The British Journal of Sociology 2005 Volume 56 Issue 4 Crompton and Lyonette (Department of Sociology, City University) (Corresponding author email: r.crompton@city.ac.uk) London School of Economics and Political Science 2005 ISSN 0007-1315 print/1468-4446 online. Published by Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA on behalf of the LSE. DOI: 10.1111/j.1468-4446.2005.00085.x The new gender essentialism ­ domestic and family `choices' and their relation to attitudes1 Rosemary Crompton and Clare Lyonette Abstract This paper critically examines two strands within contemporary gender essentialism ­ that is, the argument that men and women are fundamentally different and that it is this `difference' that explains the continuing social and material differences between the sexes. The first strand we examine is Hakim's `preference theory', which has argued that persisting sex differences in employment patterns are an outcome of the `choices' made by different `types' of women. We next examine the claims of populist conservative feminism, that has argued that women (and men) in partnerships will be happier if they adopt a gender role traditionalism in the domestic sphere. Our empirical findings suggest that neither of these theoretical explanations are supported by our data, which is derived from the samples of six countries participating in the International Social Survey Programme Family 2002 module. Keywords: Gender essentialism; gender roles; traditionalism; employment; choices Introduction Differences between men and women are extensive, undeniable, and persisting. However, there is considerable disagreement as to their origins. Gender essentialists argue that the differences between the sexes are of an intrinsic nature, closely associated with physical, physiological, and/or spiritual differ- ences.2 In contrast, feminists have argued that although there are indeed biological sex differences between men and women, much if not most of the `difference' between men and women, as expressed in gender hierarchies and patterns of inequality, is in fact socially constructed. Feminists have always been rather wary of gender essentialism, and for a very simple reason. If it can be demonstrated that the differences between the sexes (which would include persisting patterns of inequality) rest upon the innate or essential qualities of women and men, then not only may these inequalities be rendered legitimate, but also, there can be little justification for change. In this paper, no attempt will be made to arbitrate between `nature' versus `nurture', or the claims of sociobiologists and social constructionists (see Rose and Rose 2000). Rather, we will critically examine two arguments, both resting on essentialist assumptions, relating to the gendered division of market and domestic work. The first is Hakim's development of `preference theory', that seeks to explain the persisting differences between men's and women's employment patterns, and the second is populist conservative feminism, that argues that women (and men) will be happier if gender stereotypical roles are assumed within the household and intimate relationships. It will be suggested that both of these arguments are flawed, and that both serve to legitimize and justify continuing inequalities in the gendered division of labour. Over the last century, one of the major social changes in modern societies has been the transformation in the social status and economic behaviour of women. By the closing decades of the twentieth century, women in modern democratic societies had largely gained civil and economic rights, and women's claims to equality have been largely accepted.3 Increasingly, women, including mothers of young children, are entering and remaining in the labour market, thus eroding, at least in part, the conventional `male breadwinner' model of employment and family life (inequalities in employment persist, however). These changes in the status of women have been taking place in parallel with other economic and political trends. In particular, at the political level, neo-liberalism, with its emphasis on the primacy of individual `choice', has become increasingly influential (e.g. Reaganism, Thatcherism). The expansion and acceptance of women's equality claims have not taken place without resistance ­ often described as a `backlash' against feminism (Faludi 1992). `Feminism' (usually unspecified) has been targeted as responsible for the increase in single parenthood and family dissolution (Morgan 1995). Massive social disruption, it is suggested, can only be averted if women somehow rediscover their innate nurturing capacities and devote themselves to the care of their young children, rather than seek success in the employment sphere (Fukuyama 1999; Kristol 1998). More insidious, perhaps, has been the emergence of populist conservative feminism.4 Popular works such as`Men are From Mars, Women are From Venus' (Gray 1992) emphasize the emotional and interpersonal differences between men and women, with women as empathic carers and men as competitive fixers. Equally populist texts such as `The Rules' (Fein and Schneider 1995) and `The Surrendered Wife' (Doyle 2001) urge gender traditionalism in intimate relationships as the road to gender harmony. A common theme in these diverse texts is a return to gender essentialism. Indeed, gender essentialist assumptions neatly resolve a potential logical 602 Rosemary Crompton and Clare Lyonette London School of Economics and Political Science 2005 contradiction for neo-liberal thinking. It might be argued that if all individuals are to be seen as free to make choices, then why should a particular category of individuals (i.e. women), be constrained in their choices (i.e. traditionalism in gender roles would limit `choice'). However, this contradiction is resolved if it is asserted that the differences between men and women are `natural' and that the choices made by women are in accordance with this `nature' and therefore not constrained by dominant (male) norms and/or inequalities of condition. Hakim and `preference theory' Hakim has developed what she describes as `preference theory' over a number of publications (Hakim 1998; 2000; 2003a; 2003b; 2004). She argues that the characteristic patterning of women's employment, as revealed by aggregate level statistics ­ a tendency to part-time and flexible working, and overrepresentation in lower-level occupations ­ is an outcome of the fact that there are three different `types' of women (or `preference groupings'). `Homecentred' women give priority to their families, and either withdraw from the labour market altogether or work only intermittently when they have children. `Work-centred' women, in contrast, give priority to their employment and are often not married and/or childless. A further category of `adaptives'5 ­ by far the largest ­ shift the emphasis of their `preferences' over their work/family life cycles. These work/family preference categories are also found amongst men, although in different proportions, as fewer men are `home-centred' or `adaptive'. Hakim argues that `preference theory' has been created as a consequence of the contraceptive revolution, the expansion of equal opportunities, and changes in the employment sphere that have generated flexible, white-collar jobs that are attractive to women (Hakim 2004: 14). As a consequence, `. . . a new scenario in which women have genuine choices' has been created and `. . . female heterogeneity is revealed to its full extent' (Hakim 2004: 14). It predicts that `men will retain their dominance in the labour market, politics and other competitive activities, because only a minority of women (i.e. the `work-centred') are prepared to prioritise their jobs . . . in the same way as men' (Hakim 2004: 15). She argues that Britain and the USA represent a particularly favourable `test case' for the application of preference theory (Hakim 2004: 19). In Britain and the USA, women have secured equal opportunities, and the flexible, deregulated labour markets in these countries (a consequence of neo-liberal economic policies), allow women to freely exercise their `choices' ­ particularly in respect of part-time work. Although Hakim's argument might justifiably be described as voluntarist with its emphasis on the overwhelming The new gender essentialism 603 London School of Economics and Political Science 2005 significance of individual `choices', allowance is made for the impact of structural constraints and factors. Opportunities for choice are more restricted in other countries (such as Spain, the major focus of Hakim's comparative work), particularly in respect of opportunities for part-time employment. Structural constraints, however, are inexorably becoming less important, and Hakim argues that as more `good' part-time work becomes available `. . . secondary earners who are currently forced to choose full-time jobs . . . will switch to part-time or temporary jobs instead' (Hakim 2003a: 262). Thus . . . the only cleavages that will matter within the workforce in the twentyfirst century will be the continuing differences between primary and secondary earners . . . Sex and gender will cease to be important factors and are already being replaced by lifestyle preferences as the only important differentiating characteristic in labour supply. (Hakim 2003a: 261) Moreover, Hakim claims that the three lifestyle preference groups she identifies `. . . cut across social class, education, and ability differences' (Hakim 2003a: 247). Hakim's work has generated a considerable, and sometimes acrimonious, debate, which will not be reviewed in its entirety here. Her emphasis on the primacy of `choice' over contextual constraint has been challenged (Ginn et al. 1996; McRae 2003; Procter and Padfield 1998). The stability of the `preference' categories she identifies, as well as the utility of `preferences' as an explanation of work orientations or `choices', has also been questioned (Crompton and Harris 1998). Our strategy in this paper will not be to attempt an exhaustive evaluation of `preference theory'.6 Indeed, we take the view that preferences and attitudes will often be linked to behaviours. As many others have argued, (McRae 2003: 329; Glover 2002) both normative and structural constraints shape women's decisions relating to the `balance' achieved by individual women in respect of market and caring work. We also take the view that neither identities nor behaviours in respect of women's employment are fixed, but adapt to each other in a process of positive feedback (Himmelweit and Sigala 2003: 23). `Preference theory' rests upon the identification of `preference categories' that vary in their distribution between men in aggregate and women in aggregate. Thus the explanation of these variations is key to the theory. Here Hakim draws upon the work of Goldberg (1993) who argues that male hormones such as testosterone are a major source of sex differences in motivation, ambition and behaviour (Hakim 2004: 4, see also Hakim 2000: 258­62). Thus men (in aggregate) are more aggressive and competitive than women in the world of employment, and `Women accept hierarchy so long as it is men who are in positions of power and authority. Male dominance is accepted, as Goldberg argued; female dominance goes contrary to sex-role stereotypes and is unwelcome, uncomfortable, and frequently rejected' (Hakim 2004: 110). Hakim 604 Rosemary Crompton and Clare Lyonette London School of Economics and Political Science 2005 argues that her theory is `unisex', and thus not biologically essentialist (Hakim 2004: 16) in that a minority of men are `home-centred' (as well as the majority of women). However, this argument is not persuasive. As biological gender essentialists are careful to argue, they do not claim that their theories identify characteristics that apply to `all women' or `all men' (Baron-Cohen 2004), but they do not seek to reject the `essentialist' label.7 The ISSP study Our analysis will draw upon a number of individual country data sets gathered within the International Social Survey Programme's (ISSP) Family 2002 module (Britain, Finland, France, Norway, USA and Portugal). In 2002, interviews were carried out with a stratified random sample of 1475 respondents in Norway, 2312 in Britain, 1353 in Finland, 1903 in France, 1171 in the USA and 1092 in Portugal. The same questions are asked in all countries. The common language of the ISSP is English, and questions are agreed and backtranslated in order to ensure cross-national comparability of meaning.8 The ISSP Family 2002 module asked a series of questions relating to gender role attitudes, attitudes to mother's employment, as well as satisfactions with family life and life in general. If the assumptions of `preference theory' are correct, then we might expect that attitudes to mothers' employment and its impact on children and family life would be the major determinant of women's employment behaviour ­ particularly in countries such as Britain. If the arguments of populist conservative feminism are valid, (that gender traditionalism in intimate relationships is the road to gender harmony), then we would anticipate that women involved in more traditional domestic relationships will be happier than those who are not. Data analysis We classified couple households according to their working hour strategies, using a modified version of the categories devised by Moen and her colleagues (Moen 2003: 19): 1) High commitment couples: both partners work more than 40 hours a week. 2) Dual moderate couples: both partners work between 35 and 40 hours a week. 3) Neo-traditionalist couples: the man works over 40 hours a week and the woman works less than 40 hours, characteristically part-time. 4) Alternate commitment couples: both partners work under 40 hours a week but one works less than 35 hours per week.9 The new gender essentialism 605 London School of Economics and Political Science 2005 5) Finally, we have included a further category, described here as traditional couples: the man works full-time and the woman is not in employment. In our classification, we have set `full-time' hours for `traditionals' at 35 (so as not to exclude male sole earners in those countries in which working hours are regulated), but the large majority (60 per cent) of men in category five work over 40 hours a week. Table I graphically illustrates the impact of national labour market regimes on couples' working arrangements. France, which has a statutory 35-hour working week, has the lowest proportion of `high commitment' couples, followed by Finland, where average hours of full-time work are also relatively short. The low proportions of `neo-traditionalists' (where the woman typically works part-time) in Finland and Portugal reflect both traditions of full-time work and the non-availability of part-time work for women in these countries (OECD 2001). Britain, which has the second highest level of part-time working amongst women in Europe as a whole, has the greatest proportion of respondents in `neo-traditional' households. Couples' working arrangements reported by respondents, therefore, are highly conditioned by national working time regimes (indeed, when country was included in a multinomial regression on couples' working arrangements, this explained most of the variance between individuals). National or country-specific factors, therefore, clearly have an important impact on couples' working arrangements. This fact is explicitly recognized by Hakim, who has, as we have seen, argued that the lack of external constraint in the case of Britain and the USA renders these countries important `test cases' for `preference theory'. However, national factors will not be the whole of the explanation for couples' working arrangements. Attitudes and 606 Rosemary Crompton and Clare Lyonette London School of Economics and Political Science 2005 TABLE I: Couples' work arrangements x country (men and women in couple relationships) Britain Finland France Norway USA (% Portugal Total (% of (% of (% of (% of of (% of (% of country) country) country) country) country) country) total) High 99 26 39 62 60 70 356 commitments (12.1%) (6.5%) (6.1%) (7.4%) (16.9%) (17%) (10.3%) Dual moderates 134 221 233 272 46 140 1,046 (16.4%) (55.1%) (36.2%) (32.4%) (13.0%) (34%) (30.2%) Neo-traditionalists 265 50 140 209 61 49 774 (32.4%) (12.5%) (21.8%) (24.9%) (17.2%) (11.9%) (22.3%) Alternate 156 79 143 237 71 39 725 commitments (19.1%) (19.7%) (22.2%) (28.2%) (20.0%) (9.5%) (20.9%) Traditionalists 163 25 88 59 117 114 566 (20%) (6.2%) (13.7%) (7%) (33.0%) (27.7%) (16.3%) Total 817 401 643 839 355 412 3,467 preferences will play a part, as will proximate factors such as the presence of children in the household, education and age. We constructed a measure of attitudes to women's employment, using five items from the ISSP dataset, previously used by Knudsen and Waerness (2001) in a comparative analysis of national context, individual characteristics and attitudes to women's employment in Britain, Sweden and Norway: A working mother can establish just as warm and secure a relationship with her children as a mother who does not work; A pre-school child is likely to suffer if his or her mother works; Family life suffers if a woman goes out to work; Work is alright, but what a woman really wants is a home and family; A man's job is to earn money, a woman's job is to look after the home and family Scores ranged from five to twenty-five, with lower scores indicating more positive attitudes towards women's employment.10 Table II shows that respondents in traditional male breadwinner couples had significantly more negative attitudes to women's employment. If we make the assumption that attitudes are an indication of `preferences', then the significantly greater attitudinal traditionalism of respondents with a traditional pattern of work-life articulation might be taken as a prima facie confirmation of Hakim's theory. However, Table III shows that a number of other factors also have an impact on couple work arrangements: respondents in traditional households were also more likely (on average) to be older, to have a child under sixteen in the household, to be less well educated, and more likely to be manual employees. These findings raise the thorny question of the determinants of attitudes, as well as their consistency over time. The new gender essentialism 607 London School of Economics and Political Science 2005 TABLE II: Attitudes to women's employment x couples' working arrangements (all countries, men and women) Working arrangements N Mean Std. Deviation High commitments 328 12.53 4.77 Dual moderates 945 12.33 4.37 Neo-traditionalists 669 12.92 4.08 Alternate commitments 653 12.93 4.35 Traditionalists 503 15.17 4.54 Total 3,097 13.06 4.48 Note: ANOVA F = 37.142; d.f. = 4; p < 0.001. Post-hoc tests showed that traditionalists had significantly more negative attitudes towards women's employment than all other groups. Which factors affect attitudes? In the same way that couples' working arrangements are influenced by certain factors (including attitudes), attitudes may also be influenced by contextual factors, including couples' working arrangements. For example, Himmelweit and Sigala (2003: 20­1) showed that women in the UK who were in the contradictory position of being in employment but believing that pre-school children suffered as a consequence were more likely to change their attitudes than their behaviour (i.e. giving up paid work) over a two-year period. In our analysis, in order to explore the factors associated with variations in attitudes to mother's employment,a multiple linear regression analysis was first performed with attitudes to women's employment as the dependent variable. Factors included as independent variables in two blocks were: couples' working arrangements, respondent's age, education level, whether or not there was a child in the household and whether or not the respondent's mother worked. Only women were included in the regression, as it was specifically women's situations within the different couples' working arrangements which were of interest here, and it was expected (and has previously been shown) that men's and women's attitudes towards female employment are very different (Knudsen and Waerness 2001), with women significantly more positive overall than men.11 The results of the regression analysis show that being less well-educated, older, having traditional working arrangements and having a mother who did not work were the most significant predictors of negative attitudes to women's employment, explaining around 15 per cent of the variance in attitudes (adjusted r2 = 0.148). Dual moderate women were more likely than neotraditional and traditional women to have positive attitudes to women's 608 Rosemary Crompton and Clare Lyonette London School of Economics and Political Science 2005 TABLE III: Couples' working arrangements by mean age, level of education, social class, child in household (all countries, men and women) Variables High Dual Neo- Alternate Traditionalist commitment moderates traditionalists commitments Age (mean) 44.03 41.69 43.27 42.69 44.31 Education (%): Low 16.9% 14.8% 14.7% 14.3% 28.4% Middle 36.6% 51.1% 47.6% 49.4% 38.4% High 46.5% 34.1% 37.7% 36.2% 33.2% Child (%): Yes 37.7% 43.0% 50.9% 50.1% 59.0% No 62.3% 57.0% 49.1% 49.9% 41.0% Social class (%): Professional 61.0% 41.7% 43.7% 41.2% 41.2% Intermediate 18.2% 31.7% 27.9% 33.6% 24.2% Manual 20.8% 26.6% 28.4% 25.3% 34.6% employment, that is, as Himmelweit and Sigala's findings suggest, employment status will itself have an impact on attitudes. Having a child in the household did not predict women's attitudes. Separate regressions for each couples' working arrangement groups were also performed: having a child in the household was not predictive of attitudes for women in any of the working arrangement groups, whereas education level was consistently predictive. Which factors affect couples' working arrangements? We then attempted to unravel the factors having an influence on couples' working arrangements via a multinomial regression analysis, with working arrangements as the dependent variable. In this instance, attitudes to women's employment was included as a potential predictor of couples' working arrangements, along with other attitudinal variables, including attitudes to men's involvement in household tasks, using two items from the ISSP data: Men ought to do a larger share of household work that they do now; Men ought to do a larger share of childcare than they do now. Scores ranged from two to ten, with higher scores indicating more traditional attitudes towards men's involvement in household tasks (i.e. that they should not do more).12 If couples' working arrangements are indeed shaped by women's home-centred or work-centred `nature', it may be expected that those women with more traditional working arrangements (i.e. man works, The new gender essentialism 609 London School of Economics and Political Science 2005 TABLE IV: Multiple regression on attitudes to women's employment by couples' working arrangements, age, education, whether mother worked and child in household (women only, all countries, final model) Variable B Beta t-value (Constant) 9.113 17.124*** High commitments -0.314 -0.022 -0.805 Dual moderates -0.695 -0.072 -2.380* Alternate commitments -0.233 -0.022 -0.753 Traditionalists 2.184 0.177 6.319*** Education 1.819 0.199 8.512*** Mother worked -0.833 -0.093 -3.846*** Child in household -0.023 -0.003 -0.107 Age 0.064 0.181 6.998*** Notes: *p < 0.05; ***p < 0.001. Reference category = neo-traditionalists (man works full-time, wife works part-time). Education: low = 1; high = 0. Mother worked when respondent under 14: yes = 1; no = 0. Child in household: yes = 1; no = 0. Couples' working arrangement groups were dummy coded, e.g. high commitments = 1; other = 0, etc. woman stays at home) will be less likely to think that men should do more in the home. Given the importance that Hakim (2003a; 2003b) attaches to preferences in determining individual attitudes to employment, we also include data on attitudes to career development,13 as `work-centred' women should be more likely to agree that it is important to move up the job ladder than `homecentred' women. Other independent variables included whether or not the respondent's own mother had worked when the respondent was under fourteen, as this had an impact on attitudes to women's employment. Our reference category is neo-traditional households, where the man works fulltime and the woman typically works part-time (when the traditional, `male breadwinner' group was used as the reference category, all groups were significantly different from traditionalists, which masked some of the more interesting variations within the working arrangement groups). Again, only women were included in the regression analysis. Having a child in the household was not a significant predictor of attitudes to women's employment for female respondents in all categories (Table IV), but it was highly predictive of couples' working arrangements. All of the categories of women (except women in alternate couple arrangements) were significantly less likely than traditional women to have a child in the home. Neo-traditional women were the next most likely to have a child in the home and high commitment women the least likely, adding support to the claim that the presence of children in the home affects women's working arrangements. Although having a mother who worked was a significant predictor of different attitudes among women (Table IV), this was not a direct predictor of work behaviour. Education was consistently predictive of both attitudes to women's employment and actual working arrangements, with traditional women having a significantly lower level of education than all other women, and high commitment women having a higher level of education. Contrary to Hakim's argument that lifestyle preferences `. . . cut across social class, education, and ability differences' (2003a: 247), in aggregate, couples' working arrangements do appear to be affected by women's level of education. As those with higher education are also more likely to be in higher social classes (see Table III), her argument has little support from the data presented here. Age also had an impact, with dual moderate women significantly more likely to be in the younger age range than other women, especially traditional women. There were no significant effects of career aspirations and attitudes to men's involvement in the home. Although traditional women were significantly more negative in attitudes to women's employment than all other groups in which women were in employment, there were no significant differences in attitude between the `working women'. On the basis of this evidence, it would seem, therefore, that `structural' factors are at least as important, if not more important, than `attitudinal' factors in shaping the working arrangements of couples. 610 Rosemary Crompton and Clare Lyonette London School of Economics and Political Science 2005 Nevertheless, it might still be argued that our data simply reports on the impact of constraints,rather than on the operation of`preferences'.As we have seen, Hakim has argued that Britain stands out as a country where women can exercise `choice', given the extensive opportunities for part-time work in this country. We might expect, therefore, that preferences (as expressed in attitudes) might be of particular significance in the British case. Repeating the regression in Table V for separate countries was problematic given that the The new gender essentialism 611 London School of Economics and Political Science 2005 TABLE V: Multinomial regression to identify the determinants of couples' working arrangements (all countries, women only) High Dual Traditionalists Alternate commitments moderates commitments Exp (B) Exp (B) Exp (B) Exp (B) Age of respondent: 18­34 0.860 2.956*** 0.785 1.536 35­54 1.282 2.021** 0.673 1.256 55+ ­ ­ ­ Attitudes to women's employment: More positive 1.502 1.413 0.320*** 1.270 Middle range 0.969 1.094 0.657 1.188 More negative ­ ­ ­ Attitudes to men's involvement in the home: Men should do more 1.267 1.156 1.527 0.918 (+) Men should do more (-) ­ ­ ­ Importance of moving up the job ladder: Agree 1.534 0.869 1.233 0.918 Neither agree nor 1.465 1.031 1.584 0.740 disagree Disagree ­ ­ ­ Respondent's education level: Lowest or no 0.939 1.512 3.127*** 1.628 qualification Above lowest 0.456** 1.371 0.721 1.162 qualification Above higher ­ ­ ­ secondary or university degree completed Child(ren) in household: Yes 0.510** 0.667* 3.110*** 0.930 No ­ ­ ­ Mother worked when respondent under 14: Yes 1.349 0.946 1.017 1.016 No ­ ­ ­ ­ Note: p < 0.05; **p < 0.01; ***p < 0.001. Nagelkerke R2 = 0.146. Reference category is neo- traditionalists. low numbers involved meant that we could not be confident of model fit. Nevertheless, repeating the multinomial regression (not reported here) for British women did not reveal a pattern that varied from that reported in Table V; that is, having a child in the household, and level of education, were significant determinants of couples' working arrangements, as well as attitudes to women's employment. In short,the British case did not emerge as one in which attitudes were of particular importance in affecting couples' work arrangements. Without current longitudinal data, it is impossible to conclusively determine whether attitudes affect work behaviour or whether work behaviour, perhaps as a consequence of structural constraints, affects attitudes ­ although as we have seen, Himmelweit and Sigala's longitudinal analysis of the British case suggests that work behaviour has a greater impact than attitudes. Our data so far suggest that attitudes and behaviour act in a bi-directional manner, with both influencing, and being influenced by, the other. Those women who stay at home (traditionalists) may be more likely to have negative attitudes to women's employment precisely because they stay at home, rather than viceversa. As traditional women have lower levels of education than working women, meaning that their opportunities for well-paid employment are more limited, the low provision of affordable childcare in Britain is more likely to act as a constraint upon employment (see McRae 2003). We would argue, therefore, that it is unlikely that couples' working arrangements are `chosen' simply in accordance with women's attitudes, but are more likely to be the result of a complex interplay of attitudes and practical constraints. With these kinds of arguments in mind, it is instructive to examine the association between attitudes and couples' working arrangements for another of the countries included in our data set, Norway. As we have seen (Table I), there are ample part-time employment opportunities available in Norway (in 2002, 43 per cent of employed Norwegian women were classified as part-time workers). Moreover, as a `Nordic' welfare state, Norway is also characterized by the availability of low-cost, high quality childcare provision as well as general supports for families such as extensive maternity and paternity leaves (Ellingsaeter 2003; Korpi 2000).14 In short, it could be argued that in Norway parents are able to make even more unfettered `choices' relating to parental working arrangements than in Britain. If the assumptions of preference theory are correct, we might reasonably expect that attitudes and preferences would play an important role in making these `choices'.15 However, as Table VI demonstrates, for Norwegian women, there was no relationship between attitudes to women's employment and couples' working arrangements.16 This suggests that in Norway, a country in which there is extensive capacity to `choose' arrangements relating to employment and family life, individual attitudes do not appear to play a substantial part in the choices that are made. Other factors are obviously more important. Indeed, a regression analysis showed that the variations in Norwegian women's attitudes to employment were related to 612 Rosemary Crompton and Clare Lyonette London School of Economics and Political Science 2005 education (t = 5.788; p < 0.001) and age (t = 3.895; p < 0.001), rather than couples' working arrangements.17 We have not carried out an exhaustive evaluation of `preference theory', which would require longitudinal data (see McRae 2003). However, to the extent that preference theory would suggest a major role for attitudinal factors in shaping women's employment decisions, particularly in countries such as Britain, our data suggests otherwise.18 We do not dispute that preferences and related attitudes will shape individual behaviour, or that particular sets of attitudes to work and family may be found (albeit to varying extents) across a range of social positions and education levels. However, it would seem unwise to assume that women's employment behaviours may be primarily accounted for by the hypothesized existence of different `types' of women, as `preference theory' suggests. Gender role attitudes and reported levels of happiness amongst men and women We next address the question of whether traditionalism in domestic and gender roles is associated with greater happiness, particularly family happiness, for men and women. As discussed in the introduction to this paper, populist conservative feminism, which is rooted in essentialist assumptions about `natural' differences between men and women, has argued that women will achieve greater happiness and life satisfaction if they assume a traditional role in relation to men, particularly in the domestic sphere. Life and family happiness were measured using three questions from the ISSP survey:19 If you were to consider your life in general, how happy or unhappy would you say you are, on the whole; All things considered, how satisfied are you with your family life; My life at home is rarely stressful. The new gender essentialism 613 London School of Economics and Political Science 2005 TABLE VI: Attitudes to women's employment x couples' working arrangements (Norwegian women) Working arrangements N Mean Std. Deviation High commitments 40 10.83 3.83 Dual moderates 127 11.54 4.11 Neo-traditionalists 89 11.97 4.49 Alternate commitments 117 11.66 3.86 Traditionalists 28 10.57 3.99 Total 401 11.53 4.09 Note: ANOVA F = ns. There were country-specific variations in the answers given to these questions, but they did not follow a consistent pattern. American and British men were more likely to say that they were `completely' or `very' happy with life in general than men in the other four countries, as were American and British women. However, Finnish and French men were more likely than men in the other countries to say that life at home was rarely stressful, while women in Britain, the USA and Portugal were more likely to find their home lives stressful. British, American and Norwegian men were more likely to say that they were `completely' or `very' satisfied with family life than men in other countries, while British and American women were also more likely to say that they were `completely' or `very' satisfied with family life. We also examined, using the same three questions, levels of reported happiness between men and women in the different couple working arrangements described above.Again,no consistent picture emerged. Men and women in traditional and neo-traditional household work arrangements were more likely to say that they were `completely or very' happy with life in general. There were no significant differences found for women as far as the level of general satisfaction with family life was concerned, although men in traditional household working arrangements were somewhat more likely to say that they were `completely or very' satisfied with family life. However, on the third question ­ life at home is rarely stressful ­ dual moderate men were significantly more likely than other men to agree, but women with traditional working arrangements were significantly less likely to agree than women in the other categories. It would seem, therefore, that traditionalism in working arrangements might possibly bring some marginal advantages as far as men are concerned, but there is little evidence that gender traditionalism in the balance of domestic and paid work is associated with greater family happiness for women. One possibility might be that a contradiction between attitudes and practice in relation to domestic work might be a source of dissatisfaction. For example, an individual might have `liberal' or `non-traditional' gender role attitudes, but be involved in a rather traditional domestic division of labour, and this might be a source of resentment. This possibility was explored by developing a combined measure of gender role attitudes and the domestic division of labour. Responses relating to gender role attitudes (using the single question:`A man's job is to make money,a woman's job is to look after the home and family') were dichotomized and cross-tabulated with a dichotomized version of a task-based scale describing reported levels of gender traditionalism in the domestic division of labour.20 This generated four categories which were combined into three as follows: (1) `congruent liberals', those respondents with more liberal gender role attitudes, and a less traditional division of domestic labour; (2) `inconsistents' ­ a combination of those with more liberal gender role attitudes and a more traditional division of domestic labour, as well as those with less liberal gender role attitudes and a less traditional division of domestic labour, and (3) 614 Rosemary Crompton and Clare Lyonette London School of Economics and Political Science 2005 `congruent traditionals', those respondents with less liberal gender role attitudes and a more traditional division of domestic labour. These congruence categories bring attitudes and behaviour together in a single variable. Table VII summarizes the measures of life and family happiness by congruence categories, separately for men and women. It can be seen that for women, but not for men, a combination of gender role liberalism together with a less traditional allocation of domestic work is systematically associated with enhanced life happiness, satisfaction with family and lower stress at home. Congruent traditionalists (or gender conservatives) report lower levels of `happiness' on all three questions. It would seem, therefore, that contrary to the assertions of feminist conservatives, for women, domestic gender traditionalism is not associated with an enhanced level of personal and family happiness. Repeating the analysis on an individual country basis revealed that in all countries, it was the women with the least traditionalism in their domestic lives (i.e., the congruent liberals) who reported the highest levels of personal and family happiness.21 It is not being claimed that the data available from the ISSP surveys has facilitated an exhaustive exploration of the links between traditionalism in gender relations and personal happiness for women. Nevertheless, the evidence we do have suggests that for women (but not men), gender traditionalism in attitudes and/or practice (as indicated by gender role attitudes and the domestic division of labour) is not associated with greater general or family contentment and happiness, but indeed, the opposite. There is some slight indication that men find a traditional division of domestic labour more agreeable in that those men with non-employed partners are somewhat more likely to say that they are `completely' or `very' happy with life in general, but this The new gender essentialism 615 London School of Economics and Political Science 2005 TABLE VII: Life and family happiness for men (M) and women (F) by congruence categories (6country pooled sample, percentages, respondents in partnerships only) Congruent Inconsistent Congruent Total liberal traditional `Completely' or `very' M 48% 48% 49% 48% happy with life in F 57% 45% 48% 48% general (1) `Completely' or `very' M 64% 63% 64% 64% satisfied with family F 71% 55% 56% 58% life (2) `Strongly agree' or M 58% 59% 61% 59% `agree' that life at F 53% 43% 45% 45% home is rarely stressful (3) Notes: (1) Men: n.s., women: c2 = 47.909; d.f. = 6; p < 0.001. (2) Men: n.s., women: c2 = 55.666; d.f. = 6; p < 0.001. (3) Men: n.s., women: c2 = 21.472; d.f. = 4; p < 0.001. evidence is by no means conclusive. There can be no doubt that the considerable shifts in gender roles and the relationships between the sexes that have been taking place over the last half-century will have brought about considerable upheavals in the personal lives of many women and men. However, it would seem to be misleading, as gender conservatives have argued, to suggest that these transformations have brought with them any greater unhappiness for women. Discussion and conclusions In this paper, we have empirically explored Hakim's arguments that women's employment patterns may, in aggregate, be seen as an outcome of their choices. We have seen that patterns of couples' working arrangements would seem to be shaped by structural, as well as, attitudinal factors ­ even in Britain, a country that Hakim (2004: 19) has described as a `theoretically important social context' for the study of women's employment given that `. . . there are no major constraints limiting choice or forcing choice in particular directions'. Indeed, in the case of Norway, a country where, it might be argued, structural constraints relating to mothers' employment decisions are relatively minor, individual variations in (women's) attitudes to women's employment do not correlate at all with couples' working arrangements. As noted in the Introduction to this paper, gender essentialism has many affinities with conservative and/or neo-liberal thinking.22 More particularly, conservative feminism (of all kinds) serves to undermine the legitimacy of women's claims to equality, as the unequal status quo is thereby rendered `natural'. Indeed, traditional explanations of inequality have always had recourse to `naturalism': It is thus clear that there are by nature free men and slaves, and that servitude is agreeable and just for the latter . . . Equally, the relation of the male to the female is by nature such that one is superior and the other inferior, one dominates and the other is dominated . . . (Aristotle, cited in Dahrendorf 1969) As Dahrendorf argues, this kind of argument renders impossible a sociological treatment of inequality, as that which exists `by nature' does not have to be explained or challenged. Similarly, feminists have long argued that differences and inequalities between men and women do not exist `by nature', but are socially constructed. As we have seen, gender conservatism re-affirms the `naturalness' of gender differences, thus the inequalities which arise as a consequence of these differences are themselves legitimized. In taking a stance against gender essentialism, we do not intend to advocate a return to debates relating to gender `equality versus difference', which have 616 Rosemary Crompton and Clare Lyonette London School of Economics and Political Science 2005 The new gender essentialism 617 London School of Economics and Political Science 2005 taken up so much space in feminist discussions (see Bock and James 1992). As Fraser (2000) has argued, it is necessary to move beyond these polarized positions and reconstruct both equality and difference within a framework of diversity in which `participatory parity' can be achieved. This would mean the . . . creation of conditions that facilitate the meeting of human need and the exercise of caring responsibilities in such a way as to ensure that all individuals can develop and flourish as citizens. In this way difference is incorporated into strategies for gender equity without reference to potentially essentialist notions of women's qualities and nature. (Lister 2002) In a similar vein, Williams (2000: 272) argues: It is time to admit that women as a group do not perform the same as men as a group when jobs are designed around an ideal worker with men's physique and/or men's access to a flow of family work most women do not enjoy. Once we invent a language that defines this situation as the result of discrimination against women, rather than mothers' choice, we can face the facts and make new demands to restructure work. (emphasis in original). Nevertheless, although we are emphatic in our desire not to return to the polarized positions of earlier debate,we would suggest that our findings (which we would agree cannot be conclusive) of a greater degree of domestic and family happiness amongst women professing liberal gender role attitudes and involved in a less traditional division of domestic labour merits further consideration. With this in mind, and in the light of the issues raised in our previous paragraph, we will conclude with a brief reference to another of our empirical findings. The ISSP survey included a series of questions on work-life conflict that generated a robust scale (see Crompton and Lyonette, in press). We found that `congruent liberals' (that is, men and women with liberal gender role attitudes and a less traditional division of domestic labour) reported significantly lower levels of work-life conflict than either `congruent traditional' or `inconsistent' respondents. This suggests that a sustained rethinking of the division of labour between men and women, in both the domestic and market spheres, might result in improvements for both men and women. (Date accepted: August 2005) 2. Gender essentialism may be biological, arguing that `maleness' and `femaleness' derive from the impact of different levels of sex hormones, particularly in early development. See Goldberg 1993, Baron-Cohen 2004. These biological differences are held Notes 1. We would like to acknowledge ESRC support for two projects that enabled this research to be carried out:`Employment and the Family' R000239727 and `Families, Employment and work-life integration in Britain and Europe' RES 000220106. 618 Rosemary Crompton and Clare Lyonette London School of Economics and Political Science 2005 to account for behavioural differences between men and women. Psychological versions of gender essentialism locate gender differences in psychosexual development. See Chodorow 1989. Women have also been argued to be inherently more moral and spiritual than men, see Williams 1991. Biology and spirituality may often be employed in combination, see Footnote 4 below. 3. The distinction between `traditional' and `modern' societies may seem anachronistic and value-laden. It is used here simply in order to make a distinction between those societies (usually, but not always, democracies) in which the equal rights of women are at least formally accepted as legitimate, and theocracies or semi-feudal `states' such as for example ­ Saudi Arabia or Afghanistan. 4. These sentiments, of course, are not confined to populist conservative feminism. For example, the `Letter to the bishops of the Catholic Church on the collaboration of men and women in the church and in the world' (Offices of the Congregation for the doctrine of the Faith, May 31, 2004) asserts that women, `in her deepest and original being, exists "for the other"', linked to their `physical capacity to give life'. Furthermore, women live the `dispositions of listening, welcoming, humility, faithfulness, praise and waiting' with `particular intensity and naturalness'. 5. Hakim originally identified this group as `drifters', but they have undergone a change of name (and assumed considerably more importance) as she has developed her argument. 6. Hakim (2003b) would argue that we would not be in a position to do this in any case, given that we will be relying on generalized `attitude' questions rather than direct questions relating to personal `preferences' in our analysis. 7. `I would weep with disappointment if a reader took home from this book the message that `all men have lower empathy' or `all women have lower systematising skills'. Hopefully, I have made clear that when we talk about the female brain or the male brain, these terms are short for psychological profiles based upon the average scores obtained when testing women as a group, or the average scores obtained when testing men as a group' (Baron-Cohen 2004 183). 8. For a description of the ISSP programme, see Davis and Jowell 1989, also Jowell, Brook and Dowds 1993, also Jowell 1998. Methods of data collection vary amongst the participant countries, and include face-to-face interviews as well as telephone and postal surveys. For a full description of individual country surveys, see www.issp.org. A comparative analysis is not a major objective of this paper (for a discussion of the impact of societal factors, see Crompton and Lyonette, in press), and national differences will only be featured as and when they contribute to the argument that we develop. Our thanks to Gunn Birkelund, Yannick Lemel, Karin Wall and Anneli Anttonen for access to ISSP national datasets. 9. Moen's classification included a further category of`crossover' couple arrangements, in which the woman took the `breadwinner' role and worked long hours. As this category was (a) very small and (b) could not reasonably be grouped with any of the other categories, we have omitted it from our analysis. 10. Factor analysis for all countries combined showed one factor with an Eigen value of 2.69, explaining 54% of the variance. All five items loaded on the single factor >0.66. A reliability analysis recorded a Cronbach's alpha of 0.78. Separate country analyses also demonstrated good reliability. 11. It is not possible to present separate regressions for women in each country due to the small numbers in some groups. 12. Factor analysis showed one factor with an Eigen value of 1.64, explaining over 82% of the variance. A Cronbach's alpha of 0.78 demonstrated good reliability of the two-item scale. 13. The question was the level of agreement (or disagreement) expressed in relation to the statement: `It is important to move up the job ladder at work, even if this The new gender essentialism 619 London School of Economics and Political Science 2005 gets in the way of family life'. This question was not asked in the USA, which is not included in the regression. 14. Hakim (2004: 78) would argue that the extent of childcare services do not affect women's employment levels. 15. It may be noted that Hakim (2004: 20) argues that Scandinavian societies are too homogeneous in `race, religion and culture' to provide lessons for other countries. We do not hold to this view. 16. There was also no relationship between couple work arrangements and attitudes to women's employment when men were included. Couple arrangements in Norway were also evaluated against gender role attitudes, and again, there was no association. 17. These findings support those of Knudsen and Waerness (2001) who found large differences in age and education in Norwegian attitudes to women's employment, due to rapid changes in women's employment and family policy, compared with Britain and Sweden. 18. Other authors have suggested that `gendered moral rationalities' are significant in determining the employment behaviour of different groups of women. However, the origins of these different rationalities are associated with class differences (Duncan 2005). As we have seen, Hakim rejects this argument. 19. The topic of happiness has recently received increased academic attention see, for example Layard 2005. Layard employs quantitative measures of `happiness' similar to those we employ here. He does not, however, explore gender differences apart from noting that `happiness levels' are rather similar as between men and women. 20. DDL made up of 5 items: In your household, who usually does the: laundry; cares for sick family members; shops for groceries; household cleaning; prepares the meals (always me, usually me, about equal, usually spouse/partner, always spouse/partner). Scores ranging from 5­25; higher scores indicate more traditional DDL. Factor analysis showed 1 factor, Eigen value 2.699, explaining over 54% of the variance. Cronbach's alpha for all 5 items = 0.786. 21. An analysis by country resulted in some rather small cell sizes ­ for example, there are low numbers of `congruent traditional' women in Finland and Norway. The difference between the `congruent liberal' women and the other two categories was not uniformly statistically significant by country, but nevertheless, their happiness scores were uniformly higher. 22. Hakim's work has been enthusiastically taken up by right-leaning media and politicians. For example, she has been invited to advise the Australian (right wing) government (see B. Arndt, Sydney Morning Herald Feb. 7 2003), which has introduced a number of policies designed to encourage women to stay at home with their children (for example, the introduction of a `baby bonus' for women who stay at home). Bibliography Baron-Cohen, S. 2004 The Essential Difference, Penguin: Harmondsworth, Middlesex. Bock, G. and James, S. 1992 Beyond Equality and Difference: Citizenship, Feminist Politics, and Female Subjectivity. London; New York: Routledge. Chodorow, N. 1989 Psychoanalysis and Feminism, Yale University Press: New Haven and London. Crompton, R. and Harris, F. 1998 `Explaining Women's Employment Patterns: `Orientations to Work' Revisited', British Journal of Sociology 49(1): 118­36. Crompton, R. and Lyonette, C. (in press) `Work-life "Balance" in Britain and Europe'. Dahrendorf, R. 1969 `On the Origin of Inequality Among Men' in A. Beteille (ed.) 620 Rosemary Crompton and Clare Lyonette London School of Economics and Political Science 2005 Social Inequality. Penguin: Harmondsworth, Middlesex. Davis, J.A. and Jowell, R. 1989 `Measuring National Differences' in R. Jowell, S. Witherspoon and L. Brook. British Social Attitudes: Special International Report, Gower: Aldershot. Doyle, L. 2001 The Surrendered Wife: a Practical Guide For Finding Intimacy, Passion and Peace With a Man, Australia: Simon and Schuster. Duncan, S. 2005 `Mothering, Class and Rationality', Sociological Review 53(2): 2. Ellingsaeter, A.L. 2003 `The Complexity of Family Policy Reform: the Case of Norway'. European Societies 5(4): 419­43. Faludi, S. 1992 Backlash, Vintage: London. Fein, E. and Schneider, S. 1995 The Rules: Time-Tested Secrets for Capturing the Heart of Mr. Right, Warner Books. Fraser, N. 2000 `Rethinking Recognition', New Left Review, May/June: 107­20 Fukuyama, F. 1999 The Great Disruption, Profile Books, London. Ginn, J., Arber, S., Brannen, J., Dale, A., Dex, S., Elias, P., Moss, P., Pahl, J., Roberts, C. and Rubery, J. 1996 `Feminist Fallacies: a Reply to Hakim on Women's Employment', British Journal of Sociology 7(1): 167­74. Glover, J. 2002 `The Balance Model: Theorising Women's Employment Behaviour' in A. Carling, S. Duncan and R. Edwards (eds) Analysing Families, Routledge: London. Goldberg, S. 1993 Why Men Rule: A Theory of Male Dominance, Open Court: Chicago. Gray, J. 1992 Men are from Mars,Women are from Venus: How to Get What You Want in Your Relationships, Harper Collins. Hakim, C. 1998 Social Change and Innovation in the Labour Market, Oxford University Press: Oxford. Hakim, C. 2000 Work-lifestyle Choices in the 21st Century, Oxford University Press: Oxford. Hakim, C. 2003a Models of the Family in Modern Societies, Ashgate: Aldershot. Hakim, C. 2003b `Public Morality Versus Personal Choice', British Journal of Sociology 53(3): 339­46. Hakim, C. 2004 Key Issues in Women'sWork, GlassHouse Press: London. Himmelweit, S. and Sigala, M. 2003 `Internal and External Constraints on Mothers' Employment', Working paper no 27, ESRC Future of Work Programme. Jowell, R. 1998 `How Comparative is Comparative Research?', American Behavioral Scientist 42(3): 168­77. Jowell, R. Brook, L. and Dowds, L. 1993 International Social Attitudes, Dartmouth: Aldershot. Knudsen, K. and Waerness, K. 2001 `National Context, Individual Characteristics and Attitudes on Mothers' Employment: a Comparative Analysis of Great Britain, Sweden and Norway', Acta Sociologica 44(1): 67­79. Korpi, W. 2000 `Faces of Inequality: Gender, Class and Patterns of Inequalities in Different Types of Welfare States', Social Politics 7: 127­91. Kristol,W. 1998 `A Conservative Perspective on Public Policy and the Family' in C. Wolfe (ed.) The Family, Civil Society, and the State', Rowman and Littlefield: Oxford. Layard, R. 2005 Happiness: Lessons From a New Science, Allen Lane, Penguin: London. Lister, R. 2002 Citizenship: Feminist Perspectives (2nd ed.), Palgrave Macmillan: Basingstoke. McRae, S. 2003 Constraints and Choices in Mothers' Employment Careers, British Journal of Sociology 53(3): 317­38. Moen, P. (ed.) 2003 It's About Time: Couples and Careers, ILR/Cornell University Press: Ithaca and London. Morgan, P. 1995 Farewell to the Family?, IEA Health and Welfare Unit, London. OECD 2001 `Balancing Work and Family Life: Helping Parents Into Paid Employment', Chapter 4, Employment Outlook. Procter, I. and Padfield, M. 1998 Young Adult Women, Work, and Family: Living a Contradiction, Mansell: London and Washington, D.C. Rose, H. and Rose, S. (eds) 2000 Alas, Poor Darwin, Harmony Books: New York. Williams, J.C. 1991 `Domesticity as the Dangerous Supplement of Liberalism', Journal of Women's History 2(3): 69­88. Williams, J. 2000 Unbending Gender,Oxford University Press: Oxford.