively related to patterns of participation in routine housework, especially when housework is assessed in relative rather than absolute terms (Coltrane, 2000). Empirical work in this vein suggests that couple members who are generally supportive of egalitarian roles for women and men are more likely to share routine housework. According to this line of reasoning, individuals with egalitarian ideals about gender display these preferences by minimizing the extent of gender-based specialization in domestic work (Greenstein, 1996). Despite the frequency with which the causal effects of attitudes on housework have been studied, there has been a general failure to adequately demonstrate the direction of the causal relationship. It has been common for researchers to assess attitudes about gender and housework performance simultaneously and to make the strong assumption that attitudes are causally antecedent to behavior (Greenstein, 1996; Hardesty & Bokemeier, 1989; Kamo, 1988; cf. Orbuch & Eyster, 1997). The current analysis improves upon existing cross-sectional research by capturing indicators of gender socialization at the onset of adulthood prior to both the formation of a union and the assessment of the division of family work. The expected relationship between gender ideology and the gendered division of family work over time serves as the basis for my first hypothesis: Cunningham / GENDER IN COHABITATION AND MARRIAGE 1041 Hypothesis 1: Egalitarian attitudes about gender early in the life course will be positively associated with a subsequent division of family labor in which men perform a relatively greater share of routine housework. It is also likely that individuals' attitudes about gender change over time in response to experiences across the life course (Waite, Goldscheider, & Witsberger, 1986). Therefore, individuals' attitudes about gender at the time housework is measured are expected to be associated with patterns of family work. Furthermore, it is likely that the influence of early attitudes on housework allocation is transmitted by contemporaneously measured attitudes about gender. This line of reasoning leads to the second hypothesis: Hypothesis 2: Egalitarian attitudes about gender measured contemporaneously with measures of housework allocation will be positively associated with a division of family labor in which men perform a relatively greater share of routine housework and will mediate the influence of earlier attitudes. It is also important to consider the possibility that factors other than levels of gender egalitarianism mediate the influence of early gender socialization on later housework allocation. On one hand, attitudes about gender at the onset of adulthood may influence later housework allocation patterns primarily through their influence on later attitudes about gender. In this case, gender ideology would exert an independent influence on housework allocation patterns regardless of other individual or couple-level characteristics. Alternately, a more dynamic process may be involved. In this case, attitudes about gender early in the life course might be associated with individual and couple-level characteristics that have been previously linked to housework allocation. For instance, those with egalitarian attitudes at the onset of adulthood may be more likely to choose a spouse with similar career goals, thus increasing the likelihood that their eventual earnings-based resources are similar. The current analyses investigate the extent to which the influence of attitudes about gender on routine housework is mediated by relative couple resources and each couple member's time availability. A number of researchers have argued that increased resources relative to a spouse allows for greater bargaining power with regard to housework (Brayfield, 1992; Kamo, 1988; Presser, 1994). According to this exchange-based approach, a spouse with greater resources has a greater ability to avoid housework. Most often such studies have shown that 1042 JOURNAL OF FAMILY ISSUES / November 2005 women who have greater earnings or other human capital relative to their husbands have greater equality in the division of housework than do women who are more dependent on their spouses for income (Blair & Lichter, 1991; Coverman, 1985; Kamo, 1988, 1991; Shelton & John, 1993). Attitudes about gender may influence relative resources both by shaping the accumulation of resources for one member of the couple and by leading to the initiation of relationships in which resources are distributed more or less equally between members. If attitudes about gender influence housework allocation indirectly, a couple's relative resources are expected to mediate the influence of early attitudes on housework allocation. Researchers have also demonstrated that spouses or partners who have more unconstrained time perform a greater share of the routine housework (Kamo, 1988; Robinson & Godbey, 1997). The primary sources of time constraints are paid employment and children. The more hours an individual spends in the labor force, the less time they are expected to devote to housework. The presence of children increases both the degree to which housework is segregated by gender and the amount of housework performed by women relative to men (Berk, 1985; Coltrane & Ishii-Kuntz, 1992). Most researchers have found that parenting has a larger effect on women's housework time than on men's (Gershuny & Robinson, 1988; Shelton, 1992), although in general young children increase the amount of time that both women and men devote to housework (Presser, 1994; Rexroat & Shehan, 1987; South & Spitze, 1994). Women with young children may exit the labor force or reduce their work hours, leaving them with primary responsibility for housework. Conversely, the birth of a child often results in an increase in men's hours of paid work (Berk, 1985). Furthermore, research suggests that egalitarian attitudes about gender increase women's commitment to the labor force (Cunningham, Beutel, & Thornton, 2000) and decrease men's and women's childbearing (Morgan & Waite, 1987; Waite, Haggstrom, & Kanouse, 1985). If the influence of gender ideology on housework allocation is mediated by either couple-level resources or time availability, support for the following hypothesis should be found: Hypothesis 3: The positive influence of egalitarian attitudes about gender on a division of family labor in which men perform a relatively greater share of routine housework will be mediated by measures of a couple's relative resources and time availability. Cunningham / GENDER IN COHABITATION AND MARRIAGE 1043 Although some recent research has reported that men's and women's attitudes about gender are associated with patterns of housework allocation, several studies suggest that men's attitudes about gender are more highly correlated with a couple's division of housework than are women's (Blair & Lichter, 1991; Kamo, 1988). However, few studies have explicitly investigated gender differences in the influence of attitudes on the division of family work. One exception is Greenstein (1996), who demonstrates that housework allocation patterns are contingent on men's attitudes. Specifically, Greenstein finds that routine housework is likely to be shared by women and men only when men and women each hold relatively egalitarian attitudes about gender. This finding suggests that men and women may have a differential ability to put their attitudes about appropriate family roles into practice. Stated differently, although women's egalitarian attitudes constitute a sufficient condition for the sharing of domestic labor, men's attitudes constitute a necessary condition for such sharing. Ethnographic research by Hochschild (1989) also suggests that men frequently have greater power than women to establish the kinds of gendered family arrangements that they prefer. Men who want to share may choose to do so, but they have the power to do so or not regardless of the woman's ideological preferences. To the extent that the division of family work is reflective of gendered power relations in families, men's attitudes may be more likely to influence housework allocation. This body of research suggests the following interactive hypothesis: Hypothesis 4: The positive association between egalitarian attitudes about gender and a division of family labor in which men perform a relatively greater share of routine housework is likely to be stronger for men than for women. Each of the preceding hypotheses assumes that the processes through which gender ideology is linked to the division of family work are similar across union types. However, a number of researchers have suggested that the meaning of gender may vary for those who are married compared to those who are cohabiting (Blumstein & Schwartz, 1983; Brines & Joyner, 1999). Those with egalitarian attitudes about gender may be more likely to select cohabitation over marriage for their first union (Clarkberg, Stolzenberg, & Waite, 1995), and the experience of cohabitation is associated with increased levels of gender egalitarianism (Moors, 2003). Although several researchers have reported similarity in the levels of men's participation in stereotypically female housework across union 1044 JOURNAL OF FAMILY ISSUES / November 2005 types (Gupta, 1999; Stafford, Backman, & DiBona, 1977), Shelton and John (1993) report that married women spend more time performing household labor than do cohabiting women. Regardless of the levels of housework performance, however, it is possible that the factors influencing the allocation of housework between women and men differ for married and cohabiting couples. Although there is little research on differences in the determinants of housework allocation across marital statuses, the existing research suggests that gender may be a less salient factor in establishing patterns of domestic work in cohabiting couples than it is in married couples. To the extent that gendered power relations are minimized in cohabiting unions relative to marriages, the following hypothesis is suggested: Hypothesis 5: The positive association between egalitarian attitudes about gender and a division of family labor in which men perform a relatively greater share of routine housework is not expected to vary by gender among cohabiting couples. DATA AND MEASURES THE SAMPLE The analyses utilize data from the Intergenerational Panel Study of Parents and Children. Respondents were born in 1961 and are the daughters and sons of a larger panel study that was initiated in 1962. The mothers of the respondents were initially selected from a probability sample of July 1961 birth records of first-, second-, and fourth-born White children in the Detroit metropolitan area. Each of the children upon whom the initial sample was based were interviewed in 1980, when they were 18 years of age. They were reinterviewed in 1985 and 1993 at ages 23 and 31, respectively. Due to the nature of questions on the division of housework, the subsample used in this analysis is limited to individuals who were married or cohabiting at the time of the 1993 interview. The sample is composed of 282 men and 305 women for a total of 587 individuals. These data are valuable for the long time span they cover and the presence of detailed information about a wide range of family-related variables. It is important to note, however, that the sample is racially homogenous and was initially based in the Detroit area, so the conclusions of this research are not immediately generalizable to nationally representative populations. Nonetheless, the long time series available in the data justi- Cunningham / GENDER IN COHABITATION AND MARRIAGE 1045 fies the current investigation with these data given the expectation that more broadly representative panel data that could be utilized to corroborate these findings may eventually become available. MEASURES Housework allocation. To measure the allocation of housework, each cohabiting or married respondent was first read the following statement: "Now I would like to know how you and your (husband, wife, or partner) divide up some jobs around the house." This statement was followed by a set of questions assessing the division of particular tasks in the couple. For instance, one question read: "First, which category best describes how grocery shopping is divided up in your family?" Response categories included "spouse or partner usually," "spouse or partner a little more often," "equal" (if volunteered), "you a little more often," and "you usually." Similar questions were asked about doing the evening dishes and straightening up before company comes. These items were averaged into an index and coded so that a high score represents greater relative participation by men in these stereotypically female household tasks regardless of whether a man or a woman responded to the items. This measurement strategy replicates the approach used in previous studies that assess the proportional allocation of particular types of housework (Blood & Wolfe, 1960; Ferree, 1991; Huber & Spitze, 1983; see Shelton & John, 1996). Gender ideology. The indicator of gender ideology is composed of a set of eight questions representing respondents' attitudes about the appropriate roles of women and men in the family context (see the appendix for text). These measures were assessed at ages 18 and 31 and appear to measure a single underlying construct (Cronbach's a = .72 and .75, respectively). They are coded so that a high score represents support for an egalitarian division of family roles and are averaged into an index. Marital status. Respondent's marital status is assessed based on reports of whether the union is marital or nonmarital. Relative resources. Relative income is measured as a ratio of the man's 1992 income to the total income earned by both partners. This item ranges from a low of 0 to a high of 100, and a high score means that a greater proportion of the couple's income is earned by the man. 1046 JOURNAL OF FAMILY ISSUES / November 2005 Time availability. The employment hours of the respondent were reported for the previous 4 weeks, and an average of employment hours per week was created. Employment hours for the spouse were reported by the respondent based on a single question about the spouse's average hours of weekly employment. The presence of children is measured with three variables. The first assesses whether there are any children under age 6 in the household, the second measures whether there are any children between ages 6 and 18 in the household, and the third sums the total number of children living in the respondent's household at age 31. Approximately 33% of the men and 27% of the women had no children living with them at time of the interview. Parental characteristics. The analyses include controls for years of schooling completed by the mother and father at the time the child was born and for the mother's gender ideology when the child was 15. The mother's attitudes about gender are measured and coded in the same way as the items measuring the children's attitudes. METHOD The analysis begins by analyzing descriptive statistics and correlations among the variables. Next, a set of hierarchical ordinary least squares (OLS) regression models is constructed. The models estimate the effects of age-18 and age-31 attitudes about gender on respondents' housework allocation at age 31 among married and cohabiting couples, and the models control for the parental characteristics outlined above. Because several of the equations involve interaction terms, measures of attitudes about gender are standardized with a mean of 0 and a standard deviation of 1 (Aiken & West, 1991). After examining the direct effects of early attitudes, the models introduce measures of age-31 relative resources and time availability to ascertain whether the influence of gender ideology operates directly or indirectly. RESULTS Table 1 provides means and standard deviations for the key analysis variables for four separate groups: married women, married men, cohabiting women, and cohabiting men. Table 1 makes it possible to compare TABLE 1 Variable Descriptions, Means, and Standard Deviations by Gender and Union Type Married Cohabiting Variable Women (n = 271) Men (n = 260) Women (n = 34) Men (n = 22) M SD M SD M SD M SD Housework allocation: Age 31 Housework allocation 1.93a 0.82 2.22 0.90 2.03 0.87 2.33 0.99 Gender ideology Attitudes about gender index: Age 18 3.59a 0.67 3.33 0.57 3.62 0.86 3.33 0.72 Attitudes about gender index: Age 31 3.98a 0.63 3.75 0.53 3.97 0.59 3.93 0.60 Relative resources: Age 31 Spouse's relative income (man's + total) 0.71 0.25 0.70 0.22 0.63 0.27 0.67 0.20 Time availability: Age 31 Man's weekly employment hours 45.54c 14.16 45.16 15.37 35.91b 19.29 44.82 10.71 Woman's weekly employment hours 22.33ac 19.60 25.99 18.63 31.71 18.56 28.91 19.54 Any children under 6 in household 0.75ac 0.43 0.67 0.47 0.35bd 0.49 0.13 0.35 Any children 6 to 18 in household 0.30a 0.46 0.23 0.42 0.29d 0.46 0.05 0.21 Children in household 1.44ac 1.05 1.27 1.07 0.71d 1.12 0.27 0.63 Control variables Mother's education: Child age 1 12.14 1.83 12.23 1.74 11.85 1.97 11.91 1.80 Father's education: Child age 1 12.61 2.38 12.26 2.58 12.26 2.38 12.55 2.30 Mother's attitudes about gender index: Child age 15 3.29 0.67 3.33 0.63 3.38 0.64 3.49 0.60 a. Married women and married men were significantly different. b. Cohabiting men and cohabiting women were significantly different. c. Married women and cohabiting women were significantly different. d. Married men and cohabiting men were significantly different. 1048 JOURNAL OF FAMILY ISSUES / November 2005 not only gender differences in the values of the variables for each marital status but also to compare marital status differences in the values of the variables for each gender. In terms of gender differences within each marital status, men in each group report higher levels of relative participation in stereotypically female housework than women report is done by their male partners. However, this gender difference is not significant among cohabiting partners. In contrast to married couples, cohabiting men report working significantly more hours than are worked by the male partners of cohabiting women. However, women have more children than men in both married and cohabiting couples. It is important to keep in mind, however, that although the men and women in the sample are all the same age, men marry at older ages than women. Therefore, it is likely that gender differences in age at marriage are associated with these gender differences, including those observed in patterns of housework allocation. Finally, women's attitudes about gender are slightly more egalitarian than are men's, although again the difference is only significant for married individuals. When we compare respondents of a similar gender across marital statuses, we see that cohabiting women are employed nearly 10 hours more per week, on average, than are married women. Married respondents have more children living in their households at the time of the interview than do cohabiting respondents, and this is especially true for men. Mean levels of the man's share of the total family income are lower among cohabitors than among married respondents, but this difference is not statistically significant. At the low end, men earn 63% of the couple's income among cohabiting women. At the high end, men earn 71 % of the couple's income among married women. Table 2 presents the zero-order correlations of the predictor variables with the measure of men's relative participation in stereotypically female housework for each of the four groups identified in Table 1. There are several notable results. First, gender ideology at age 18 is positively associated with men's participation in routine housework at age 31 for three of the four groups, providing initial support for Hypothesis 1. Although the correlations are not uniformly large, they are nonetheless nontrivial given the 13-year time span between measurement of the attitudinal and behavioral measures. The only group for whom there is not a positive association between early attitudes and later housework performance is married women. Furthermore, the strong positive influence of cohabiting women's early attitudes on later housework allocation patterns is the only statistically significant coefficient for cohabitors. In addition, with the Cunningham / GENDER IN COHABITATION AND MARRIAGE 1049 TABLE 2 Correlations of Age 31 Household Labor Allocation With Independent Variables by Gender and Union Type Married Cohabiting Women Men Women Men (a = 271) (n = 260) (a = 34) (n = 22) Gender ideology Attitudes about gender index: Age 18 -.01 .12* 43** .09 Attitudes about gender index: Age 31 .11* 3^*** .22 .12 Relative resources: Age 31 Spouse's relative income (man's total) _ i9** — 28*** -.32 -.23 Time availability: Age 31 Man's weekly employment hours -.14* -.09 .13 .05 Woman's weekly employment hours .15* 31*** .22 .11 Any children under 6 in household .02 _ i7** -.19 -.14 Any children 6 to 18 in household .00 -.13* -.25 -.30 Number of children in household .04 -.11 -.24 -.25 Control variables Mother's education: Child age 1 .15* .10 -.14 .22 Father's education: Child age 1 .12* .04 .00 .09 Mother's attitudes about gender index: Child age 15 .08 .16** .31 .16 NOTE: Tests are two-tailed. Tests for gender ideology measures are one-tailed. *p < .05. **p < .01. ***/> < .001. exception of cohabiting women, attitudes about gender measured contemporaneously with housework allocation are more highly correlated with the division of labor than are attitudes measured at age 18. Taken as a whole, these correlational results provide original evidence of the importance of gender ideology for housework allocation over long periods of time. Table 2 also demonstrates that the more a married man makes in relation to his spouse or partner the less likely he is to perform a large proportion of the stereotypically female tasks. Similarly, married women's employment hours are positively associated with men's relative levels of participation in these kinds of tasks. It is important to note that the magnitude of the coefficients for many of the variables are similar among married and cohabiting couples, and in the cases of the measures of the presence and ages of children, they are substantially larger. However, the small sample size for cohabitors increases the standard errors associated with 1050 JOURNAL OF FAMILY ISSUES / November 2005 these coefficients and therefore reduces the likelihood of rejecting the null hypotheses of no association. In Table 3, a series of multivariate models that are designed to confirm the relationships observed in the bivariate context and to examine the processes through which the key variables operate are presented. Table 3 provides unstandardized OLS regression coefficients estimating the influence of age-18 and age-31 attitudes about gender on men's relative participation in stereotypically female household tasks at age 31 for married and cohabiting respondents. Given the directional hypotheses identified above regarding the influence of gender ideology on housework allocation, one-tailed tests are used to establish statistical significance for coefficients associated with measures of gender ideology and interactions of gender and gender ideology. To reduce the likelihood that the observed relationships are spurious, each of the models in Table 3 controls for the mother's and father's education and the mother's attitudes about gender. Furthermore, to clearly establish the causal ordering of gender ideology, union type, and housework, respondents who married (Models 1-5) or cohabited (Models 6-10) prior to age 18 are excluded from the analyses. I begin by discussing the results for married respondents, and then I discuss findings for cohabiting respondents. The first model in Table 3 examines the influence of attitudes about gender at age 18 on men's relative participation in stereotypically female household tasks, controlling for parental characteristics and the respondent's gender. Model 1 suggests that attitudes about gender at the onset of adulthood are not related to greater relative participation by men in stereotypically female housework. Model 2 includes a term capturing the interaction of the respondent's gender and attitudes about gender. The statistically significant interaction term demonstrates that the influence of early attitudes on men's relative participation in stereotypically female housework differs by gender. Model 2 shows that a standard deviation increase in gender egalitarianism is associated with a .09 increase in men's relative participation in routine housework, whereas the comparable effect for women is-.06 (-.15-.09). Although the main effect for men falls just shy of statistical significance at conventional levels, it suggests that married men's attitudes about gender are more highly linked to later patterns of housework allocation than are women's. Furthermore, it is important to keep in mind that the measurement of the independent and dependent variable is separated by 13 years. There are a number of implications of the coefficients in Model 2. First, the results offer support for Hypothesis 4 in the marital context. Hypothesis 4 predicted that the influence of attitudes about gender on the TABLE 3 Unstandardized Ordinary Least Squares Regression Coefficients From the Regression of Household Labor Allocation at Age 31 on Gender Ideology at Ages 18 and 31 and Other Predictors: Married and Cohabiting Respondents Married (a = 531) Cohabiting (n = 56) Independent Variable Model 1 2 3 4 5 6 7 8 9 10 Gender ideology Attitudes about gender index: Age 18 .01 .09 .09 .06 -.04 .20* .09 .16 .17 .14 Gender * attitude about gender index: Age 18 -.14 * —15** .16 Attitudes about gender index: Age 31 31*** .26*** .07 .06 Gender * attitude about gender index: Age 31 - 23** _ 24*** Relative resources Spouse's relative income (man's + total) -.38 -.39* -.88 -.87 Time availability Man's weekly employment hours .00 .00 .01 .01 Woman's weekly employment hours 0i*** .01* .00 .00 Any children under 6 -.07 -.10 .20 .19 Any children 6 to 18 -.13 -.12 -.53 -.54 Number of children in household .11 .10 .06 .08 Control variables Gender (1 = female) -.31 _27*** _ 35*** _ 32*** -.37 -.34 .35 -.36 -.27 R2 .05 .06 .13 .10 .16 .16 .17 .21 .16 .21 2 Adjusted R .04 .05 .11 .09 .13 .07 .07 .01 .06 -.01 NOTE: Estimates control for mother's education, father's education, and mother's attitudes about gender. Tests are two-tailed. Tests for coefficients of gender ideology and interaction terms are one-tailed. *p < .05. **p < .01. ***p < .001. 1052 JOURNAL OF FAMILY ISSUES / November 2005 division of family work would differ by gender. Second, the findings are notable because they suggest that existing single-generation studies of the predictors of housework allocation that report an association between gender ideology and housework may be based on findings that are at least partially due to the joint influence of parental characteristics on adult children's gender ideologies and patterns of housework allocation. Finally, because of the relatively small number of tasks included in the data, as well as the fact that the included tasks are among the most frequently shared by men (especially grocery shopping and washing dishes), it is likely that this coefficient underestimates the magnitude of the relationship between attitudes about gender and the full range of tasks likely to be carried out in most households. Similarly, it is likely that the relatively low explained variance in Model 2 is in part a result of the smaller-than-ideal set of tasks that are used to measure housework allocation. Having demonstrated partial support for Hypotheses 1 and 4, the next step is to identify the processes through which married men's early gender socialization influences subsequent housework allocation patterns. Model 3 provides a test of Hypothesis 3, which suggests that couple members' relative income and time availability mediate the influence of early gender socialization on the division of labor later in the life course. Model 4 provides a test of Hypothesis 2, which suggests that early gender socialization operates through its influence on later levels of gender egalitarianism. The equation represented by Model 3 of Table 3 includes measures of the parental control variables, gender, gender ideology, the interaction of gender and gender ideology, and measures of couple members' relative resources and time availability. The coefficients in Model 3 demonstrate that married women's employment hours are positively associated with men's relative participation in stereotypically female housework. More importantly, the magnitude of the influence of married men's early gender egalitarianism on age-31 housework allocation remains unchanged when the couple characteristics are added to the model. This suggests that relative couple resources and time availability do not mediate the influence of married men's attitudes on their participation in routine housework. Hypothesis 3 predicted that the influence of attitudes about gender on housework allocation would be transmitted by a couple's relative resources and time availability, but Model 3 offers little support for this hypothesis among married respondents. This finding demonstrates that the influence of married men's early gender socialization does not operate by shaping the relative income or time availability of couple members later in the life course. Rather, early measures of gender ide- Cunningham / GENDER IN COHABITATION AND MARRIAGE 1053 ology exert an independent influence on subsequent levels of housework allocation. Hypothesis 2 suggests that indicators of gender ideology are associated with housework allocation when the two variables are measured contemporaneously and that later attitudes about gender would mediate the influence of early attitudes. Model 4 of Table 3 includes measures of gender ideology at ages 18 and 31 along with a term for the interaction of gender and age-31 attitudes. This model provides support for Hypothesis 2, demonstrating that when gender ideology and housework allocation are measured at the same time, attitudes about gender are strongly associated with men's relative participation in stereotypically female housework. Specifically, a standard deviation increase in married men's egalitarian-ism is associated with a predicted increase of .29 units in men's relative participation in stereotypically female housework. As in Model 2, the interaction of gender and age-31 gender ideology is again highly significant, providing additional support for Hypothesis 4. Furthermore, the influence of early gender ideology is substantially reduced, providing evidence that married men's early gender egalitarianism influences housework allocation through its effect on men's later attitudes. Model 5 includes each of the predictors in a single equation. This model offers a second test of Hypothesis 3, which suggests that gender ideology operates by influencing couple characteristics. In support of Hypothesis 3, Model 5 provides some evidence that the influence of married men's gender ideology on their participation in routine housework operates indirectly. The coefficient for men's gender ideology is reduced by roughly one sixth, from .31 to .26, when measures of relative resources and time availability are included in the model with age-31 attitudes. Although it is possible to infer support for Hypothesis 3 from Model 6 of Table 3, it is important to keep in mind that in this model gender ideology and couple characteristics are captured contemporaneously. Therefore, attitudes about gender at age 31 may be reflective of the variables measuring relative resources and time availability. Having assessed Hypotheses 1 through 4 among married individuals, I now turn to an analysis of cohabiting couples, which is presented in Models 6 through 10 of Table 3. Hypothesis 5 predicted that the relationship between gender ideology and housework allocation patterns would not be conditional on the gender of the respondent among cohabiting couples. To test this hypothesis, a series of multivariate models similar to those in Models 1 through 5 of Table 3 were constructed for cohabiting respondents. Although the sample size for cohabiting respondents is small, Model 6 demonstrates that among cohabitors egalitarian attitudes about 1054 JOURNAL OF FAMILY ISSUES / November 2005 gender at age 18 are positively related to men's relative participation in stereotypically female housework 13 years later. Model 7, which includes the interaction of gender and gender ideology, provides support for Hypothesis 5. Neither the main effect of gender ideology nor the interactive effect of gender and gender ideology are statistically significant in Model 7. In addition, the direction of the effects suggests that women's attitudes are slightly more influential than are men's, a finding that is corroborated by the larger zero-order correlation between early attitudes and housework allocation for cohabiting women compared to cohabiting men that is shown in Table 2. It is also interesting to note that the magnitude of the coefficient for the influence of cohabiting men's attitudes in Model 7 is identical to that of the coefficient in Model 2 for married men (.09). This suggests that it is the strong influence of cohabiting women's early attitudes on their later housework allocation that is responsible for the absence of a significant gender-by-attitude interaction in Model 7. These findings show that the relationship between early attitudes and housework behaviors is not contingent on gender among cohabitors. Model 8 provides a test of Hypothesis 3 among cohabitors by adding measures of relative resources and time availability. This model demonstrates that when couple characteristics are added to the model, the magnitude of the influence of early attitudes on men's later participation in stereotypically female housework is reduced by approximately one fifth (from .20 to .16), and the influence of attitudes is no longer statistically significant. Model 8 provides evidence in support of Hypothesis 3 among cohabitors, suggesting that part of the influence of gender ideology early in adulthood on housework allocation later in adulthood is due to the influence of early attitudes on the contextual characteristics of cohabiting couples. Furthermore, when measures of relative resources and time availability are added to the model separately, the results suggest that each factor is responsible for approximately half of the total reduction in the influence of early attitudes on later housework (results not shown). Model 9 provides a test of Hypothesis 2 for cohabitors. In contrast to the models for married respondents, contemporaneous measures of cohabitors' gender ideology exert a smaller influence on the allocation of stereotypically female housework than do the earlier measures, and they do not appear to mediate the influence of early gender socialization. A model testing the interaction of gender and age-31 gender ideology found that this interaction was not statistically significant, offering additional support for Hypothesis 5 (results not shown). Model 10 includes the main effect of early and late gender ideology along with measures of relative resources and time availability and provides little evidence that the influ- Cunningham / GENDER IN COHABITATION AND MARRIAGE 1055 ence of contemporaneously measured indicators of gender ideology is mediated by couple characteristics among cohabitors. CONCLUSION The analyses presented here provide several original contributions to our understanding of the way attitudes about appropriate roles for men and women influence the gendered division of household labor. First, the results demonstrate complex but important relationships between early gender socialization and the subsequent division of routine housework. Among married respondents, women's attitudes about gender at the onset of adulthood are not associated with their subsequent patterns of routine housework. However, there is suggestive evidence that married men's early attitudes are related to the gendered division of labor in families many years later. Among cohabitors, egalitarian attitudes early in the life course are positively associated with men's relative participation in routine housework allocation later in the life course. Furthermore, this relationship does not vary by gender. Second, the analyses shed light on the process through which gender socialization and housework allocation are linked. Specifically, among cohabitors, a nontrivial proportion of the influence of early gender ideology operates by shaping couple members' subsequent levels of relative resources and time availability. However, the influence of early gender socialization on subsequent patterns of housework allocation is not mediated by couple characteristics among those who are married. Finally, the analyses showed that contemporaneously measured indicators of gender ideology are highly associated with married men's housework behaviors and that current attitudes mediate the influence of earlier attitudes. There are a number of theoretical implications of these findings. By testing for differences in the influence of gender ideology on the gendered division of labor by gender and marital status, the current analysis contributes to existing research that directs attention to the centrality of men's attitudes for contemporary family arrangements. In support of Greenstein's (1996) research, these analyses suggest that in the context of marriage, men's attitudes about gender are more strongly associated with their relative participation in routine housework than are women's. Although the magnitude of the influence of men's attitudes about gender at the onset of adulthood is not large in absolute terms, the analyses provide original evidence about the extent to which men and women are able to put their attitudes into practice. The results suggest that power dynam- 1056 JOURNAL OF FAMILY ISSUES / November 2005 ics in marriage operate in such a way that egalitarian men may participate more frequently in the kinds of tasks that have historically been assigned to women but that housework performance by the husbands of egalitarian women is not responsive to women's attitudes. Future research in other domains may be able to expand on this insight by analyzing the ways in which gendered power dynamics in marriage operate by producing differences in the relative ability of women and men to implement their preferences. The analysis of longitudinal data facilitated a more dynamic interpretation of the processes through which early gender socialization shapes the routine production of gender in families. First, the analyses demonstrated the long-term influence of married men's attitudes about gender on their participation in routine housework. Because gender ideology was assessed 13 years prior to the measurement of the division of routine housework, the analyses were able to more definitively isolate the causal influence of attitudes on later behavior than has been possible in previous research. However, future studies able to assess changes in housework allocation associated with changes in levels of gender egalitarianism would provide even stronger support for this hypothesis. Furthermore, the analyses showed that the influence of married men's early attitudes on later housework allocation patterns is largely mediated by men's attitudes at the time housework participation was assessed. Second, and more strikingly, the influence of married men's attitudes operates independently of other factors commonly associated with the division of household labor in previous research. Although men's share of the couple's income and women's employment hours were linked to housework allocation among married couples, these contextual characteristics of the couple did not reduce the magnitude of the impact of married men's earlier attitudes. These findings point attention to the importance of early gender socialization for men, suggesting that the influence of gender ideology is likely to endure regardless of wives' earnings or participation in the labor force. However, there was somewhat stronger evidence that attitudes were related to couple-level characteristics when each of the variables was assessed simultaneously. Research designs with smaller time intervals between measurements of the key variables might be better able to identify the processes through which gender ideology, union formation, employment and childbearing patterns, and gendered family interactions are causally linked. Several significant contributions of this investigation emerged through the comparison of the relationship between gender ideology and housework allocation among married and cohabiting couples. Most impor- Cunningham / GENDER IN COHABITATION AND MARRIAGE 1057 tantly, the analysis demonstrated a positive association between early gender egalitarianism and subsequent levels of men's relative participation in stereotypically female housework for cohabitors, and the magnitude of this relationship did not vary by gender. Indeed, there was suggestive evidence that the relationship was stronger for women than for men among cohabitors. Considered in light of the findings for married individuals, the results provide compelling evidence that gendered power relations operate differently for cohabiting couples than they do for married couples. This conclusion is supported by the results showing that among cohabiting couples women and men were equally able to put their gender ideology into practice. In terms of the most salient comparison, cohabiting women's relative participation in routine housework was reflective of their attitudes about gender, but this was not true for married women. It is possible that some women choose cohabitation because they believe they are more likely to establish the kind of gendered family practices they desire in a cohabiting union than in a marriage. As other researchers have noted, the institution of marriage carries with it a relatively strong set of assumptions about gendered family roles (Nock, 1998). Although most cohabitating unions in the United States are short lived, it seems that a segment of the population may choose cohabitation because it is perceived to minimize gender-related obligations relative to marriage. It is also important to recognize that some cohabiting women in this sample had previously been married, and they may have dissolved earlier marriages in part out of dissatisfaction with gendered expectations in those relationships. However, the influence of previous marriage on housework allocation among cohabitors was not significant, nor did the inclusion of this variable alter the influence of the other variables in the model (results not shown). Similarly, when a variable measuring previous experiences with cohabitation among the married respondents was included, it did not alter the observed relationships (results not shown). In combination, these results suggest it is the context of marriage that affects the nature of the observed relationship between attitudes and behaviors rather than the experience of cohabitation. Further supporting the idea that power relations between women and men are minimized in cohabiting unions, the results suggested that the process through which gender ideology is linked to the division of labor differs for cohabiting and married individuals. Specifically, in contrast to the direct influence of early measures of gender ideology on housework allocation observed among married men, there was greater evidence that this influence operates indirectly for cohabitors. Despite the relatively 1058 JOURNAL OF FAMILY ISSUES / November 2005 small sample of cohabitors, when measures of relative earnings and time availability were added to the models of housework allocation for cohabitors, the magnitude of the influence of early gender ideology was reduced by one fifth. This finding suggests that individuals with egalitarian gender ideologies in early adulthood are more likely to establish unions in which their cohabiting partners earn a relatively similar share of the income and have more similar amounts of time available for housework. In support of existing research, these findings suggest that the mechanisms through which gender operates differ among cohabiting and married couples (Brines & Joyner, 1999). This analysis has demonstrated the theoretical value of a more dynamic conceptualization of the relationship between attitudes about gender and housework allocation. The findings serve as a starting point from which we can attempt to broaden our specification of the processes through which housework allocation patterns are established over time and through which gender socialization influences gender-related behaviors more generally. It may also be important for researchers to devote greater attention to the way housework allocation and other gendered behaviors produce attitude change. Furthermore, this research highlights the importance of accounting for relationship context when studying the way that gender is produced through housework-related interactions. In support of existing research, this analysis uncovered little evidence that union type is associated with variation in housework allocation patterns. However, the factors and processes associated with the gendered division of labor differ for married and cohabiting couples, and these differences appear to provide cohabiting women with relatively greater voice in the negotiation of gendered behavioral displays relative to their married peers. APPENDIX Text for Attitude Measures Attitudes About Gender Index (1 = strongly agree; 5 = strongly disagree) 1. Most of the important decisions in the life of the family should be made by the man of the house. 2. It's perfectly alright for women to be very active in clubs, politics, and other outside activities before the children are grown up (reverse coded). 3. There is some work that is men's and some that is women's, and they should not be doing each other's. Cunningham / GENDER IN COHABITATION AND MARRIAGE 1059 4. A wife should not expect her husband to help around the house after he comes home from a hard day's work. 5. A working mother can establish as warm and secure a relationship with her children as a mother who does not work (reverse coded). 6. It is much better for everyone if the man earns the main living and the woman takes care of the home and family. 7. Women are much happier if they stay at home and take care of their children. 8. It is more important for a wife to help her husband's career than to have one herself. REFERENCES Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage. Atkinson, J., & Huston, T. L. (1984). Sex role orientation and division of labor early in marriage. Journal of Personality and Social Psychology, 46, 330-345. Barnett, R. C, &Baruch, G. K. (1987). Determinants of fathers' participation in family work. Journal of Marriage and the Family, 49, 29-40. Barnett, R. C, & Shen, Y. (1997). Gender, high- and low-schedule-control housework tasks, and psychological distress: A study of dual-earner couples. Journal of Family Issues, 18, 403-428. Berk, S. F. (1985). The gender factory: The apportionment of work in American households. New York: Plenum. Blair, S. L., & Lichter, D. T. (1991). Measuring the division of household labor: Gender segregation of housework among American couples. Journal of Family Issues, 12,91-113. Blood, R. O., & Wolfe, D. M. (1960). Husbands and wives. Glencoe, IL: Free Press. Blumstein, P., & Schwartz, P. (1983). American couples: Money, work, sex. New York: William Morrow. Brayfield, A. (1992). Employment resources and housework in Canada. Journal of Marriage and the Family, 54, 19-30. Brines, J., & Joyner, K. (1999). The ties that bind: Commitment and stability in the modern union. American Sociological Review, 64, 333-356. Clarkberg, M. E., Stolzenberg, R. M., & Waite, L. J. (1995). Attitudes, values, and entrance into cohabitational versus marital unions. Social Forces, 74, 609-634. Coleman, M. T. (1988). The division of household labor: Suggestions for future empirical consideration and theoretical development. Journal of Family Issues, 9, 132-148. Coltrane, S. (2000). Research on household labor: Modeling and measuring the social em- beddedness of routine family work. Journal of Marriage and the Family, 62,1208-1233. Coltrane, S., & Ishii-Kuntz, M. (1992). Men's housework: A life course perspective. Journal of Marriage and the Family, 54, 43-57. Coverman, S. (1985). Explaining husbands' participation in domestic labor. Sociological Quarterly, 26, 81-97. Cunningham, M., Beutel, A., & Thornton, A. (2000, March). The influence of gender role attitudes on life course events in early adulthood. Paper presented at the annual meetings of the Population Association of America, Los Angeles. 1060 JOURNAL OF FAMILY ISSUES / November 2005 Deutsch, F. M., Lussier, J. B., & Servis, L. J. (1993). Husbands at home: Predictors of paternal participation in childcare and housework. Journal of Personality and Social Psychology's, 1154-1166. Ferree, M. M. (1991). The gender division of labor in two-earner marriages. Journal of Family Issues, 12, 158-180. Gershuny, J., & Robinson, J. P. (1988). Historical changes in the household division of labor. Demography, 25, 537-552. Goldscheider, F. K., & Waite, L. J. (1991). New families, no families? The transformation of the American home. Berkeley: University of California Press. Greenstein, T. N. (1996). Husbands' participation in domestic labor: Interactive effects of wives' and husbands' gender ideologies. Journal of Marriage and the Family, 58, 585-595. Gupta, S. (1999). The effects of marital status transitions on men's housework performance. Journal of Marriage and the Family, 61, 700-711. Hardesty, C, & Bokemeier, J. (1989). Finding time and making do: Distribution of household labor in nonmetropolitan marriages. Journal of Marriage and the Family, 51, 253-267. Hochschild, A. (1989). The second shift. New York: Avon Books. Huber, J., & Spitze, G. (1983). Sex stratification: Children, housework, and jobs. New York: Academic Press. Kamo, Y. (1988). Determinants of household division of labor: Resources, power, and ideology. Journal of Family Issues, 9, 177-200. Kamo, Y. (1991). A nonlinear effect of the number of children on the division of household labor. Sociological Perspectives, 34, 205-218. Moors, G. (2003). Estimating the reciprocal effect of gender role attitudes and family formation: A log-linear path model with latent variables. European Journal of Population, 19, 199-221. Morgan, P. S., & Waite, L. J. (1987). Parenthood and the attitudes of young adults. American Sociological Review, 52, 541-547. Nock, S. L. (1998). Marriage in men's lives. New York: Oxford University Press. Orbuch, T, & Eyster, S. L. (1997). Division of household labor among black couples and white couples. Social Forces, 76, 301-332. Presser, H. B. (1994). Employment schedules among dual-earner spouses and the division of household labor by gender. American Sociological Review, 59, 348-364. Rexroat, C, & Shehan, C. L. (1987). The family life cycle and spouses' time in housework. Journal of Marriage and the Family, 49, 737-750. Robinson, J. R., & Godbey, G. (1997). Time for life: The surprising ways Americans use their time. University Park: Pennsylvania State University Press. Shelton, B. A. (1992). Women, men, andtime: Gender differences in paidwork, housework, and leisure. Westport, CT: Greenwood. Shelton, B. A., & John, D. (1993). Does marital status make a difference? Journal of Family Issues, 14, 401-420. Shelton, B. A., & John, D. (1996). The division of household labor. Annual Review of Sociology, 22, 299-322. South, S. J., & Spitze, G. (1994). Housework in marital and nonmarital households. American Sociological Review, 59, 327-347. Stafford, R., Backman, E., & DiBona, P. (1977). The division of labor among cohabiting and married couples. Journal of Marriage and the Family, 50, 595-618. Cunningham / GENDER IN COHABITATION AND MARRIAGE 1061 Starrels, M. E. (1994). Husbands' involvement in female gender-typed household chores. Sex Roles, 31, 473-491. Waite, L. J., Goldscheider, F. K., & Witsberger, C. (1986). Nonfamily living and the erosion of traditional family orientations among young adults. American Sociological Review, 51, 541-554. Waite, L. J., Haggstrom, G. W., & Kanouse, D. (1985). The consequences of parenthood for the marital stability of young adults. American Sociological Review, 50, 850-857. West, C, & Zimmerman, D. H. (1987). Doing gender Gender and Society, 1, 125-151.