Parental Influences on the Gendered Division of Housework

by Mick Cunningham
Citation
Title:
Parental Influences on the Gendered Division of Housework
Author:
Mick Cunningham
Year: 
2001
Publication: 
American Sociological Review
Volume: 
66
Issue: 
2
Start Page: 
184
End Page: 
203
Publisher: 
Language: 
English
URL: 
Select license: 
Select License
DOI: 
PMID: 
ISSN: 
Abstract:

PARENTAL

INFLUENCES ON THE GENDEREDDIVISIONOF HOUSEWORK
MICK CUNNINGHAM

Western Washington University

Despite a recent emphasis on contextual explanations for the gendered division of housework, early socialization may also be important. Data from a 31-year panel study of white mothers and children are zlsed to examine parental predictors of the division of household labor among the adult children. Parental inflzlences are as- sessed when the children were ages 1 and 15; characteristics of the adult children are measzlred at ages 23 and 31. The effects of the parents'division of housework and parents' education on children's division of housework are considered, as well as the effects of the mother's gender-role attitudes and employment. The parental division of labor measured when the sons were very young has a positive effect on the sons' later participation in rolltine housework, while the mother S employment during their dazlghters' early years is a more important predictor of the allocation oj housework among the daughters. Parental influences are transmitted partially through the children's gender-role attitudes, but there is also evidence of enduring direct effects of parental characteristics on children's housework allocation, espe-

cially for men.

Direct correspondence to Mick Cunningham, Department of Sociology, Western Washington University, Arntzen Hall 532, Bellingham, WA 98225-9081 (cunningm@cc.wwu.edu). I thank Arland Thornton, Pamela Smock, Duane Alwin, Jay Teachman, Georgina Binstock, Sanjiv Gupta, four anonymous ASR reviewers, and the ASR Editors for helpful comments on early drafts of this article. Support for this research was pro- vided by an NICHD traineeship. A version of this article was presented at the annual meeting of the American Sociological Association, San Fran- cisco, August 1998.

household contexts. Several researchers claim that individuals who change their housework performance when they experi- ence these shifts in context are "doing gen- der" (S. F. Berk 1985; Perkins and DeMeis 1996; South and Spitze 1994; West and Zimmerman 1987).

Context-based differences in housework performance have also been drawn on to re- fute explanations of housework allocation based on gender socialization. Perkins and DeMeis (1996) argue:

The equal participation found among single women and men in domestic activities chal- lenges the view that gender differences in household work simply reflect a culturally learned pattern whereby women are taught that doing housework for oneself as well as for others is part of the gender identity. If women were acting out this type of gender socialization learned during childhood, then gender differences should occur at all stages of the family including the singles stage; however, significant differences appeared only at marriagelcohabitation and parent- hood stages. (P. 87)

This argument, like much recent work by interactionist-oriented scholars, favors con- textual explanations over those based on childhood socialization (S. F. Berk 1985; Ferree 1990; West and Zimmerman 1987).

I do not dispute the importance of contex- tual factors in shaping gendered behavioral patterns. Rather, I contend that the discov- ery of contextual effects does not disprove an explanation based on socialization. As Greenstein (1996) argues, there may be variation across individuals regarding the kinds of behaviors that signify gender. Parents' behaviors and attitudes during an individual's childhood may operate as an important source of variation in the way that individuals display gender.

I measure socialization in the domain of household labor by analyzing a regionally based probability sample of white mothers and their adult children. I focus specifically on attributes of the parental family during the early years of the children's lives. I as- sess the causal influences of parents' house- work allocation patterns, the mother's gen- der-role attitudes and employment, and the parents' education. By drawing on informa- tion about the parents from the children's in- fancy and adolescence, I am able to test hy- potheses concerning the relative influence of parental characteristics at different points in the children's lives. Further, by including predictors of housework allocation among the second generation and paying careful at- tention to potential gender differences in so- cializing influences, I am able to specify some of the mechanisms through which pa- rental characteristics influence children's housework allocation.

FAMILY SOCIALIZATION AND THE DlVlStON OF
HOUSEHOLD LABOR

There are several potential mechanisms through which parents can influence their children's household labor allocation pat- terns. There is substantial evidence that par- ents assign different tasks to their male and female children (Blair 1992; White and Brinkerhoff 1981), and that parents also dif- ferentially praise or criticize daughters and sons for performing housework (Block 1984; but see Maccoby and Jacklin 1974).' Most researchers who have exam- ined intergenerational influences on house- work allocation among adults have focused on the potential influence of housework modeling (Hochschild 1989; Koopman- Boyden and Abbott 1985; Stafford, Backman, and Dibona 1977; Thrall 1978).

According to the modeling hypothesis, children learn about gender appropriate be- haviors by observing and imitating the be- haviors of their parents (Bandura 1977). It is not clear, however, whether we should ex- pect observed behaviors to be reproduced once the source of modeling is no longer present (for instance, when children move out of their childhood homes). Symbolic interactionists have responded to shortcom- ings in the modeling explanation by arguing that gender is produced on the everyday level through ongoing behavioral displays. Fur- ther, they argue that these context-dependent behavioral displays (such as housework per- formance) are imbued with symbolic signifi- cance for gender (S. F. Berk 1985; West and ~immerman 1987). It is generally assumed, however, that there is widely shared agree- ment about what it means to "do gender."

Here I attempt to link the modeling and interactionist positions. I argue that children learn a particular set of gender-symbolic meanings for behaviors such as housework by observing their parents' daily interac- tions, then later draw on that repertoire of previously modeled behaviors to represent themselves as masculine or feminine in simi-

In one of the earliest statements applying so- cial learning theory to gender learning, Mischel (1966) wrote:

Social-learning theory makes no assumptions about the generality or consistency of behavior patterns across stimulus situations. The extent to which similar response patterns are manifested in a variety of situations is viewed as a function of the deg;ee of learned stimulus generalization for that response pattern and the specific contingen- cies in the situation rather than as a function of relatively situation-free underlying traits. (P. 66)

To the extent that research on gender social- ization specifies the scope of the learned behav- ior under examination and the context in which it is being analyzed, questions about the role of so- cialization should be treated as empirical prob- lems to be addressed rather than as inadequate models for understanding the nature of gender.

lar behavioral contexts (S. F. Berk 1985; Goffman 1977). I hypothesize that the gendered meaning of household tasks is "ac- tivated" upon entry into a particular context, but I explicitly acknowledge that family ex- periences may produce different understand- ings of the extent to which particular types of housework are viewed as opportunities to display gender.

Despite a paucity of long-term studies, some empirical evidence suggests that pa- rental modeling shapes the behavior of young children. For instance, although pa- rental reinforcements are not effective at changing the toys with which children choose to play, parents' own toy choices do influence the choices their children make (Eisenberg et al. 1985). It is possible that the household labor behaviors children observe in their parents may be even more signifi- cant for children's behavior than whether they were reinforced for performing (or not performing) particular household tasks.

In studies of adults, results have been mixed. In a sample of 50 dual-career couples, Hochschild (1989) finds that men who par- ticipated frequently in routine housework were no more likely to report having a father who modeled this kind of behavior than were men who did not share housework. Thrall (1978), on the other hand, finds evidence suggesting that "the best single predictor for the pattern of division of labor in any family may be what the pattern was in the adults' own families of orientation" (p. 249). Simi- larly, Stafford et al. (1977) present evidence that among married and cohabiting college students, rating of their father's housework participation was an important predictor of both the amount of time respondents spent on housework and the degree to which tasks were segregated by gender.

Whether these studies support or discon- firm the modeling hypothesis, they all are limited by relatively small convenience samples and retrospective reporting of the allocation of parental housework. The ab- sence of data gathered directly from parents makes retrospective indicators susceptible to potential bias in the child's perceptions of the parents' behavior and also in the child's ability to accurately recall behaviors that oc- curred many years earlier. I hypothesize that housework allocation for an adult is posi-

I tively related to his or her parents' house- work allocation patterns during that adult's childhood. However, by using data collected from the parents at multiple time points in the child's upbringing, I am able to estimate effects of parents' behavioral modeling that are not filtered through the child's percep- tions or memory. In addition to examining the influence of parents' housework performance on children's housework performance, I also explore other mechanisms involved in the process through which parents influence children's housework allocation. First, I con- sider the possibility that additional parental factors, such as attitudes about gender or maternal employment, may have a direct ef- fect on the way children divide or share housework in adulthood. Second, I attempt to identify how this array of parental char- acteristics can exert an indirect influence on children's housework allocation by influenc- ing other attitudes and behaviors of children that may be associated with the children's own patterns of housework allocation.

ADDITIONALPARENTALFACTORS

INFLUENCING CHILDREN'S

HOUSEWORK
ALLOCATION

I hypothesize that parental factors other than parents' housework performance may have a direct influence on children's housework al- location. Mothers who devote a relatively large proportion of their time to paid employ- ment may contribute to the erosion of gen- der-stereotypical housework behaviors in their children. This mechanism might oper- ate much like housework modeling-by minimizing the extent to which sons and daughters make an association between gen- der and the performance of particular kinds of work. Children of employed mothers may also be responsible for more housework, in the process developing skills in a broad range of household tasks. For instance, research has shown that the number of hours mothers are employed in the paid labor force is posi- tively related to the amount of time children spend on housework, especially for daugh- ters (Benin and Edwards 1990; Blair 1992).

Parents' attitudes may also be related to children's attempts to divide or share house- work. Egalitarian gender-role attitudes among parents have been linked to the per- formance of housework by children who are still living in the parental household (Blair 1992; Goldscheider and Waite 1991). Grow- ing up with parents who support equality be- tween women and men may lead children to be more likely to share housework. Parents' educational attainment is also expected to influence children's housework performance. High levels of education may lead to greater housework sharing among parents, and could also contribute to parents' attempts to treat sons and daughters similarly in terms of task assignment and reinforcement.
INTERVENING MECHANISMS AMONG THE SECONDGENERATION

In addition to considering the direct effects of varental characteristics on children's housework behaviors, I also examine poten- tial mediating factors among the children. If an individual's gender-role attitudes influ- ence the amount of housework he or she per- forms, for instance, it is important to under- stand the extent to which parental influences on housework allocation are the indirect re- sult of parents' influences on the individual's gender-role attitudes. The model that I propose includes measures of children's gender ideology, relative resources, and time de- mands (Kamo 1988; Shelton and John 1996). I posit that parents' characteristics influence not only the allocation of housework, but also each of these causal factors as well.

GENDERIDEOLOGY.One source of varia- tion in housework performance is the extent to which individuals support a role-special- ized model of the family. Many researchers argue that socialization into a particular set of attitudes toward "appropriate" family roles partially determines the way a given couple divides the labor in its own house- hold (Coltrane and Ishii-Kuntz 1992; Goldscheider and Waite 199 1 ; Greenstein 1996; Kamo 1988). Most studies find that men's attitudes about gender roles are im- portant determinants of household task per- formance (Blair and Lichter 1991; Kamo 1988; Ross 1987), and several studies report that women's attitudes are important corre- lates of housework allocation as well (Brayfield 1995; Presser 1994). I hypothesize that not only are individuals' attitudes about gender associated with the way they allocate housework, but also that children's attitudes are influenced by parental charac- teristics. Previous research has provided evi- dence that adult children's gender-role atti- tudes are associated with parents' housework allocation (Cunningham 2001), parental gen- der-role attitudes (Booth and Amato 1994; Thornton, Alwin, and Camburn 1983), ma- ternal labor force participation (Mortimer and Sorensen 1984; Wilkie 1987), and pa- rental education (Thornton et al. 1983). I expect that part, but not all, of the relationship between these parental characteristics and children's housework allocation is a result of the influence of parents' characteristics on children's gender-role attitudes.

Further, most studies of household labor have measured the division of housework and the attitudes of one or both spouses simultaneously (Coltrane and Ishii-Kuntz 1992; Coverman 1985; Kamo 1988). The current study performs a robust test of the effect of gender-role attitudes on housework performance by measuring these attitudes several years prior to the assessment of housework allocation patterns.

RELATIVERESOURCES. Research based on an exchange perspective demonstrates that women who have large earnings or hu- man capital resources relative to their male partners have greater equality in the division of housework than do women who are more "dependent" on their spouses for income or support (Blair and Lichter 1991; Kamo 1988; Shelton and John 1993). I hypothesize that not only does the relative income of the spouses affect the way they allocate house- work, but also that a couple's relative in- come is partially caused by parental charac- teristics. Parental influences on relative in- come might occur by directly influencing the accumulation of resources of one member of the second-generation couple, but parents may also influence characteristics in their children that would lead to a more equal di- vision of work responsibilities and employ- ment income more generally.

TIMEDEMANDS. Finally, the time de- mands faced by households also affect housework allocation. A significant cause of increased time demands is the presence of children (S. E Berk 1985; Gershuny and Robinson 1988). The presence and number of children in a household not only raises the total household work load, but also leads to greater segregation of tasks by gen- der and proportionately more housework hours for women compared with men (Hochschild 1989; Shelton and John 1993). I expect that as the number of children in a household increases, men will participate in a smaller proportion of the routine house- work. Further, I hypothesize that the par- ents' attitudes and behaviors shape the num- ber of children in the second-generation household as well.

Retrospective reporting of the parental divi- sion of labor does not allow for the assess- ment of effects related to the age of the child when the parental characteristics are mea- sured. Not only is the parental division of labor likely to change over time (Rexroat and Shehan 1987), parental characteristics may be particularly important at specific points in the life course of the child.

Child psychologists provide extensive evi- dence of the early development of gender- related psychological and behavioral pro- cesses, including the ability of children to classify themselves as girls or boys, to label others as girls or boys, and to understand the constancy of gender categories over time (Fagot and Leinbach 1993; Huston 1983). Further, research demonstrates that parental characteristics during very early childhood are critical determinants of later educational attainment and cognitive ability (Alwin and Thornton 1984; Duncan et al. 1998). I hypothesize that parental characteristics early in a child's life affect that child's eventual pattern of housework allocation. However, because children may not understand com- plex sets of attitudes at young ages, it is ex- pected that early parental behaviors will be more important to children's later behaviors than will early parental attitudes.

Parental influences may also be important later in the life course. Specifically, adoles- cence may constitute a period in which children's lives are directly affected by par- ents' behaviors and attitudes regarding housework. Adolescents are commonly re- sponsible for more household tasks than are young children, so the experience of per- forming housework may contribute to the formation of beliefs and behavioral scripts about that work. While this study does not measure adolescent housework performance, previous research shows that young men and women are typically assigned different kinds of housework (Blair 1992; White and Brinkerhoff 1981), that girls are responsible for more housework than are boys (Peters 1994; Timmer, Eccles, and O'Brien 1985), and that teenagers perform greater amounts of housework if their mothers are employed (Benin and Edwards 1990; Goldscheider and Waite 1991; Shelton 1992). Thus, adoles- cence may be a crucial time for socialization with regard to household labor. Parents' atti- tudes, in particular, are expected to take on greater significance as older children be- come better able to understand and interpret more complex ideas about gender.2

Despite an ongoing debate over the extent to which parents treat their daughters and sons differently (e.g., Block 1984; Maccoby and Jacklin 1974), some versions of socialization theory argue that the influences of the same- sex parent are especially important to gen- der-related socialization (Johnson 1975; Lynn 1969). If children identify more closely with the same-sex parent, we might expect to find higher levels of parent-child similarity between either the mother-daugh- ter dyad or the father-son dyad. However, few studies have tested these hypotheses over long periods of time, and several have shown little evidence of gender differences in parents' influences on children's gender- role attitudes (Acock and Bengtson 1978; Thornton et al. 1983). Nonetheless, sons' and daughters' eventual housework alloca- tion patterns may be shaped by different pa- rental attributes. Throughout this analysis I investigate the moderating effects of gender

Because the measures I use are separated by many years, I capture "early" and "late" child- hood influences crudely. The measures of paren- tal characteristics are proxies for parents' attributes across a range of early and late years rather than as tests of highly specific age-related hypotheses.

on the relationship between parental charac- teristics and children's behaviors.

In summary, I test three sets of hypoth- eses. First, I argue that parents' housework allocation, mother's gender-role attitudes, mother's paid work experience, and parents' education are related to the relative propor- tion of routine housework performed by men. Second, I argue that these parental in- fluences may occur directly, but are partially mediated by characteristics of the children, including children's gender-role attitudes, childbearing patterns, and economic resources relative to their spouses or partners. Finally, I hypothesize that gender may con- dition the relationship between parental at- tributes and children's housework allocation.
DATA

Data relevant to the study of intergener- ational patterns in household labor are avail- able in the Intergenerational Panel Study of Parents and Children (IPSPC). The data set has two components. A first-generation sample, composed of a cohort of mothers, was interviewed initially in 1962, then again in 1977. These women were selected from a probability sample of July 1961 birth records of first-, second-, and fourth-born white chil- dren in the Detroit metropolitan area. The second-generation cohort, or "children," was initially interviewed in 1980 (age 18), then again in 1985 (age 23) and 1993 (age 31). The sample used in this analysis includes all mother-child pairs in which the mothers were interviewed in 1962 and 1977 and whose 31-year-old daughter or son was in- terviewed in 1993. It includes 887 mothers- child dyads, representing 95 percent of those children originally interviewed in 1980. Be- cause of the nature of cluestions on the divi- sion of housework, the sample is further re- stricted to children who were married or co- habiting at the time of the 1993 inter vie^.^

Seventy percent (612) of the children were members of married or cohabiting heterosexual unions in 1993. However, the sample was re- duced by an additional 70 cases for several mod- els because of the exclusion of cases in which the mother was not married at the time of the 1977 interview. The 1977 division-of-labor questions were not administered to mothers who were single or cohabiting. Sensitivity tests show no

The data provide parental information that was assessed when the children were age 1 and 15 (1962 and 1977, respectively). The final sample is made up of 284 men and 324 women, for a total of 608 pairs of mothers and children.

These data are particularly valuable be- cause they cover a long time span and in- clude directly reported information from two generations. Further, the IPSPC also in- cludes rich measures of family behaviors and attitudes. Nonetheless, several limita- tions of the data may be relevant to this analysis. The sample was initially drawn from a relatively small geographic area, and although respondents have spread beyond the Detroit metropolitan area since the first wave of data collection, generalizability may be limited for this reason. However, most characteristics of the sample match up favor- ably with those of national-level studies (Thornton and Axinn 1996). Perhaps more serious, the sample is limited to white re- spondents. Generalization about intergenera- tional processes among other raciallethnic groups must await replication with more broadly representative data.

MEASURES
HOUSEWORK

ALLOCATION. 1 ~OCUSon the relative participation of men and women in stereotypically female household tasks. Al- though the ability to examine a wide range of tasks performed by both men and women is ideal for studies of this kind, data limitations restrict the analysis to the examination of "fe- male tasks." Despite the absence of data on "male tasks," the approach used here is worthwhile for several reasons. First, the housework stereotypically defined as "men's" constitutes a relatively small propor- tion of the total housework load (S. F. Berk 1985; Robinson 1997). Second, housework stereotypically performed by women is more repetitive and ongoing than is men's house- work, in addition to offering less discretion as to when, where, and how often it is per-

significant differences in the mean levels of housework performance or gender-role attitudes of the children dependent on the marital status of the mother in 1977 (results available on request from the author).

formed (S. F. Berk 1985). Finally, the hy- potheses I test here are designed to examine the learning of gendered meanings for tasks as much as the total allocation of housework.

The division of household labor among both parents and children is assessed using an index composed of items tapping partici- pation in specific household tasks, including "shopping for groceries," "doing the evening dishes," and "cleaning up before company

come^."^ The questions are proportional in nature, and are coded so that a score of 5 means the man "usually" performs the task and a score of 1 means the woman "usually" performs the task. The items are averaged into an index of participation in the three tasks. The text of these items is shown in Appendix Table A. Unfortunately, the avail- able measures represent tasks that are rela- tively more likely to be performed by men than are other "female" tasks, such as cook- ing and laundry, so the conclusions may be limited to a restricted domain of tasks. Fur- ther, although the measures cover a small range of tasks, the advantages of directly re- ported data from both parents and children over a 31-year time span justify the consid- eration of conclusions based on these data.

Because the analyses involve the exami- nation of housework allocation only among those 31-year-old children who were married or cohabiting at the time of the interview, respondents were selected into the sample in a way that might produce biased parameter estimates due to censored sampling (R. Berk 1983). There may be differences in single individuals compared with those included in the subset examined in the analyses that are correlated with the dependent variable. Coltrane and Ishii-Kuntz (1992), for in- stance, suggest that those who marry early are more likely to imitate the housework pat- terns of their parents. Because respondents in this analysis were selected based on their relatively early union formation in compari- son to their peers, selected respondents may be more likely to be influenced by their par- ents' behaviors than would the population as

4~n

both 1962 and 1977, the items asked of the mothers were identical in substance to those asked of the children, although an additional measure concerning "putting the baby to bed" was included in 1962.

a whole. It is important to bear in mind, however, that while approximately 30 per- cent of the children are excluded from the sample because they were not in a union at the time of the age-31 interview, only 14 percent of the children had never formed a marital or cohabitational union by age 31. Although some respondents were not selected for the analysis because they were not currently cohabiting or married, their exclu- sion does not mean that they had never formed such a relati~nship.~
ADDITIONAL

MEASURES. Gender-role at- titudes of the parents and children are mea- sured using a set of questions representing attitudes about appropriate roles of women and men in the family context (see Appen- dix Table A). These items were measured in 1962, 1977, and 1985.6 They are coded so that a high score represents support for egalitarian roles in the family.

Relative resources are measured as the ra- tio of the man's annual income to the couple's combined annual income (expressed as a percentage). I assess the time demands associated with the presence of children with a measure of the total number

One way to examine the selectivity of the sample is to look at variables not included in the models that might correlate with the dependent variable. To accomplish this, I compared the gen- der-role attitudes of the two groups (singles and marriedlcohabiting) measured in 1993. Three of

-,

five items showed statistically significant differ- ences in mean values. Children who were not members of a couple at age 31 had slightly more egalitarian attitudes than did those who were married or cohabiting. While the differences are not completely consistent across all five attitudi- nal measures, the possibility exists that there are substantively important differences in the two groups that would produce inaccurate estimates of the intergenerational processes at work. At the same time, Stolzenberg and Relles (1990) dem- onstrate the difficulties inherent in using existing statistical techniques to make corrections for se- lection into the sample.

Some respondents did not participate in the 1985 wave of the IPSPC. For those few cases, values on the gender-role attitude items from the 1980 wave of the study were substituted for the missing 1985 information. Three items in the 1977 mother's and 1985 child's gender-role atti- tude factors were not asked of the mothers in the 1962 interview.

Early ParentalVariables Parents' division of labor Mother's gender-role attitudes Mother's paid employment hours Parents' education
Late Parental Variables

Parents' division of labor

* Mother's gender-role attitudes Mother's paid employment hours

Child's lntervening Variables

Gender-role attitudes

Child's Contemporaneous lntervening Variables

Man's share of household income

Number of children in household

I1

Outcome Variable

Adult child's household division of labor

Figure 1. Heuristic Model of the Influence of Parents on Their Adult Children's Household Division
of Labor

of children living in the household at the time of the 1993 interview.

Two indicators of the mother's paid work experience are included. The total hours the mother was employed between 1962 and 1966 represents a measure of early work experience, while the hours she worked from 1967 to 1977 represents the mother's work experience when the child was 01der.~Finally, the education level of both the father and the mother in 1962 is measured in years. These education measures are treated as individual indicators of a common factor of parental education.

The examination of the timing of parental influences is accomplished by comparing ef-

The coding scheme used here means that a mother who worked outside the home full-time for one year (2,080 hours) would receive the same value as a mother who was employed halftime for two years. In that sense, it measures total time away from the home rather than capturing the difference in full-time employment, part-time employment, and no employment.

fects of variables measured during the infancy and mid-adolescence of the children. The general model is that the parents' characteristics measured when the children were age 1 (1962) and age 15 (1977) exert direct effects on the children's housework allocation, and that the parents' time-1 (1962) measures have an indirect effect on the children's housework allocation that is mediated by the parents' time-2 (1977) measures and by the children's gender-role attitudes, relative resources, and time demands (see Figure 1).

The 1962 measures of housework allocation, gender-role attitudes, maternal work experience, and education are used to capture early childhood influences, as all children were almost one year old when the initial interview took place. From 1977, the parental division of housework, the mother's gender-role attitudes, and the mother's paid work are used to measure influences that occurred later in the life course, when the children were 15 years old.
METHODS

Analyses are carried out using structural equation modeling techniques that are able to estimate parameters in a dynamic context and account for measurement error in the observed variables. Further, these techniques easily facilitate the analysis and testing of group differences and the decomposition of
effect^.^

Structural equation modeling is used to estimate a measurement model and a struc- tural model. The general model for the mea- surement model is y = h,q + E for endog- enous variables and x = hxc+ 6 for exog- enous variables; the structural model may be written as q = B(g)q+ T(g)c+ <; the y and x variables are observed indicators, the q and 6 are unobserved or latent variables, the E and 6 are random errors of measurement, the < are random disturbances in the structural equations linking the latent variables, the A,, Ax,r,and B are coefficient matrices, and g identifies the comparison groups of interest (Lomax 1983). Items with only one mea- sured variable are treated as singly measured factors with perfect mea~urement.~

Analyses were carried out using LISREL 8.12 (Joreskog and Sorbom 1996).

The key measures of parents' and child's housework allocation are treated as singly mea- sured indexes rather than as multiple indicators of a latent construct. A latent factor of task seg- regation may exist, in which case each of the be- havioral indicators of housework performance could be considered "effects" of this latent vari- able (Bollen 1989). However, it is also possible that instances of housework performance func- tion as a cause. Just as causally independent stressful events may contribute to overall levels of stress, individual performances of particular kinds of household tasks may accumulate to pro- duce a heavier or lighter housework load. Ac- cording to such a conceptualization, each indica- tor of housework is considered a cause of house- work allocation rather than an effect. Without data on a wider range of tasks, it is impossible to know whether a particular man who never washes the dishes also never washes clothes, never cooks dinner, and never cleans the bath- rooms, or if he "buys out" of the dish washing precisely because he does all of these other kinds of work frequently. Thus, it becomes difficult to conceive of these individual items as effects of a latent variable. In the end it is impossible to ad-

In addition to estimating a measurement model, the models estimate direct and indi- rect structural effects of the parental con- structs on the children's characteristics. Be- cause no causal assumptions are made be- tween any intratime variables, the early (1962) constructs are allowed to correlate, as are the errors of prediction of the late (1977) factors and the errors of prediction between the children's relative income and childbear- ing. Each factor is allowed to influence all subsequently measured factors.

Measurement model properties are pro- vided in Appendix Table A. Appendix Table B provides tests of measurement equiva- lence for sons and daughters. Within the measurement model, two error variances for attitudinal items measured at age 23 are sta- tistically different for sons and daughters, and these are left unconstrained in the mod- els. Although an omnibus test of structural equivalence did not provide clear evidence that structural parameters differed by gender, there is a moderate decline in model fit when parental predictors are constrained to equal- ity. Hence, the subgroup analyses are per- formed for the structural model and variable- specific differences are individually tested.

RESULTS

Mean values of the variables, shown in Table 1, indicate that men report higher levels of sharing in 1993 than women do (t = 4.32; p < .001). This result may partially reflect social desirability effects in question re- sponses on the part of men and women. However, some degree of this gender differ- ence may result from the fact that women marry and have children at earlier ages than men and thus have slightly different house- hold contexts at age 31. One example is sug- gested by the larger number of children in daughters' households (1.38)compared with sons' (1.20).1°Although the number of chil-

judicate definitively between these two measure- ment approaches. Relatively low factor loadings of the housework measures when they are treated as "effect" indicators provide at some additional evidence supporting the measurement strategy used in these analyses.

lo The difference in the mean number of chil- dren living in the household by gender is statisti- cally significant (t = -2.13, p c .05).
1962 to 1993
Age of Variable Child Mean     Son Standard Deviation     Daughter Standard Mean Deviation
Child's Characteristic Household division-of-labor index 31
           
Child's gender-role attitude:
Housework 23
           
Men's worWwomen's work 23
           
Men at worldwomen at home 23
           
Women happier at home 23
Wife should help husband's career 23
           
Man's relative share of household income 30
(expressed as a percentage)
Number of children in household
31             
Parents' Characteristic Early household division-of-labor index
1 15Late household division-of-labor index a
           
Mother's early gender-role attitude:
1Housework
           
1Men's worWwomen's work
           
Mother's late gender role attitude:
15Housework
           
15Men's worldwomen's work
           
15Men at worWwomen at home
           
15 15 Women happier at home
Wife should help husband's career
           
Mother's total early paid employment
4 Mother's total late paid employment
15
1
1
hours (in 1,000s), 1962 to 1966
1 to hours (in 1,000s), 1967 to 1977
5 to Father's education level (in years)
Mother's education level (in years)
           
Number of cases             
a For this variable, N = 250 for sons and N = 292 for daughters.             

dren in the daughters' households is 13 per- predictive measures and the children's cent greater than the number of children in housework allocation at age 3 1 provide pre- the sons' households, the sons' reports of liminary evidence that parents' characteris- their relative participation in housework are tics have a long-lasting impact on the 15 percent greater than the daughters' re-children's housework allocation (Table 2). ports of their husbands' relative housework Specifically, the parental division of labor performance. Importantly, the children re- when the child is age 1 and maternal gen- port higher mean levels of task sharing in der-role attitudes when the child is age 15 1993 than did their mothers in 1977. are significantly correlated with the son's re- The zero-order correlations between the ported division of housework, while early

Table 2. Unstandardized Zero-Order Factor Correlation Coefficients between Predictor Variables and Sons' and Daughters' Household Division of Labor at Age 31: IPSPC, 1962 to 1993

Child's 1993
Household
Division of Labor

Predictor Variable Son Daughter

Parental Characteristic

Early household division
of labor

Late household division
of labor a

Mother's early gender-
role attitude

Mother's late gender-
role attitude

Mother's early paid
employment hours

Mother's late paid
employment hours

Parents' education level

Child's Characteristic

Gender-role attitude
at age 23

Man's relative share of
household income
at age 30

Number of children in
household at age 3 1

Number of cases

Note: Numbers in parentheses are standard errors. a For sons, N = 250; for daughters, N = 292 *p< .05 **p< .O1 xxxp< ,001 (one-tailed tests)

maternal labor force activity and parents' education are associated with the daughter's division of labor.

EARLYAND LATE PARENTAL INFLUENCES ON CHILDREN'S HOUSEWORK

ALLOCATION

Table 3 presents structural equation model- ing coefficients that estimate the direct ef- fect of parental variables on the division of household tasks. The models are designed to test the relative influence of the three paren- tal characteristics that are measured at two points in time. Model 1 includes the early and late measures of the parental division of labor. This model shows that among the sons there is a strong positive effect of the par- ents' division of labor when the child was one year old on the son's allocation of housework at age 3 1. The relative amount a man's father contributed to the stereotypi- cally female housework when the son was very young predicts the relative amount the son participates in the same type of work in adulthood. Controlling for the later parental division of labor does not reduce the effect of the parents' early division of labor, sug- gesting an important role for modeling in very early childhood. There is no evidence of household labor modeling effects among the daughters.

Model 2 in Table 3 replaces the division- of-labor variables with the multiple indica- tor factors representing mother's gender-role attitudes. In this case, the sons' division of labor is predicted by the mothers' gender- role attitudes measured during the children's mid-adolescence." Greater support for egalitarian roles between women and men by the mother when the son is 15 years old increases the relative contribution of men to the stereotypically female housework at age 31, controlling for the mother's gender-role attitudes when the son is age 1. Among the daughters, results are consistent with the zero-order models-neither measure of ma- ternal attitudes is associated with the daughter's housework allocation.

Model 3 of Table 3 estimates the effect of measures of the mother's labor force partici- pation. When the mother's hours of paid em- ployment between 1967 and 1977 are con- trolled, the magnitude of the direct influence of the mother's employment when the child is age 1 to 4 is reduced only slightly in rela- tionship to its standard error. This finding suggests that increases in the mother's early

l1 In fact, the relative magnitude of the stan- dardized effect of the mother's gender-role atti- tudes increases from .17 in Table 2 to .23 in Table 3. This unexpected result is most appropri- ately interpreted in statistical rather than substan- tive terms. Because of the relatively high corre- lation between the two parental measures com- pared with the correlations between the parental measures and the child's division of labor, the magnitude of the direct effects is altered (Gor- don 1968).

Table 3. Unstandardized Structural Equation Model Coefficients of Direct Effects of Parents' Characteristics on Child's Household Division of Labor at Age 31: IPSPC, 1962 to 1993
Predictor Variable Parents' early household division of labor     Son Model 1 Model 2     Model 3     Daughter Model 1 Model 2 Model 3
Mother's early gender-role attitude             
Mother's early paid employment hours             
Parents' late household division of labor             
Mother's late gender-role attitude             
Mother's late paid employment hours             

Goodness-of-fit index Standardized root mean square residual

Note: Numbers in parentheses are standard errors.

* p < .05 "'p < .O1 '"p < .001 (one-tailed tests)

employment decrease the daughter's relative contribution to stereotypically female house- work. Further, this effect is not mediated by later maternal employment, pointing again to the importance of parental behaviors early in the child's life.

The comparison of early and late parental measures demonstrates variability in the tim- ing of parental effects. Parents' behaviors during early childhood appear to have a greater influence on children's housework allocation than do parents' behaviors during mid-adolescence, but early maternal attitudes are not as important as later attitudes. Although young children are not affected by their mothers' abstract attitudes about gen- der, parental behaviors during early child- hood manifest their effects many years later in the children's patterns of housework allo- cation. Table 4 presents an analysis includ- ing both parents' and children's characteris- tics that facilitates further insights into the intergenerational processes at work.

DIRECT EFFECTS OF PARENTAL PREDICTORS CONTROLLING

FOR CHILD'SCHARACTERISTICS

Model 1 in Table 4 is a baseline model that includes all parental characteristics with the exception of the late division of labor and the early maternal gender-role attitudes.12 Overall, the pattern of relationships observed in the simpler models holds. Among the daughters, the effects of parents' educa- tion and mother's early labor force partici- pation are slightly attenuated compared to

l2 These variables are excluded for two rea- sons. First, they have little direct or indirect in- fluence on the child's housework allocation. Sec- ond, issues of collinearity arise when models in- clude all early and late parental measures. The data suggest that collinearity between parents' housework allocation, maternal work, and par- ents' education are not a major concern. Early parental housework allocation is not correlated with parental education, (r = .00), and it is corre- lated with early maternal employment at only a moderate level (r = .07,p < .lo). However, some other correlations among parental variables are larger. There are no models, for instance, that si- multaneously assess the effects of all early and late measures of housework allocation, gender- role attitudes, and maternal employment. Draw- ing on the parameter estimates in Table 2 as a baseline, I attempt to ensure that the coefficients in subsequent tables make sense in light of these zero-order associations. In general, the estimates are stable across models, so it is doubtful that they are significantly biased because of collinear- ity among parental characteristics.

Table 4. Unstandardized Structural Equation Model Coefficients of Direct Effects of Parents' Characteristics and Child's Characteristics on Adult Household Division of Labor at Age

31: IPSPC, 1962 to 1993

Son Daughter

Model Model Model Model Model Model Model Model Model Model Predictor Variable 1 2 3 4 5 1 2 3 4 5

Parents' early household .25'" '25 " "26"" .26"\25.25""03 -.03 -.01 -.03 -.01 division of labor (.09) (.09) (.08) (.09) (.08) (.08) (.08) (.08) (.08) (.08)

Mother's early paid .OO .01 .01 .OO .01 .04* .04' .04 .04' .04* employment hours (.03) (.03) (.02) (.03) (.02) (.02) (.02) (.02) (.02) (.02)

Parents' education level .02 -.01 .OO .01 -.03 .05 .03 .05 .05 .04 (.03) (.04) (.04) (.03) (.04) (.03) (.04) (.03) (.03) (.04)

Mother's late gender-role .30** .22 .26" .31" 19 .05 .02 .03 .06 .01 attitude (.14) (.14) (.13) (.13) (.13) (.12) (.12) (.12) (.12) (.12)

Mother's late paid .OO .OO .OO .OO -.01 .OO .OO .OO .OO .OO employment hours (.Ol) (.Ol) (.Ol) (.Ol) (.Ol) (.Ol) (.Ol) (.Ol) (.Ol) (.Ol)

Child's gender-role -,51*** --.47""" .32" --.24

A

attitude at age 23 (.I81 (.17) (.I71 (.I61

Man's relative share of household income ---1,13*"" --1.08"" ---.80q"' --0.79"" at age 30 (.21) (.20) (.21)

Number of children in ---.09" -.01 ----.01 .06

A

household at age 3 1 (.05) (.05) (.04) (.04)

Goodness-of-fit index .97 .96 .97 .97 .96 .98 .97 .97 .98 .96

Standardized root mean .055 ,052 .053 ,053 ,046 ,049 .048 ,049 .048 .045 square residual

Note: Numbers in parentheses are standard errors.

'p < .05 "p < .O1 "*" p < ,001 (one-tailed tests)

the correlations in Table 2, and the effect of direct effect of the mother's attitudes on parents' education is no longer statistically housework allocation is reduced from .30 to significant (p< .lo). .22 and is no longer statistically significant.

Model 2 introduces the child's gender-role However, the influence of the early parental attitudes at age 23 (measured eight years division of labor is not affected by inclusion prior to the division-of-labor measure, at age of the son's attitudes. Among the daughters, 31). Son's and daughter's gender-role atti- the standardized parameter estimate for par- tudes at age 23 are statistically significant ents' education is reduced by almost 50 per-predictors of their housework allocation at cent (.l2 to .07, results not shown), suggest- age 31, net of parental factors. However, ing that the effect of parents' education on Model 2 also demonstrates that despite con- their daughter's housework allocation is trols for multiple parental characteristics and partly mediated by the daughter's attitudes the child's gender-role attitudes, the positive about gender. effect of the parental division of labor when Models 3 and 4 include all parental pre- the son is age 1 on the son's division of la- dictors as well as measures of the children's bor remains. Similarly, there continues to be relative income and childbearing, respec- a positive influence of the mother's early tively. Again, the effect of the parents' early employment on her daughter's housework division of labor on the son's housework al- allocation. Model 2 also demonstrates that location is not diminished. Further, Model 3 the effect of the mother's late gender-role at- shows that as the man's relative share of in- titudes on the son's housework allocation is come increases he does less of the stereo- partially mediated by the son's attitudes. The typically female housework. However, rela- tive income has a smaller attenuating effect on other parental characteristics than does the measure of children's attitudes. Model 4 shows a continuing negative effect of the presence of children on son's relative par- ticipation in housework when parental char- acteristics are controlled. However, the pres- ence of children has a relatively minor in- fluence on the magnitude of the parental ef- fects.

Model 5 includes all three of the child's intervening variables. There continues to be a strong effect of the early parental division of labor on the son's division of labor. In the full model, the effect of the daughter's gen- der-role attitudes is reduced slightly and is no longer statistically significant. Part of the influence of the daughter's gender-role atti- tudes on housework allocation may be due to the effect of her gender-role attitudes on the couple's relative income and childbearing.

The results in Tables 3 and 4 provide tenta- tive evidence that housework modeling is an important factor in understanding the extent to which men participate in "female" tasks relative to their spouses, but appears to have little to do with women's relative participa- tion in those tasks. A series of nested model comparisons was performed to further inves- tigate differences in the effects of parental characteristics on daughters and sons. The nested models approach to interaction exam- ines the change in model chi-square when a given parameter is first allowed to vary across groups (as in Tables 3 and 4) and then constrained to equality. A significant decline in model fit is evidence of a statistical inter- action (Hoyle 1996; Lomax 1983; Schu- macker and Marcoulides 1998). Despite only moderate evidence of gender differences in the structural model using an omnibus test for group differences (Appendix Table B), a series of multigroup comparisons were per- formed using Model 5 in Table 4 as a base- line. Each parameter estimate was individu- ally constrained to equality while the re- maining estimates were allowed to vary. The results show that the influence of the early parental household division of labor varies by the sex of the child (change in x2 = 6.20,

d.f. = 1, p < .05). Thus, I conclude with greater certainty that the early parental divi- sion of labor is a significant predictor of the division of labor for sons but not for daugh- ters. Tests for gender differences in the in- fluence of other predictive variables showed no additional evidence of significant differ- ences between sons and daughters (results available on request from the author). These findings suggest that the evidence of gender differences in intergenerational processes should be interpreted cautiously.
CONCLUSIONS

Using independent reports from parents and children and a probability sample, this analysis provides the best evidence to date regarding parental influences on household labor allocation. Nonetheless, it is important to consider several limitations of the present study. Information gathered directly from the fathers would have facilitated more rig- orous tests of some of the hypotheses. For example, a father's gender-role attitudes may constitute an unobserved factor that af- fects both the mother's and child's attitudes and behaviors (Greenstein 1996). For simi- lar reasons, the findings would be bolstered by data from both members of the second- generation couple as well as from the par- ents of both couple members. The data set also represents a relatively narrow segment of contemporary U.S. society. Future analy- ses using data with greater raciallethnic and regional heterogeneity are necessary before these results can be more widely general- ized. Nevertheless, the findings of this study, using intergenerational data spanning more than 30 years, provide a unique contribution to our knowledge of family processes. Com- parable analyses using the National Survey of Families and Households will not be pos- sible for at least another decade.

This analysis has demonstrated the impor- tance of parents' early behaviors for their adult children's allocation of hou~ework.'~

l3 Other models examined but not presented explored the influence of different parental char- acteristics including religious affiliation, reli- gious participation, and income. None of these parental attributes displayed any relationship with the child's division of labor.

For sons, parents' household labor patterns during their early years are significant deter- minants of sons' household labor allocation many years later. The results show that the extent to which fathers participate in a greater share of stereotypically female tasks when their sons are very young influences sons' relative participation in those tasks 30 years later, controlling for multiple addi- tional characteristics of both the parents and the child. However, this analysis made little progress toward understanding the mecha- nisms through which parents' early division of housework influenced their sons' behav- iors. Supplementary analyses suggested that parents have almost no effect on their sons' enjoyment of housework. It is possible that parents shape men's skill at performing housework efficiently through task assign- ment during childhood. Or, parents may con- tribute to sons' perceptions of the "appropri- ateness" of the performance of particular tasks by women or men. Future research that can further investigate these processes is clearly needed.

A mother's early labor force participation is an important predictor of her daughter's housework allocation. Though gender differ- ences in this case must be interpreted tenta- tively, the data suggest that increased hours of paid work among the mothers may have led to a decrease in the daughter's relative contributions to stereotypically female housework. For both sons and daughters, pa- rental behaviors early in the life course had long-term effects on the allocation of house- work. In each instance, the same-sex parent's early involvement in nonstereotypical pursuits, whether in the home or in the work force, influenced the children's later allocation of stereotypically female housework. Viewed in combination with other work on the causal influences of early parental characteristics, my findings rein- force the importance of early parental behav- iors for children's adult lives (Alwin and Thornton 1984; Duncan et al. 1998). Further, they extend the affected domains from cog- nitive abilities and educational attainment to gendered family practices.

The analyses also showed enduring ef- fects of the mother's gender-role attitudes during mid-adolescence on the child's housework performance at age 31, espe

cially among sons. This finding suggests that parental attitudes are more important than parental behaviors during children's adolescence. Although the long-term effects of parents' behaviors seem to dominate in early childhood, as children age they be- come increasingly responsive to ideas as opposed to actions. Additional research in- volving other ideational components of family life, such as religious beliefs or po- litical attitudes, might further enhance our understanding of the changing influence of ' parental attitudes or beliefs across the

' course of children's lives. While there was mixed evidence of gen- der differences in parental effects, it is im- portant to understand the reasons behind these differences. Because the observed ef- fects of parental characteristics were some- what stronger for men than for women, it seems reasonable to conclude that structural changes outside of the home are more im- portant for understanding women's side of the household labor equation. Women may gain a degree of power and independence by working outside the home, so their motiva- tion for changing the allocation of paid work and family work are more evident. For men, on the other hand, behaviors modeled in the family are clearly important in shaping their housework allocation patterns in adulthood. Perhaps the experience of living in a family that has pursued a more egalitarian house- work strategy makes it more likely that men see the potential gains of their contributions to domestic life, or at least will not interpret their own participation in family work as highly stigmatizing. Future research on pat- terns of family work need to take careful ac- count of gender differences in the social and psychological processes that are driving change.

Finally, it is important to note that a large portion of the parental effects were not the result of parents' influence on their children's attitudes. Rather, they were the direct effect of parents' behaviors and atti- tudes on children's later housework behav- iors. The inclusion of the children's attitudes reduced the magnitude of parental effects only partially in some cases and not at all in others. This finding constitutes a serious cri- tique of many previous studies of housework allocation, which have commonly opera- tionalized "socialization" using indicators of gender differences in these processes. By a respondent's gender-role attitudes. My recognizing that early experiences in the analyses show that such an approach is family contribute to individuals' conceptions likely to miss a large proportion of the ex- of what it means to "do gender" in a particu- planatory power of childhood socialization. lar context, we can approach a more compre- To achieve a comprehensive understanding hensive understanding of the mechanisms by of gender socialization in the family, we which housework takes on symbolic impor- must account for familial factors apart from tance for gender. individuals' attitudes.

By focusing on contextual effects at the Mick Cunningham is Assistant Professor of So- level of the couple, critics of socialization ciology at Western Washington University. His theories have ignored variation in the learn- general research interests include gender, family ing of gender-symbolic behaviors that oc- demography, and the life course. He recently curs between children and parents. Although published "The Influence of Parental Attitudes

and Behaviors on Children's Attitudes toward

context clearly operates as an important trig-

Gender and Household Labor in Early Adult-

ger for certain kinds of gendered behaviors,

hood" (Journal of Marriage and Family, 2001,

it is also true that family characteristics dur-

vol. 63, pp. 11 1-22). His current projects exam

ing childhood shape individuals' understand-

ine the causes and consequences of gender ideol-

ing of the symbolic meaning of particular

ogy among young adults (with Ann Beutel, Jenni-

behaviors for identifying and enacting gen-

fer Barber, and Arland Thornton), the influence

der. This study has demonstrated that in or-

of father-child relationships on gender-related der to account for variation in housework be- attitudes and belzaviors, and the reciprocal links haviors, it is important to consider influ- between employment and housework allocation ences from familial sources and to attend to over the life course.

Appendix Table A. Factors, Variable Labels, Factor Loadings, Error Variances, and Text for Primary Items: IPSPC, 1962 to 1993
Factor     Error
Factor     Variable     Loadine Variance     Text of Variable
Parents' early     Household division-     1.OO     -    Now, who shops for the groceries?
household division of     of-labor index             Who does the evening dishes?
labor                 Who straightens up before company is coming?
            Who puts the children (baby) to bed?
            (1 = Wife almost always, 2 = Wife mostly,
            3 = Husband and wife exactly the same, 4 =
            Husband mostly, 5 = Husband almost always)
Mother's early     Housework     .85     A wife should not expect her husband to help
gender-role             around the house after he comes home from a
attitude             hard day's work.
            (1 = strongly agree; 2 = agree; 3 = don't
            know; 4 = disagree; 5 = strongly disagree)"
    Men's work/    .86     .26     There is some work that is men's and some that
    women's work            is women's, and they should not be doing each
            other's.
Parents' education     Father's education     .79     .37     How many grades of school did your husband
level                 finish?
    Mother's education    .6 1     .62     How many years of school did you finish?
Parents' late     Household division-     1.OO     -    First, which category best describes how
household     of-labor index             grocery shopping is divided up in your family?
division of labor                 How about straightening up before company comes?
            (Appendix Table A continued otz next page)

(Auuendix Table A continued from urevious uane)

Factor Error Factor Variable Loading Variance Text of Variable

Parents' late (How about) doing the evening dishes? household

(1 = wife usually, 2 = wife a little more often, division of

3 = equal [if volunteered], 4 = husband a labor (cont.)

little more often, 5 = husband usually)

Mother's late Housework See "housework" above. gender-role

Men's work/ See "men's work/women's work" above.

attitude women's work

Men at work/ It is much better for everyone if the man earns women at home the main living and the woman takes care of the home and family.

Women happier Women are much happier if they stay at home at home and take care of their children.

Wife should help It is more important for a wife to help her husband's career husband's career than to have one herself.

Child's gender- Housework See "housework" above. role attitude

Men's work/ See "men's work/women's work" above. women's work

Men at work/ See "men at worklwomen at home" above. women at home

Women happier See "women happier at home" above. at home

Wife should help See "wife should help husband's career" above. husband's career

Child's household Household division- See "Parents' late household division-of-labor division of of-labor index index" above. labor

(1 = Woman usually, 2 = Woman a little more often, 3 = Equal [if volunteered], 4 = Man a little more often, 5 = Man usually)

Note: Perfect measurement is assumed for all factors with only one observed variable. Coding is in parentheses.

a Response format for all gender-role attitude questions is identical.

Error variances are significantly different. Values for unconstrained parameter estimates are included in paren- theses for sons/daughters.

Appendix Table B. Tests of Invariance of the Measurement Model and the Structural Model for Sons and Daughters: IPSPC, 1962 to 1993

Goodness-of-Fit Index

Degrees of Probability Model Hypothesis x2 Freedom Son Daughter Ax2/d.f. (Ax2/d.f.)

A Model form unconstrained 218.55 213 .96 .96 -

B A + parallel factor loadings equal 228.01 222 .96 .96 9.4619 NS

C B + parallel measurement error 257.91 234 .95 .96 29,9112 .01 variances equal

D C + measurement error variances 238.30 232 .96 .96 10.29110 NS of child's "housework" and "women happier at home" items free

(Appendix Table B continued on next page) (Appendix Table B continued from previous page)

Index

Degrees of Probability Model Hypothesis x2 Freedom Son Daughter Ax2/d.f. (Ax2/d.f.)

E D + parallel equation error variances 245.10 238 .96 .96 6.8016 NS equal

F E + parallel factor correlations equal 247.06 243 .96 .96 1.9615 NS

G F + parallel structural models equal 283.89 274 .95 .96 36.83131 NS

Note: Nested models compare change in model fit when elements of the model are constrained to equality for the two groups. Model C is rejected, and analysis of individual error variances suggests the best model allows the error variances of the child's measures of "housework" and "women happier at home" to remain unconstrained (Model D). The omnibus test of structural equivalence in Model G does not attain statistical significance at con- ventional levels, but tests of equivalence of individual structural parameters across sons and daughters were per-

formed (results available on request from the author).

REFERENCES

Acock, Alan C. and Vern L. Bengtson. 1978. "On the Relative Influence of Mothers and Fathers: A Covariance Analysis of Political and Reli- gious Socialization." Journal of Marriage and the Family 40:5 19-30.

Alwin, Duane and Arland Thornton. 1984. "Fam- ily Origins and the Schooling Process: Early versus Late Influence of Parental Characteris- tics." American Sociological Review 49:784

802. Bandura, Albert. 1977. Social Learning Theory. Englewood Cliffs, NJ: Prentice Hall.

Benin, Mary H. and Debra A. Edwards. 1990. "Adolescents' Chores: The Difference between Dual- and Single-Earner Families." Journal of Marriage and the Family 52: 361 -73.

Berk, Richard. 1983. "An Introduction to Sample Selection Bias in Sociological Data." American Sociological Review 48:386-98.

Berk, Sarah Fenstermaker. 1985. The Gender Factory: The Apportionment of Work in American Households. New York: Plenum.

Blair, Sampson L. 1992. "The Sex-Typing of Children's Household Labor: Parental Influ- ence on Daughters' and Sons' Housework." Youth and Society 24: 178-203.

Blair, Sampson L. and Daniel T. Lichter. 1991. "Measuring the Division of Household Labor: Gender Segregation of Housework among American Couples." Journal of Family Issues 12:91-113.

Block, Jeanne H. 1984. Sex Role Identity and Ego Development. San Francisco, CA: Jossey-Bass.

Bollen, Kenneth. 1989. Structural Equations with Latent Variables. New York: John Wiley and Sons.

Booth, Alan and Paul R. Amato. 1994. "Parental Gender Role Nontraditionalism and Offspring Outcomes." Journal of Marriage and the Fam- ily 56:865-77.

Brayfield, April. 1995. "Juggling Jobs and Kids: The Impact of Employment Schedules on Fa- thers' Caring for Children." Journal of Mar- riage and the Family 57:321-32.

Coltrane, Scott and Masako Ishii-Kuntz. 1992. "Men's Housework: A Life Course Perspec- tive." Journal of Marriage and the Family 54:43-57.

Coverman, Shelley. 1985. "Explaining Hus- bands' Participation in Domestic Labor." Sociological Quarterly 26: 8 1-97.

Cunningham, Mick. 2001. "The Influence of Pa- rental Attitudes and Behaviors on Children's Attitudes toward Gender and Household Labor in Early Adulthood." Journal of Marriage and Family 63:1 1 1-22.

Duncan, Greg J., W. Jean Yeung, Jeanne Brooks- Gunn, and Judith R. Smith. 1998. "How Much Does Childhood Poverty Affect the Life Chances of Children?" American Sociological Review 63:406-23.

Eisenberg, Nancy, Sharlene A. Wolchik, Robert Hernandez, and Jeannette F. Pasternack. 1985. "Parental Socialization of Young Children's Play: A Short-Term Longitudinal Study." Child Development 56: 1506-1 3.

Fagot, Beverly I. and Mary D. Leinbach. 1993. "Gender-Role Development in Young Chil- dren: From Discrimination to Labeling." Developmental Review 13:205-24.

Ferree, Myra Marx. 1990. "Beyond Separate Spheres: Feminism and Family Research." Journal of Marriage and the Family 52:866

84.

Gershuny, Jonathan and John P. Robinson. 1988. "Historical Changes in the Household Division of Labor." Demography 25:537-52.

Goffman, Erving. 1977. "The Arrangement be- tween the Sexes." Theory and Society 4:301

32.

Goldscheider, Frances and Linda Waite. 1991.

New Families, No Families? The Transforma- tion of the American Home. Berkeley, CA: University of California Press.

Gordon, Robert A. 1968. "Issues in Multiple Re- gression." American Journal of Sociology 73:592416.

Greenstein, Theodore N. 1996. "Husbands' Par- ticipation in Domestic Labor: Interactive Ef- fects of Wives' and Husbands' Gender Ideolo- gies." Journal of Marriage and the Family 58: 585-95.

Gupta, Sanjiv. 1999. "The Effects of Transitions in Marital Status on Men's Performance of Housework." Journal of Marriage and the Family 61:700-11.

Hochschild, Arlie. 1989. The Second Shift. New York: Avon Books.

Hoyle, Rick H. 1996. "The Structural Equation Modeling Approach: Basic Concepts and Fun- damental Issues." Pp. 1-15 in Structural Equa- tion Modeling: Concepts, Issues, and Applica- tions, edited by R. H. Hoyle. Thousand Oaks, CA: Sage.

Huston, Aletha C. 1983. "Sex Typing." Pp. 387- 468 in Handbook of Child Psychology, vol. 4, Socialization, Personality, and Social Devel- opment, edited by E. M. Hetherington. New York: Wiley.

Johnson, Miriam M. 1975. "Fathers, Mothers, and Sex Typing." Sociological Inquiry 45: 15-

26. Joreskog, Karl and Dag Sorbom. 1996. LISREL

8: User's Reference Guide. Chicago, IL: Sci- entific Software International.

Kamo, Yoshinori. 1988. "Determinants of Household Division of Labor: Resources, Power, and Ideology." Journal of Family Is- sues 9: 177-200.

Koopman-Boyden, Peggy G. and Max Abbott. 1985. "Expectations for Household Task Allo- cation and Actual Task Allocation: A New Zealand Study." Journal of Marriage and the Family 47:211-19.

Lomax, Richard G. 1983. "A Guide to Multi- Sample Structural Equation Modeling." Behavior Research Methods and Instrumentation 15:580-84.

Lynn, David B. 1969. Parental and Sex-Role Identification: A Theoretical Formulation. Berkeley, CA: McCutchan.

Maccoby, Eleanor E. and Carol N. Jacklin. 1974. The Psychology of Sex Differences. Stanford, CA: Stanford University Press.

Mischel, Walter. 1966. "A Social-Learning View of Sex Differences in Behavior." Pp. 56-81 in The Development of Sex Differences, edited by

E. E. Maccoby. Stanford, CA: Stanford Uni-

versity Press. Mortimer, Jeylan and Glorian Sorensen. 1984.

"Men, Women, Work, and Family." Pp. 137- 67 in Women in the Workplace: Effects on Families, edited by K. M. Borman, D. Quarm, and S. Gideonse. Norwood, NJ: Ablex.

Perkins, H. Wesley and Debra K. DeMeis. 1996. "Gender and Family Effects on the 'Second Shift' Domestic Activity of College-Educated Young Adults." Gender and Society 10:78-93.

Peters, John F. 1994. "Gender Socialization of Adolescents in the Home: Research and Dis- cussion." Adolescence 29:913-34.

Presser, Harriet B. 1994. "Employment Schedules among Dual-Earner Spouses and the Di- vision of Household Labor by Gender." American Sociological Review 59:34844.

Rexroat, Cynthia and Constance Shehan. 1987. "The Family Life Cycle and Spouses' Time in Housework." Journal of Marriage and the Family 49:737-50.

Robinson, John R. 1997. Time for Life: The Sur- prising Ways Americans Use Their Time. University Park, PA: Pennsylvania State Univer- sity Press.

Ross, Catherine E. 1987. "The Division of Labor at Home." Social Forces 65:816-33.

Schumacker, Randall E. and George A. Marcoulides, eds. 1998. Interaction and Non- linear Effects in Structural Equation Model- ing. Mahwah, NJ: L. Erlbaum.

Shelton, Beth A. 1992. Women, Men, and Time: Gender Differences in Paid Work, Housework, and Leisure. Westport, CT: Greenwood.

Shelton, Beth A. and Daphne John. 1993. "Does Marital Status Make a Difference?" Journal of Family Issues 14:401-20.

. 1996. "The Division of Household La- bor." Annual Review of Sociology 22:299-322.

South, Scott J. and Glenna Spitze. 1994. "House- work in Marital and Nonmarital Households." American Sociological Review 59:32747.

Stafford, Rebecca, Elaine Backman, and Pamela Dibona. 1977. "The Division of Labor among Cohabiting and Married Couples." Journal of Marriage and the Family 39:43-57.

Stolzenberg, Ross M. and Daniel A. Relles. 1990. "Theory Testing in a World of Constrained Re- search Design: The Significance of Heckman's Censored Sampling Bias Correction for Nonexperimental Research." Sociological Methods and Research 18:395415.

Thornton, Arland, Duane F. Alwin, and Donald Camburn. 1983. "Causes and Consequences of Sex-Role Attitudes and Attitude Change." American Sociological Review 48:211-27.

Thornton, Arland and William G. Axinn. 1996. "A Review of the Advantages and Limitations of the Intergenerational Panel Study of Parents and Children." Working paper, Institute for Social Research, University of Michigan, Ann Arbor, MI.

Thrall, Charles A. 1978. "Who Does What? Role Stereotyping, Children's Work, and Continu- ity between Generations in the Household Di- vision of Labor." Human Relations 3 1 :249-65.

Timmer, Susan G., Jacquelynne Eccles, and Kerth O'Brien. 1985. "How Children Use Time." Pp. 353-82 in Time, Goods, and Well- Being, edited by F. T. Juster and F. P. Stafford. Ann Arbor, MI: Institute for Social Research.

West, Candace and Don H. Zimmerman. 1987.

"Doing Gender." Gender and Society 1: 125-

51.

White, Lynn K. and David B. Brinkerhoff. 1981. "The Sexual Division of Labor: Evidence from Childhood." Social Forces 60: 170-8 1.

Wilkie, Jane R. 1988. "Marriage, Family Life, and Women's Employment." Pp. 149-66 in Women Working, 2d ed., edited by A. H. Stromberg and S. Harkess. Mountain View, CA: Mayfield.

Comments
  • Recommend Us