Psychology of Women Quarterly, 1989, 13, 113-125. Printed in the United States of America. GENDER DIVISION OF TASKS BY PARENTS AND THEIR CHILDREN Ailsa Burns and Ross Hornel Macquarie University The division of parenting tasks between wives and husbands was explored in a sample of 279 Australian families containing a 9-1 1-year-old child, and the children’s performance of household tasks was compared with their parents’ division of labor. Three parenting factors (domestic care, leisure/enrichment, and psychological), and four child factors (handyman, domestic, maturity, and babysitting) were obtained. Mothers were more active than fathers on all three parenting factors, but significant differences were found between parents of differing SES and cultural backgrounds. Among children, the greatest sex difference was found on the handyman factor, where boys greatly outperformed girls. Boys’ performance on this factor was not related to parental division of labor or social background, but girls’ performance was higher when they were only children and when they lived in higher SES families. Girls outperformed boys on the domestic and maturity factors. Parenting style, SES, and cultural background also were significantly related to these factor scores. Apart from child’s sex, cultural background was the strongest predictor of children’s task performance. In Australia as in other industrialized countries there has been a movement o f m ar r ied wome n i n t o th e workforce in recent decades. This has been accompanied by growing assent t o the doctrine of sexual equality, by m e n as w e l l as among women. N a ti o n a l surveys o f 25-34-year-olds conducted in 1972 a n d 1982 f o u n d large changes in the endorsement of such items as “Im por t ant decisions should b e made by the h u s b a n d (endorsed by 43% of This research was supported by a grant from the Australian Research Grants Scheme and by a Macquarie University Research Grant. We are grateful to the Foundation of Child Development for permission to use questionnaire material, to Jacqueline Goodnow for valued advice, to John P. Murray of Kansas State University who helped design the project, and to Susan Cairns who acted as research assistant. Requests for reprints should be sent to: Dr. A. Burns, School of Behavioural Sciences, Macquarie University, N.S. W., Australia, 2109. Published by Cambridge University Press 0361-6843/89 $5.00 + .OO 113 114 BURNSAND HOMEL women in 1971, but only 17% in 1982) and “Whatever career a woman may have, her most important role in life is still that of becoming a mother” (78% in 1971, but only 47% in 1982, at which time similar proportions of men and women disagreed with this statement; Glezer, 1984). However, the change in attitudes has not translated into great changes in behavior within the home, where most families continue to adhere to the “traditional” division of labor, with women taking primary responsibility for household tasks (Adler, 1966; Clemenger, 1984; Harper & Richards, 1979; Oeser & Hammond, 1954; Russell, 1983). The one domestic area that does seem to have drawn greater paternal input is that of looking after children (Harper & Richards, 1979; Russell, 1983). How do parenting patterns affect children’s induction into the gendered division of labor? This is the question addressed by the present study. In line with the findings of Harper and Richards (1979) and of Russell (1983) that fathers’ involvement is most likely to be in parenting rather than in household work, we explored differences in the division of parenting tasks between families and then asked whether these differences are reflected in child behavior. As children do not of course engage in parenting behaviors themselves, we looked at some everyday children’s tasks where a marked gendered division of labor has been found from an early age (White & Brinkerhoff, 1981; Zill & Peterson, 1982). For the sake of clarity, the study is organized around three questions. First, we asked which childrearing tasks Australian parents are most likely to share, and which they are most likely to allocate to the male or female parent. Second, we explored sociodemographic differences between families in terms of the ways in which they divide parenting tasks. Third, we looked at children’s performance of sex-typed tasks and the extent to which this is associated with family background and with parental division of labor. Our general hypothesis is that some parents will be more traditional than others and that patterns of greater and lesser sex typing will be transmitted from one generation to the next. It is expected that differences between parents in performance of parenting tasks will not be random. Research on sex roles has regularly revealed that more egalitarianism is found among higher status parents, particularly when the mother is highly educated (Ericksen, Yancey, & Ericksen, 1979; Scanzoni & Szinovacz, 1980) or employed (Ericksen et al., 1979); among younger parents (Geerken & Gove, 1983) and parents of larger families (Brody & Steelman, 1985); and among those of certain cultural backgrounds (Scanzoni & Szinovacz, 1980). These variables are accordingly included in the study, in the expectation that they may influence parenting behaviors as well as general sex roles. Cultural background is defined in terms of country of origin (Australia has a large immigrant population). There is some evidence that children’s task performance also varies with family background. Zill and Peterson (1982) and White and Brinkerhoff 115 Gender Division of Tasks (1981) reported sex-role variations that in some ways parallel those noted for adults, in other ways not. Both sets of authors found that children of more educated parents were less sex-stereotypedthan others, both in terms of daughters’ engagement in male-typed tasks and of sons’ engagement in female-typed tasks, but White and Brinkerhoff found that low-income parents also made less distinction between boys’ and girls’ work. Maternal employment and larger family size were also associated with more femaletyped tasks being done by both girls and boys. However, neither study explored the association between the division of labor practiced by particular parents and that practiced by their children. It is therefore unclear whether, for example, the lessened sex-stereotyping of the higher status children resulted from the models presented by parents or from broader social influences. In the present study we explore this issue, by ascertaining whether parenting division of labor has effects on children over and above sociodemographic differences. In addition, we take White and Brinkerhoff‘s (1981) point that there are at least four classes of children’s task performance that can be affected (girls’ performance of female-typed tasks, boys’ performance of femaletyped tasks, boys’ performance of male-typed tasks, and girls’ performance of male-typed tasks) and that these need to be considered separately. Furthermore, we are aware that there may be cultural differences between Australia and the United States in what are considered to be suitable tasks for boys and girls. We accordingly make use of factor analysis to establish groupings among 14 everyday tasks (washing dishes, going to the shops, fixing things, etc.) and consider the performance of boys and girls separately on each of these groupings. METHOD Subjects The data come from a larger study of families living in 18 suburbs of Sydney, chosen to represent the socioeconomic and cultural diversity of the population (Vinson & Homel, 1976). For the present analysis, single-parent families were excluded, giving a sample of 279 9-ll-year-olds (133 boys and 146 girls) and their parents (or in some cases, stepparents, as all married couples with responsibility for children were included). Within the selected areas, children’s names were randomly selected from school rolls, and notes sent home to parents inviting their participation. Interviews with children were conducted at school, and those with parents at home, 2 to 8 weeks after the child had been interviewed. No more than one child from any particular family was included in the sample. 116 BURNSAND HOMEL Measures Division of parenting labor. This variable was measured by a 5-point Likert scale which asked parents to judge whether a task was done by mother only or far more often, mother usually, mother and father about equally often, father usually, or father only or far more often. The following parenting tasks were assessed: disciplining the children, transporting them to sports activities and functions, comforting them when sick in the night, seeing that children are clean and tidy, playing games with them, buying books for them to read, getting their breakfast, organizing their clothes, listening to them talk about the day’s activities, buying presents for them and their friends, and taking them on outings. These items were taken from the interview schedule developed by the New York Foundation for Child Development (Lash & Sigal, 1976). Children’s household task performance. Parents were asked whether the child had done each of the following tasks by him or herself, without anyone’s help: put batteries in a torch or toy; baked a cake or pie; built something out of wood with real tools, like a hammer and saw; changed sheets on a bed; cooked a complete meal for the family; ironed things; made a model like a model car, boat, or airplane; painted some object other than a picture; babysat; fixed a bicycle, truck, or roller skates; set the time on a clock or watch; gone to the shops by him or herself; washed dishes; or written a letter and posted it. These items also came from the Foundation for Child Development interview schedule (Lash & Sigal, 1976; Zill & Peterson, 1984). Demographic variables. Occupational status was scored in line with Congalton’s 7-point scale of Australian occupations, with an additional category for unemployed (Congalton, 1969). When both mother and father were employed, the higher status occupation was used (almost always that of the father). Paternal and maternal education were included as separate variables, using a 6-point scale. Family income was calculated by dividing total income by the number of persons dependent on this income, using a weighted formula developed by Saunders (1980). Mother’s employment status was classified as employed in the workforce (full- or part-time) or not employed. Cultural background was collapsed into four categories: AA (both parents native Australian born), NN (both parents born overseas in a non-English speaking country), AN/NA (mixed marriage between an Australian born and a non-English speaking born parent), and EE/E+ (one or both parents born in an overseas English-speaking country). Second generation immigrants where both were born in Australia were scored as AA. Most of the NN families were of Mediterranean origin, Greek and Greek-Cypriot in particular. Maternal age was scored as <30, 30-40, or > 40, and family size as five categories (3-7 persons). Child’s age (9-11) was treated as three categories, and child’s ordinal position as only, eldest, 117 Gender Division of Tasks Table 1 Means,o standard deviations, and rotated factor solution* for eleven parenting tasks Factor Loadings Parenting Task Mean SD Care 2 Leisure 3 Psychological Discipline Drive Comfort Clean Play 2.78 3.09 2.17 1.81 3.13 2.31 2.07 1.48 2.71 2.00 2.87 1.03 1.36 1.16 0.95 0.95 1.20 1.09 0.90 0.79 1.03 0.55 - - .73 - .39 - .48 .87 - - - .39 .53 2.40 0.53 1 Books Breakfast Clothes Listen Gifts Outings Total Cumulative variance (%) ff .34 .64 .84 ’ - - - - .43 .46 - .30 .52 .43 30 .74 44 .56 54 .55 - - Note: N = 279. al=Mother only, S=Fatheronly. bVarimax rotation, orthogonal. Only factors with eigenvalues greater than unity included. second, or later born. In light of the sampling procedure adopted, area of residence was also included, coded numerically as the SES level of the suburb, using the formula devised by Vinson and Home1 (1976), which in the present instance gave a range from very poor or “high r i s k (+3.63)to very good or “low risk” (-2.06). RESULTS Division of Parenting Tasks Table 1 contains mean values for the eleven parenting items and factor analysis of the items. With an overall mean value of 2.4 (2=mothers mostly, 3=both parents about equaZZy) and SD of 0.53, mothers were clearly carrying most responsibility for these children. Only on two items - playing with children and driving them to various activities - did the mean score fall above 3 (i.e., toward the father does end of the scale), and then only modestly (3.20, 3.14). Fathers were least likely to be involved in organizing children’s clothing, and keeping them neat and clean. The factor analysis describes three factors with eigenvalues greater than unity. Within the general framework of greater mother than father respon- 118 BURNSA N D HOMEL sibility for parenting tasks, the three factors represent three overlapping sets of tasks. The first factor (representing 30% of the variance) is the most stable (a=-74) and has its highest loadings on those tasks involving domestic care (cleaning, getting breakfast, organizing clothes). The second factor has its highest loadings on leisurelenrichment tasks (buying books and gifts, taking children on outings, driving them places, and playing with them); and the third on discipline and listening to children. The mean values for the items loading on each factor show that Factor 1 describes those tasks in which fathers are least likely to be involved, and Factor 2 those in which they are the most likely. Factor 3 brings together what could be considered the two most psychologically salient of the parenting tasks measured. Demographic Predictors of Division of Parenting Tasks A multivariate analysis of twelve demographic variables against the parenting factors gave a highly significant result, A = .54, x2 (117)=158.58, p = .006. Analyses of each predictor fitted independently against the three factors indicated that cultural background and three measures of socioeconomic status (occupational status, mother’s education, and residential area) were significantly associated with parenting factors. Cultural background, F (3, 275) =8.6, p = .OOO, r2= .086, occupational status, F (7, 271)=2.78, p=.OOl, r2=.067, maternal education, F (5, 273)= 5.33, p=.OOO, r2=.089, and residential area, F (3, 275)=3.75, p=.02, rz= .039, all correlated with Factor 2 (leisure/enrichment) scores, with higher paternal involvement associated with lower SES. Thus fathers with the lowest occupational and residential status were the most active in Factor 2 tasks, as were those who were married to mothers who had received only primary school education. Fathers of non-Anglo immigrant background were also more involved in these tasks, in particular in buying children gifts and transporting them to outings. Area also was related to Factor 1 (domestic care) scores, F (3, 275)=3.2, p=.O3, r2=.034, but in this case the pattern was less clear. Although child‘s sex was not a significant predictor when analyzed against the parenting factors, it did discriminate in respect to one item, play, F (1, 277)=8.84, p = .003, r2=.031, with fathers much more involved when the target child was a boy. Mothers’ workforce participation, not a significant predictor of any of the factors, was associated with fathers’ doing more listening to children, F (1, 277)=3.83, p=.05, r2=.014, and more chauffeuring, F (1, 277)=6.17, p=.Ol, r2=.022. Naturally, the various measures of socioeconomic status were correlated. In addition, cultural background was associated with occupational status and maternal education, due to the fact that most non-Anglo immigrants were in low-skill jobs and the mothers in particular lacked education be- Gender Division of Tasks 119 yond primary level. A series of multivariate linear models were accordingly constructed with the three parenting factors as dependent variables, in order to locate an adequate set of predictors. The best reduced model gave an overall A value (variance left unexplained) of .74, the variables retained being cultural background, A=.91, x 2 (9)=24.9, p=.002, and area of residence, A = .93, x 2 (9)= 19.2, p = .03, with occupational status also marginally significant, A = .89, x2 (21) =31.1, p = .06. Thus although the socioeconomic variables and cultural background are intercorrelated, each independently predicted the way that parents divided up their labor. Children’s Tasks The proportions of children who had accomplished the various children’s tasks ranged from 25% (cooked a complete meal for the family) to 93% (replaced batteries). Sex differences on all but three items (setting the time, babysitting, and cooking a meal) were significant beyond the .05 level. The greatest differences occurred on making a model (79%of boys versus 33% of girls), t (277)=9.10, p<.OOl; building something out of wood (81%, 33%),t (277)=9.54, p < .001; and fixing possessions like a bicycle (71%, 23%), t (277)=9.54, pC.001; ironing (86% of girls, 53% of boys), t (277)=6.42, p<.OOl; and baking (57%,28%),t (277)=5.35,p<.OOl. A factor analysis of the items gave four factors with eigenvalues greater than unity. The first comprised “handyman” activities, the second “domestic” items, the third comprised “maturity”-related items, and the fourth loaded on a single item, babysitting (Table 2). Parenting and Children’s Task Performance The association between parenting division of labor and children’s activities was very modest. Where fathers were more involved in leisurelenrichment tasks, daughters’ scores on the maturity factor were significantly lower, F (1, 277) =4.43, p = .04, r2= .03. Fathers’ contribution to parenting was not associated with boys’ performance of any of the tasks. Sociodemographic Variables and Children’s Task Performance Table 3 shows that parents’ cultural background, occupational status, mother’s education and age, family size, and child’s age, sex, and ordinal position all correlated with children’s task performance, with the strongest predictor being the child’s sex. Although sex differences on Factors 1, 2, and 3 were all highly significant, the greatest difference occurred on Factor 1 (the handyman scale), F (1, 277) = 128.7, p = .OOO, r2= .317. Sex differences on the items compris- BURNSAND HOMEL 120 Table 2 Rotated factor solution for children's household achievements" Factor Loadings Household Achievement Bake cakelpie Iron Change beds Cook meal Wash dishes Make model Fix things Paint things Carpentry Replace batteries Set time Shop alone Write and post letter Babysit Cumulative variance (X) Reliability CY 1 2 3 4 Handyman Domestic Maturity Babysit .66 .58 - .78 .70 .78 .64 - -83 .69 .84 .60 .75 .67 .57 30 .67 46 .53 45 .37 Note: N=279. aVarimax rotation, orthogonal. Only factors with eigenvalues values greater than 1.O included. ing this scale were strikingly larger than those on the domestic scale (Factor 2), F (I, 277) =38.05, p=.OOO, @=.121, paralleling the sex differences noted above on the different items. Girls also outperformed boys on Factor 3 (maturity) scores, despite a lack of age difference between boys and girls in thesample, F (1, 277)=10.87, p=.OOl, r2=.038. Handyman factor. The boys' superiority on this factor was expected. More interesting was the finding that none of the sociodemographic variables was related to the boys' performance of these tasks. In the same way that the boys' performance was unrelated to paternal involvement in parenting tasks, it proved impervious also to the effects of social class, cultural background, maternal employment, family size, and ordinal position. Even the child's age (across the 9-11 year range) was not a significant predictor. By contrast, the background of girls did appear to be associated with their acquisition of handyman skills. Those girls who lived in more advantaged areas, F (3, 142)=3.89, p=.Ol, r2=.076, those having more educated mothers, F (5,140)=2.98, p=.Ol, r2=.086, and those who were only children, F (3, 142) =2.62, p=.05, r2=.052, outperformed other girls in handyman skills. They still, however remained well below the perfor- Note: The sample consisted of 133 males and 146 females * p < . O 5 , **p<.Ol, ***p<.OOl. - - - - 5.2* - - - - - - - 31.7*** - - 4.9* - - - - - - - 3.5* - - - - - Girls - - 14.1*** 2.8* - 12.3*** Boys - 2.7* 12.1* - 9.4*** 3.3* - 9.3*** 5.2* Total - - 5.2* - - 3.9* - 5.8* - Boys - - 3.0" - Girls - 3.4** 3.8* - Total Children's Household Achievement Factors 2 3 Domestic Maturity - - - - - - 7.6** - - - - Total - - - - - - - - 9.6** - Girls - Boys 1 Handyman - - - - Girls - - - - Total 15.8** 12.4*** 13.6*** - - Boys 4 Babysit Table 3 (expressed as percentage) for associations between family variables and children's tasks performance: Total sample and by sex of child Family Occupation Maternal education Paternal education Income Area Cultural background Maternal employment Maternal age Family size Parental division of labor Factor 1 Factor 2 Factor 3 Child Age Sex Ordinal position R2 values 122 BURNSAND HOMEL mance level of the average boy. The mean score for daughters of the most highly educated mothers was 3.0, compared with a mean of 4.1 for sons of the least-educated mothers (which was the same as the average for all boys). Daughters who were only children had a mean score of 3.2, compared to 2.4 for other girls. Girls from families of higher occupational status and higher paternal education were slightly superior to other girls, but these effects were not statistically significant. Older girls performed no better than younger. Domestic and maturity factors. Both girls’ and boys’ scores on Factors 2 and 3 were related to sociodemographic factors. Cultural background proved the most powerful predictor, with sons in the non-Anglo families far less likely than other boys to have accomplished domestic tasks, F (3, 129)=7.07, p=.OOO, r2=.141, and somewhat less likely to have accomplished maturity tasks, F (3, 129)=2.66, p=.O5, r2=.058. The nonAnglo girls did not differ significantly from other girls on these factors, but their scores were a little lower. Two SES measures-occupational status, F (7, 271) =3.95, p = .001, r2= ,093, and mother’s education, F (5, 273) =4.6, p = .01, r2= .052, -also correlated with Factor 2 scores, with both sons and daughters (but particularly sons) of higher occupational status families and more highly educated mothers having mastered more domestic tasks. When mothers were in the workforce, both sons, F (1, 131)=3.75, p=.05, r2=.028, and daughters, F (1, 144)=5.17, p=.03, r2=.035, had achieved more domestic skills. The mean values, however, indicate that the sex difference remained consistent across the two groups (means of 2.6 and 3.6 for sons and daughters of the employed mothers, compared with 2.2 and 3.1 for other sons and daughters). A child’s age was correlated with both domestic, F (2, 276)=3.82, p=.02, r2=.027, and maturity, scores F (2, 276)=4.81, p=.O1, r2=.034, mainly because of improvement with age among the boys. Bearing in mind that girls scored higher on these two factors, it appears that girls accomplish these tasks earlier than boys and show minimal gains across the 9-1 1 year range, whereas this age range is a “learning band” for boys. Boys’ acquisition of maturity skills was also accelerated by being a member of a larger family, F (4, 128)=3.13, p=.02, r2=.039. Babysitting. Factor 4, which included babysitting one’s own siblings, was not associated with any of the above factors, except that the non-Anglo girls were more likely than other girls to have experience in this regard, F (3, 142)=3.86, p=.Ol, r2=.075. Most of the explained variance was associated with being the eldest child, F (3, 275) =14.37, p=.OOO, r2= .136 for the total sample, and to a lesser extent with family size, F (4, 274) = 2.67, p=.04, r2= .038. These two variables were of course correlated, Cramer’s V= .63, p = .OOO. 123 Gender Division of Tasks Relative Contribution of the Predictor Variables In order to establish the relative influence of the various predictor variables, a series of multivariate linear models was estimated. None of the parenting factors retained explanatory power over and above the effect of demographic variables. Thus, the significant association between paternal involvement in leisure/enrichment tasks and girls’ maturity scores was derived largely from variance shared with the demographic variables. It proved necessary to retain six of these in the final reduced model. Total A value (variance left unexplained) for the best final model was .30, with the six significantly contributing factors being child’s sex, A = .51, x2 (4)= 175.7, p = .OOO; child’s ordinal position, A= .83, x2 (12)=48.8, p = .OOO; mother’s education, A=.86, x2 (20)=39.7, p = .005; cultural background, A= .88, x2 (12)=33.5, p=.OOl; child’s age, A=.92, x2 (8)=21.8, p=.003; and maternal employment, A = .95, x2 (4)= 13.4, p = $005. Thus child’s sex accounted for the greatest portion of the variance, and socioeconomic status, cultural background, child’s ordinal position and age, and maternal employment made independent contributions. DISCUSSION Parental division of labor showed only a limited relationship to children’s gender division of labor. This was not because of lack of variability in either measure. The variations were considerable, and were systematically related to socioeconomic status, cultural background and some other variables. However, greater paternal input had no effect on boys’ achievement of any of the tasks measured and a detrimental effect on the girls’ achievement of maturity skills. Closer inspection of the data explained this detrimental effect. It was paternal involvement in driving children to activities and in purchasing gifts and books that proved principally responsible for this effect on daughters. The protocols showed that in a number of these families the mothers were unable to drive a car, or had little access to nonhousekeeping money, so that the daughters’ lesser possession of skills parallels that of the mothers. The fact that the effect disappeared when family background variables were included indicates that these patterns were class- and culture-linked. Thus we find no evidence that gender division of parenting labor (in terms of the tasks measured) is related to children’s sextyping, over and above the effects of family background. Family background, however, appears to be a significant predictor. At least four aspects of family background were important. The first is social status. Higher status parents had both daughters and sons with more maleand female-typed skills. Daughters from such families outperformed other girls on both types of skilb, and sons outperformed other boys on the domestic tasks. As these better-off families presumably had no greater need of children’slabor than other families, parental values can be considered to 124 BURNSAND HOMEL be the motive force here. Of all the status variables, mothers’ education was the strongest predictor of children’s skills, suggesting that it is maternal values that are salient and that higher education leads mothers to value the early acquisition of useful skills and to downplay sex differences. The second aspect is seen in terms of the non-Anglo families, who practiced a sharper division of parenting tasks than did other families, with mothers more involved in domestic tasks and fathers more involved in leisure/enrichment tasks. These differences reflect the Mediterranean origins of the majority of non-Anglo families. Mediterranean immigrant groups being generally considered to adhere to a more “patriarchal” life style than the general Australian population (Storer, 1985). The belowaverage domestic and maturity competencies of the non-Anglo boys could then result from a patriarchal cultural belief that these are not appropriate for boys. An alternative explanation comes from Goodnow, Knight, and Cashmore’s (1985) finding that non-Anglo mothers differed from Anglo-Australians in seeing no merit in children’s early acquisition of various skills and behaviors. As these mothers saw it, their children would learn quickly enough when the need arose, or the appropriate time arrived (“Time enough later”). This explanation fits with the finding that the non-Anglo girls also had somewhat less domestic and maturity skills than their samesex peers. We conclude that the non-Anglo families were implementing a different set of values to Anglo parents, one that was grounded in a different theory of child behavior. A third direction of influence comes from family needs. Maternal employment provides a clear example. When mothers were in the workforce, both sons and daughters had mastered more domestic tasks, but other skills were not affected. Another “needs” example is that of babysitting, where family size and child’s ordinal position were the major predictors. Finally, there is the interesting case of the elevated handyman scores of the girls who were “only” children. It seems that these girls owed their skills not so much to family values or family needs, but to the fact that they were treated to some extent as surrogate sons. An example was the “only” daughter - very close to her father and better-equipped with mechanical and handyman than domestic skills - whose parents commented that, “It would probably have been different if we’d had a boy as well.” The effects of age also merit comment. On both the domestic and the maturity tasks, where girls were more advanced, older boys outperformed younger boys, but older girls did no better than younger girls. We interpret this to mean that girls acquire a body of these competencies before age 9, while age 9-11 is a learning band for boys. However, on the handyman tasks age was not associated with boys’ or girls’ scores. Much of the boys’ learning had taken place by age 9; for girls it seems that learning comes at a later age, or perhaps never. This raises the question of the relative value of the different competencies to parents, sons, and daughters. Gender Division of Tasks 125 Our children’s scores are strikingly similar to those reported by Zill and Peterson (1982) in their large U.S. sample. Both show continuing very large sex differences, with the greatest difference being in the acquisition of handyman skills. Feminists have stressed the importance of girls’ ability to do their own maintenance jobs, as a source of empowerment, and an escape from dependency status (Barrett, 1980). Our findings lead us to conclude that establishing the conditions under which girls do or do not learn these skills constitutes the most important line of further inquiry. REFERENCES Adler, D. L. (1966).The contemporary Australian family. Human Relations, 19, 265-282. Barrett, M. (1980). Women’s oppression today. London: Verso/NLB. Brody, C. J., & Steelman, L. C. (1985). Sibling structure and parental sex-typingof children’s household tasks. Journal of Marriage and the Family, 47, 265-273. Clemenger, J. (1984). Beyond the stereotypes. Melbourne: John Clemenger Pty. Congalton, A. (1969). Status and prestige in Australia. Melbourne: Cheshire. Ericksen, J. A., Yancey, W. L., & Ericksen, E. P. (1979). The division of family roles. Journal of Marriage and the Family, 41, 301-313. Geerken, M., & Cove, W. R. (1983). At home and at work: The family’s allocation of labor. London: Sage. Glezer, H. (1984). Changes in marriage and sex role attitude among young married women: 1971-1982. Proceedings of the Australian Family Research Conference. Melbourne: Institute of Family Studies. Goodnow, J. J., Knight, R., & Cashmore, J. (1985). Adult social cognition: Implications of parents’ ideas for approaches to development. In M. Perlmutter (Ed.), Social Cognition: Minnesota Symposium on Child Development (Vol. 18, pp. 287-324). Hillsdale, NJ: Erlbaum. Harper, J., & Richards, L. (1979).M o t h and working mothers. Ringwood, Australia: Penguin. Lash, T., & Sigal, H. (1976). State of the child: New York. New York: Foundation for Child Development. Oeser, O., & Hammond, S . (1954). Social structure and personality in a city. London: Routledge & Kegan Paul. Russell, G. (1983). The changing role of fathers? St. Lucia: University of Queensland Press. Saunders, P. (1980). The poverty line: Methodology and measurement. Kensington: University of New South Wales Social Research Centre. Scanzoni, J,, & Szinovacz, M. (1980). Family decision-making: A developmental sex role model. London: Sage. Storer, D. (Ed.) (1985). Ethnic family values in Australia. Melbourne: Prentice-Hall. Vinson, T., & Homel, R. (1976). Indicators of community wellbeing. Canberra: Australian Government Publishing Service. White, L. K., & Brinkerhoff, D. B. (1981). The sexual division of labor: Evidence from childhood. Social Forces, 60, 170-181. Zill, N., & Peterson, J. L. (1982). Learning to do things without help. In L. M. Laosa & I. E. Sigel (Eds.), Families as learning environments for children (pp. 211-243). New York: Plenum. First draft receiued: September 4 , 1987 Final draft received: March 23, 1988