The gender(ed) division of labour in Europe: patterns of practices in 18 EU countries

Drawing on cross-national data from ISSP 2012, this paper aims to identify and characterize the patterns of gender division of labour in European families with children. Analysis focuses on gender role practices at country level, assuming that welfare regimes frame the gendered allocation of time to paid work, care work and household work. From a cluster analysis based on time-use (weekly hours and sex asymmetries) in 18 EU countries, six patterns are identified. Findings disclose the relevance of disentangling care work from household work in order to understand in more detail the diversity of patterns across Europe. sur la comparaison des pratiques utilisées dans chaque pays. L’analyse de clusters a identifié six modèles d’emploi du temps (heures par semaine et les asymétries entre les sexes) distincts. Les résultats soulignent l’importance de séparer le travail domestique du travail familial, afin de mieux comprendre la diversité trouvée. artículo objeto identificar patrones de del trabajo familias países de la (ISSP protección asignación de tiempo el trabajo pagado, de cuidar y doméstico de hombres y mujeres, estudio aquí presentado en la comparación de las prácticas instituidas en cada país. El análisis de clusters permitió identificar seis patrones de uso del tiempo (horas semanales y asimetrías entre hombres y mujeres) distintos. Los resultados evidencian la importancia de separar el trabajo doméstico del trabajo de cuidar, para comprender mejor la diversidad encontrada.


Introduction 1
Work-life balance has undergone constant change since the 1950s, at the dawn of the post-War makeover of Western societies. Social scientists have tracked these developments closely and have offered a wide range of interpretations for the diversity of paths and pace of change in different countries (e.g. Cousins, 1999;Crompton, 1999;Crompton, Lewis and Lyonette, 2007;Esping-Andersen, 1990). There are at least four consensual assumptions on this subject. The first is that the foundation of modern welfare states, in the context of post-War economic prosperity, was complicit with the social prominence of the male breadwinner/female homemaker model, which established a work-life balance based on distinctive gender roles in paid and unpaid work. The second is that the weakening of this model is substantially related to the "female revolution" (Esping-Andersen, 2009), i.e. women's increasing participation in paid work and quest for economic independence and full civil rights from the 1960s onwards (Aboim, 2010;Lewis and Lewis, 1996;Lewis, 2001). The third is that developments towards the dual earner/ dual carer family model (Gornick and Meyers, 2003) are making work-life balance in couples with young children less gendered, even if with national and cross-national specificities. The final assumption is that, despite overall changes, men and women in all European countries are far from spending equal amounts of time in paid and unpaid work, since men still have a higher share of the former, while women have a higher share of the latter (Amâncio, 2007;Perista et al., 2016;Wall et al., 2017). 2 On the other hand, the resilience of gender stereotypes and gender role essentialism is deeply ingrained in gender cultures (Aboim, 2010;Gaunt, 2006;Pfau-Effinger, 1998), giving rise to tensions between the ideals of gender equality and notions of male and female roles, which are expressed at the individual level of beliefs, attitudes and practices regarding the division of labour, but also at the institutional level of daily human existence. In this context, the type of welfare state plays a major role, not only by conveying, to a greater or lesser extent, an ideal model of division of labour at a given moment, but also by offering a framework of more or less consistent objective conditions for the work-life balance of couples with small children.

3
Drawing on cross-national data from ISSP 2012 "Family and Changing Gender Roles", this article aims to explore the current gender division of labour in 18 EU countries, taking into account the time allocated by men and women to paid work, care work and household work and respective asymmetries. By selecting a subsample of partnered men and women at working ages (18-64 years old) and with dependent children (up to 18 years old), and by applying a country-cluster analysis of nine indicators of gender role practices and asymmetries expressed in mean weekly hours, we came up with six patterns of gender division of labour in families with children. Subsequently, we characterized the patterns by analysing indicators of male and female practices in paid and unpaid work, as well as some indicators of gender role attitudes, in order to apprehend how changing practices are influencing attitudes or contradicting them.
Theoretical framework 4 The sexual division of labour, along with the assumption that gender qualifies men and women for specific roles, has always been at the core of the social organization of contemporary societies, framing their relationship to paid and unpaid work. Indeed, in the 1950s, in the context of post-War economic prosperity, this state of affairs supported the functionalist family model characterized by a complementary specialization of gender roles that was understood as the most suitable for the purpose of family functioning, as well as for its relationship to the social system, which was ensured mainly by men (Parsons and Bales, 1955). The male breadwinner/female homemaker family model, which became popular with the middle classes in the US and in western European countries that experienced post-war recovery and growth, was supported by the labour market and the state: on the one hand, Fordism, a modern economic and social system of mass production and consumption, enabled stable full-time employment for men that secured the family-wage (Cousins 1999, Ramos 2015; on the other hand, due to their breadwinning responsibility, these economically active men became the direct beneficiaries of public policies in emergent welfare states, including 'family allowances', while economically dependent women were supposed to benefit indirectly through marriage (Crompton, Lewis and Lyonette, 2007;Wall, 2011). 5 However, from the 1960s/1970s onwards, with the economic slowdown that depressed the male family wage, and the emerging feminist movements that were pressing for women's independence and full citizenship, there was a consistent decline of this family model, owing to women's growing enrolment in paid work (Aboim, 2010;Cousins, 1999;Esping-Andersen, 2009). The changes in women's lives did not entirely challenge the traditional gender order, and it would still take time for men to change their commitment to unpaid work and for greater public awareness of men's role in work-life balance to develop (Altintas and Sullivan, 2017;Wall et al., 2017). This state of affairs led to the well-known women's double-burden and difficulty in reconciling paid and unpaid work in most countries; and the shift towards the dual earner/dual carer model (Gornick and Meyers, 2003), pointing to a symmetric allocation of time in paid and unpaid work by men and women, is far from being a reality. At this level, social class matters. Women from poorer and less educated social strata have a higher probability of experiencing double-burden and overload in unpaid work, while highly educated and qualified women are more likely to avoid it by outsourcing unpaid work (Amâncio, 2007;Lyonette, Crompton and Wall, 2007). However, their ability to improve gender equality at home, by reducing conjugal homemaking and childcare, does not necessarily bring societal gains in terms of gender equality, since these occupations are overwhelmingly performed by other women in a highly segregated labour market (Wall et al., 2017). Welfare states, gender cultures and gender division of labour 6 Broadly speaking therefore, the decline of the male breadwinner model has led to two main forms of the conjugal division of paid work, according to women's greater or lesser commitment to the labour market: while some countries have seen a steady increase in the "dual earner" model, with both partners working full-time, in other countries the "one and a half earner" model is more prevalent, with partnered women reducing their participation in the labour market, typically working part-time, when there are preschool children in the household. These alternative models are consistent with the pathways that welfare state regimes have been treading in Europe, offering different sets of objective opportunities for couples with small children to cope with work-life balance. Indeed, if some countries sponsored public policies highly committed to a gender equality agenda, promoting women's full-time employment, along with public and universal childcare facilities and men's role in family care through a more inclusive parental leave architecture, other countries set up policies in line with the principle of individual "choice", relying on the market to provide family services, while others explicitly encouraged maternal care through lengthy maternity leaves and cash for care, resulting in women's career slowdown or interruption and "mommy tracks" (Wall and Escobedo, 2013). According to the original proposal of Esping-Andersen in the early 1990s, these policy frameworks portray the three main welfare regimes that arose in affluent Europe with the decline of the male breadwinner model (respectively, the Social-Democratic regime in the Scandinavian countries, the Liberal Regime in Anglo-Saxon countries, and the Conservative-Corporatist regime in Central Europe, e.g. Germany), but the author overlooked the paths followed by less affluent countries such as the Southern and former Socialist ones.

7
If subsequent works have attempted to close the gap by addressing the characteristics of these regional-based regimes (Ferrera, 1999;Torres et al., 2005;Esping-Andersen, 2009), the awareness of crucial cross-national differences in gender divisions of labour have required authors to grasp the singularities of national pathways (Wall and Escobedo, 2013;Aboim, 2010;Fenger, 2007). For instance, similarities in the social and political developments of the countries of Southern Europe during the twentieth century -late industrialization and societal modernization, the power of the Church and lengthy dictatorships -legitimized clustering them into a Mediterranean welfare regime (Cousins, 1999;Ferrera, 1999) or, in more substantive ways, into a defective and unequal welfare regime (Torres et al., 2005), and a corporatist/familialist welfare regime (Crompton, Lewis and Lyonette, 2007), due to the inability of these States to provide extensive and universal social protection, entrusting to private solidarities the decommodification goal through the provision of family welfare (Esping-Andersen, 2009) and, therefore, failing to put an end to structural social inequalities (Wall et al., 2001;Torres et al., 2005). However, a closer look reveals Portugal's singular path to the dualearner/dual-carer model, since the prevalence of Portuguese women in the labour market on a full-time basis is quite distinct in the Southern context, and even one of the highest in Europe among mothers of pre-school children.

8
The institutional setting for change and diversity in gender relations, and, in particular, women's relationship with the labour market, comprises the various types of welfare state which have arisen in Europe since the mid-20th century and their respective labour markets. Welfare states differ not only in their objective ability to provide social protection and public investment in work-life balance, but also in ideological choices that are deeply rooted in gender cultures, i.e. dominant conceptions of masculinity and femininity, which "form a particular system or structure that underpins policy making, labour markets and individual practices" (Aboim, 2010: 173). This means that gender cultures, which are embedded in all realms of collective existence, are complicit with the gender division of labour in societies, legitimizing persistent sex asymmetries in paid and unpaid work (Knudsen and Waerness, 2008), and stalling the "gender revolution" (England, 2010).

9
According to Esping-Andersen (2009), this status quo is a hallmark of the foundation of a modern welfare state based on familialism. Gender cultures and institutional settings have assigned to women the unpaid work associated with family caregiving and domestic duties, even when they were entering the labour market and becoming the breadwinner. As the author writes, most European countries, with few exceptions, "face intensifying tensions because the female revolution has not been met with a reformed family policy.
[Thus, a] paradox of our times is that familialistic social policy is anathema to family formation" (2009: 80).
10 A different approach to the relationship between welfare states, gender cultures and gender division of labour is offered by Saxonberg (2013), who pinpoints the misleading perspective of Esping-Andersen on the degree of defamilialization of welfare states, wherein the Scandinavian countries would be the most defamilialized and, as a result, the most gender equal. But as Saxonberg highlights, public policies in these countries in fact encourage gender equality by providing public childcare -a recognized defamilializing policy -and generous and lengthy parental leaves for fathers -which may be seen, in fact, as a familializing policy. This author offers a typology that takes into account the degree of degenderization of public policies, distinguishing those "that promote different gender roles for men and women […] [from those] that promote the elimination of gender roles" (2013: 33). Additionally, genderization policies may be explicit, such as those adopted by conservative regimes, but also implicit, like those of liberal regimes that rely on private childcare provision which not all families can afford. By demanding time- consuming dedication to professional life, the labour market imposes major difficulties on work-family balance, undermining gender equality practices in family life , especially for families from lower strata.
Gender stereotypes, gender-role attitudes and gender division of labour 11 In sum, within the goal of enabling families to pursue work-life balance, different welfare regimes put into practice the ideals of masculinity and femininity that stem from a particular gender culture, and consequently prescribe different gender divisions of labour in societies. However, as Aboim (2010) points out, gender cultures are not only undergoing constant change, through a process of assimilation and normalization of institutional and individual practices, but may also have inner tensions at any given moment, upholding a not necessarily consistent constellation of social values; at the same time, these tensions rebound at the institutional level, finding expression in contradictory policies i at the individual level of gender role attitudes, where notions of masculinity and femininity may embody conflicting perspectives.
12 Clearly, sharp tensions in gender cultures have a major impact on men's and women's lives, sustaining an unequal division of labour. In spite of the changes that developed societies have made in the direction of the dual-earner/dual carer model, and an attitudinal climate that encourages less gender specialization, the uneven division of labour is still anchored in highly stable essentialist gender stereotypes. By disseminating collective beliefs on the proper roles of men and women in society, based on natural masculine and feminine attributes and skills, gender stereotypes continue to assign breadwinning responsibilities to men and caregiving responsibilities to women . The effects of this state of affairs on women's lives are widely recognized, namely the double-burden, the onus of reconciliation and lifelong discrimination in the labour market and income (Amâncio, 2007;Wall et al., 2017), especially among women from lower strata who cannot afford to outsource unpaid work. Less well known, though no less pernicious, are the effects on men's lives: by conforming to the dominant "androcentric career model" (Lewis, 2010), characterized by an extended working day and maximum availability in professional life, they have, at least until recently, been deprived of the emotional privileges and rewards of family life and hands-on parenting (Marinho, 2011;. 13 Attitudes to gender roles may in part explain the resilience of asymmetries even when institutional frameworks explicitly and consistently encourage the dual-earner/dualcarer model through degenderizing public policies. As Knight and Brinton (2017) highlight, most national and cross-national longitudinal studies on gender role attitudes have recurrently emphasized the general move from more traditional to more egalitarian attitudes regarding gender roles and the division of labour in Europe, but with a puzzling slowdown trend since the mid-1990s. According to these authors, gender scholars are having trouble in explaining this reality, because they have been understanding gender revolution as an inexorable civilizational drift, with forerunner and laggard countries and attitudes "falling along a linear continuum from traditionalism to liberal egalitarianism" (2017: 1487), i.e. from the support of absolute asymmetry in gender roles that underpinned the male breadwinner/female homemaker model to the support of absolute symmetry that underpins the dual-earner/dual-carer model. Going back to survey data on gender role attitudes from the 1990s and 2000s in 17 EU countries, Knight and Brinton came up with a novel insight on the issue, by identifying three pathways of attitudinal change that resulted from the declining support of "traditionalism", i.e. the constellation of values that uphold, at the same time, male primacy in paid work, the homemaking role of women and gender essentialism: the "liberal egalitarianism" underpinning strict gender equality; the "egalitarian familism" that supports women's participation in paid work, primacy in family life and gender essentialism; and the "flexible egalitarianism" that rejects gender essentialism, while supporting women's choice to take on more traditional roles. Even if it is controversial to speak of egalitarianism regarding attitudinal patterns that are favourable to women's double-burden, the truth is that this proposal unties the Gordian knot that has been stalling the gender revolution: attitudes towards gender equality are changing much faster regarding women's role in the public sphere than their role in the private sphere. Therefore, since they underscore ambivalence concerning women's roles in society, these attitudinal tensions are complicit with the persistence of women's overload, multi-assignment and role conflict, but also with men's minor role in caregiving and homemaking, making asymmetries in paid and unpaid work quite resilient.
Data and methods 14 The following analysis is based on cross-national data from ISSP 2012 -'Family and Changing Gender Roles', ii a survey which examined changes in gender role attitudes and practices concerning the paid and unpaid work of adult men and women aged 18 and over. Family life, professional life and work-life balance were, therefore, at the heart of the questionnaire. Of the participating countries in ISSP 2012, 18 are included in the present analysis, all of them EU member states. iii In order to correct the sample biases at national level, a post-stratification weighting was applied for each selected country, taking into account the distribution of population in terms of sex, age and educational level.
Sampling and dependent variables 15 According to the main goal of this article -to capture the diversity of patterns in the current gender division of labour in European families with children -the analysis focuses on gender role practices in paid work, care work and household work. From the total sample (n=24029), we extracted a subsample of men and women of working ages (18-64 years old), living in a couple (married or in a civil partnership) and with at least one dependent child under 18 in the household (n=6273). With these criteria, we ensured the homogeneity of the sample regarding the high probability of these individuals combining paid and unpaid work (including caring for dependent children), negotiating gender roles, and coping with work-life balance on a daily basis. In fact, it is particularly important to investigate this crucial period of the family life cycle with dependent children: on the one hand, because public policies are responsive to the work-life balance of these families, even if they offer different solutions to help them cope with childcare (mother-centred, public services, private sector); on the other hand, because the negotiation of gender roles in the private and the public spheres is also at its peak, with The gender(ed) division of labour in Europe: patterns of practices in 18 EU c...
Sociologia, Problemas e Práticas, 90 | 2019 major impacts on the conjugal division of paid and unpaid work among heterosexual couples. iv 16 Taking into account this specific population, we analysed the respondents' self-report of the amount of time allocated to paid work, care work and household work, the three key domains of work-life balance in which gender roles are displayed. v In fact, it is broadly recognized that the volume of hours is a particularly useful and reliable tool for assessing work-life imbalances and gender role practices in cross-national studies (Bianchi et al., 2012;Knudsen and Waerness, 2008;Sullivan, 2013), by putting into perspective the asymmetries and the degree of (de)genderization of the countries. We were thus able to produce nine indicators at country level: men's mean weekly hours of paid work, care work, and household work; women's mean weekly hours of paid work, care work, and household work; mean weekly hours of asymmetry between men and women in paid work, care work, and household work.
17 A methodological remark must be made regarding our decision to take into account only the respondents' self-report of time-use, and not their partners' reports. On the one hand, we have to acknowledge the different assessments that individuals make in relation to their own and their partner's contribution, which is usually reflected in underestimating the time spent by the other, or even the inability to make any estimate at all. vi On the other hand, time use is, above all, a subjective experience (Durán, 2013), and particularly significant in unpaid work (but not exclusively so) in which there are multitasking and overlapping activities, less control over time spent on each task, and no work schedule regulation. In fact, separate analysis of the time allocated by men and women to household work and care work is not only imperative, due to the way information was collected, vii but also sheds light on the different rationalities that underpin these distinct domains of family life, which are usually found under the same umbrella of unpaid work (Altintas and Sullivan, 2017;Sullivan, 2013). Moreover, it is important to emphasize that if the respondents self-reported zero hours, that was considered valid and was included in the analysis, on the assumption that the sampling criteria ensure their eligibility to take on professional, domestic and parenting duties; viii however, missing data on respondents' allocation of time meant it was necessary to exclude this case from the analysis of the specific indicator.
18 Drawing on the above-mentioned indicators of gender role practices and asymmetries at country level, expressed in mean weekly hours, we proceeded to a hierarchical cluster (Ward method) with the nine scores of the 18 selected countries, in order to identify different patterns of the gender division of labour in European families with children. This procedure enabled us to identify six patterns, each of them grouping a different number of countries. All the subsequent analysis was performed with the scores of the clusters as dependent variables.
Independent variables and hypothesis 19 The following step was to characterize the clusters, by taking into account additional indicators of gender role practices in paid and unpaid work (types of conjugal division of paid work and proportion of housewives), but also indicators of gender role attitudes (indexes of gender division of paid and unpaid work and of employment-motherhood conflict). To identify the prevailing models of conjugal division of paid work in Europe, we computed a variable from four indicators that depicts the situation of the respondents and their partners in the labour market (if they are currently in paid work and the number of hours/week), and we came up with four types of conjugal division of paid work with different distributions in the clusters: dual-earner couples (both employed full-time); one and a half earner couples (one employed full-time and one employed part-time); breadwinner couples (one employed full-time and one not employed); and half-earner/not employed couples (both not employed, both employed part-time, or one not employed and one employed part-time).
21 After characterizing the clusters in relation to the (de)genderization of practices in paid work, we proceeded with the analysis of attitudes, for the purpose of assessing the inconsistencies between actual patterns of gender division of labour in Europe and attitudes to gender roles. Hence, by selecting a set of consensual indicators on gender role attitudes, we came up with two indexes ranging from 1 (traditional) to 5 (modern). These indicators and indexes are commonly employed in national or cross-national studies which classify social groups or groups of countries along an axis that goes from traditional to modern attitudes, implying that the Southern and post-Socialist countries, on one side, and the Scandinavian countries, on the other, are at opposite poles (Aboim, 2010;Knight and Brinton, 2017). With slight differences regarding the indicators included, these indexes have already been tested in the analysis of ISSP 2002 (Aboim, 2010;Wall, 2007) and have also been tested in an analysis of the Portuguese case in the current issue of this journal (Ramos, Rodrigues and Correia, 2019). 22 The first indexgender division of paid and unpaid work -gathers respondents' replies to four statements: "A job is all right, but what most women really want is a home and children"; "Being a housewife is just as fulfilling as working for pay"; "Both the man and woman should contribute to the household income"; "A man's job is to earn money; a woman's job is to look after the home and family". While the first two statements focus on and compare women's roles and identities as mothers, housewives and professionals, the last two focus on the gender division of labour in couples. The second indexemployment-motherhood conflict -is composed of the responses to three statements that assess the impact of women's employment on the wellbeing of children and the family: "A working mother can establish just as warm and secure 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"; "All in all, family life suffers when the woman has a full-time job". Respondents answered according to a 5-point Likert scale: strongly agree; agree; neither agree nor disagree; disagree; strongly disagree. Two additional remarks are required: it was necessary to invert the scale of two statements in order to cover all the answers, from traditional to modern attitudes; we excluded Spain from the analysis, because the Spanish survey adopted a 4-point scale. 23 In line with the goal of this cross-national analysis -to identify and characterize the current patterns of gender division of labour in European families with children, taking into account practices in paid and unpaid work and their relationship to gender role attitudes -our hypotheses are twofold: 24 Welfare regimes, with their enabling conditions and gender culture model for work-life balance (Aboim, 2010;Esping-Andersen, 2009;Pfau-Effinger, 1998), influence current patterns of gender division of labour in Europe. As some authors claim, the driving force for change relies on consistent and all-encompassing public policies endorsing gender equality. The conservative regimes that offer a policy framework embedded in an essentialist perspective of gender roles, and the liberal regimes that perceive women and men as having the ability to make free choices, putting on the onus of change on individuals themselves, tend to foster the permanence of gender inequality, owing to the status quo of institutional and cultural settings that underpin gender relations (Aboim, 2010;Amâncio, 2007;Esping-Andersen, 2009).
25 · Along with the objective contextual conditions at the institutional level, gender role attitudes at individual level are likely to mobilize gender role practices (Aboim, 2010;Pfau-Effinger, 1998). Accordingly, the most gender-unequal patterns of the gender division of labour are those that embody higher levels of gender essentialism regarding men's primacy in paid work and women's primacy in unpaid work, as they are likely to be related to lower levels of egalitarianism in gender role attitudes (Knight and Brinton 2017).

Results and discussion
Paid work, care work and household work: a country-level portrait 26 In order to depict the patterns of gender division of labour in families with children in Europe, we start by portraying, at country level, the amount of time (mean weekly hours) men and women allocate to paid work, care work and household work, as well as the resulting sex asymmetries in each domain (figure 1). Unsurprisingly, in the 18 selected countries, men spend more time in paid work than women (on average, 13.9 hours more per week) while women spend more time in care work and household work than men (on average, 14.5 and 9.3 hours more per week respectively).  consuming patterns and lower in countries with less time-consuming patterns. In addition, in Scandinavian countries and Great Britain, men allocate more than half the time compared to women. These findings also draw attention to the existence of different cultural models of domesticity in Europe (Aboim, 2010;Crompton, Lewis and Lyonette, 2007), which doubly penalize women by embodying high standards: in the amount of time they allocate to this domain and in the extent of the unequal distribution of household work among couples.
30 Briefly, it is important to underline the following findings: the striking gendered division of labour in most of the countries, with an evident trade-off between paid work and care work, which accounts for the persistence of gender role practices in couples with dependent children; and the distinctive nature of household work, which is not necessarily a side effect of women's primacy in care work or subsidiary status in paid work. In fact, it is an autonomous domain of family life, driven by cultural domesticity standards that may be time-consuming and involve major asymmetries, even in countries where women are committed to paid work; while less time-consuming standards, even in countries where women are more committed to care work, result in lower asymmetries.
Patterns of gender division of labour: a cluster analysis 31 Drawing on the gender role practices (mean weekly hours) in paid work, care work and household work and respective sex asymmetries at country level, we carried out a hierarchical cluster analysis (Ward method) with the nine scores of the 18 selected countries. The procedure enabled us to identify six patterns of gender division of labour ( Figure 2): Cluster 1 -Sweden, Denmark, France and Portugal (n=1562); Cluster 2 -Latvia, Slovakia and Slovenia (n=837); Cluster 3 -Great Britain and Finland (n=546); Cluster 4 -Spain, Hungary, Poland and Lithuania (n=1623); Cluster 5 -Czech Republic, Bulgaria, Germany and Austria (n=1313); and Cluster 6 -Ireland (n=393). The following analysis starts by portraying the six patterns and then offers a characterization of the patterns taking into account a set of independent variables that account for additional practices and gender role attitudes.  Figure 2 shows the different patterns of gender division of labour according to the average gender role practices and asymmetries (translated into mean weekly hours) in the six clusters of the selected EU countries. The clusters were organized according to sex asymmetry in paid work, from lower to higher asymmetry. 33 Cluster 1 brings together Sweden, Denmark, France and Portugal and reveals a low genderunequal with high commitment to paid work pattern. This cluster has the lowest asymmetries in the three domains and has three distinctive traits: the highest allocation of time to paid work and the lowest allocation of time to care work among European women, thereby highlighting more weekly hours in paid work than in care work; the low allocation of time to household work, especially by women; and the significant allocation of time to care work by men. This pattern suggests that trade-offs are taking place in paid work and care work towards more gender-equal practices, which are supported by policies that promote public childcare provision.
34 Cluster 2 brings together Latvia, Slovakia and Slovenia, and reveals a moderate genderunequal with commitment to paid work pattern. In this pattern asymmetries are slightly lower than average in paid work and care work, mainly due to women's less gendered practices in these domains, since men's allocation of time to care work is, indeed, one of the least substantial. But asymmetry is slightly higher in household work, owing to women's time-consuming standards.
35 Cluster 3 brings together Great Britain and Finland and shows a moderate gender-unequal with commitment to care work pattern. In this pattern, asymmetry in paid work is average. Asymmetry is high in care work and low in household work, however, reflecting a clear investment of time in the former instead of the latter, by both women and men. Indeed, the allocation of time to household work is the lowest of all the clusters.
36 Cluster 4 brings together Spain, Hungary, Poland and Lithuania, and reveals a moderate gender-unequal with female commitment to care work and household work pattern. In this pattern, asymmetry is average in paid work, but is high in both domains of unpaid work, mainly due to women's overload. However, it is the cluster where the commitment to household work is more time-consuming, not only for women, but also for men. 37 Cluster 5 brings together the Czech Republic, Bulgaria, Germany and Austria, and reveals a high gender-unequal with male high commitment to paid work pattern. This pattern stands out due to a more traditional gender division of labour, reflected in high asymmetries in all domains, where men's practices seem to contribute more to the status quo than women's. Indeed, this is the pattern where men combine the highest allocation of time to paid work with the lowest allocation to care work and one of the lowest to household work. Therefore, even if women record one of the lowest levels of participation in the labour market, clearly in line with the 'one and a half earner model', men's scanty allocation of time to care and household tasks contributes to women's overload in unpaid work.
38 Finally, cluster 6 includes Ireland only and reveals a high gender-unequal with high commitment to care work pattern. This pattern has features of the male breadwinner/ female homemaker model, with maximum asymmetries in paid work and care work, owing to women's residual allocation of time to the former and extreme allocation of time to the latter. However, this pattern has some interesting traits, as it seems that male practices are less compliant with traditional gender roles than female ones. Indeed, they Gender-role practices and attitudes: characterizing the clusters 39 Figure 3 portrays the distribution, in the clusters, of four types of conjugal division of paid work, highlighting the distinctiveness of each cluster compared to the overall distribution. Additionally, we estimate the proportion of housewives in each cluster. x Figure 3 Conjugal division of paid work and proportion of households with housewives (%) -CLUSTERS, 2012/14 (n = 5494) 40 Dual-earner couples, which involve full-time employment for both partners, comprise the most prevalent conjugal division of paid work in the selected EU countries, totalling 45.8% of the sample. With a substantially lower weight, 27.1%, one-earner couples, in which one of the partners works full-time and the other is not employed, represent the second most frequent situation. After that, with 16.1%, come one-and-a-half earner couples, in which one partner works full-time and the other part-time. Finally, half-earner/not employed couples, with 11.0%, combine different conjugal situations that most probably represent more precarious relations with the labour market: both not employed; both employed part-time; and one not employed and the other employed part-time. These data confirm that dual-earner couples undoubtedly predominate, but one-earner couples are a not insignificant reality, even exceeding one-and-a-half earner couples, and accounting for more than a quarter of the sample. However, the proportion of housewives represents 13.8% of the overall sample of women. If we consider that ISSP 2012 was carried out at the peak of the economic crisis, we should not underestimate the impact of unemployment, not only on the most precarious situations of couples as far as their position in the labour market is concerned, but also on the prominence of one-earner couples, which may reflect traditional as well as involuntary male breadwinning situations. xi 41 Looking at the conjugal division of paid work in the clusters, the first point to note is that dual-earner couples constitute a crucial category for seeing the big picture of the conjugal division of paid work in Europe, as it splits the clusters in two: while in the first three, this type is overrepresented, standing for the pervasiveness of women's entitlement to paid work, in the other three, the type is underrepresented, owing to greater levels of exclusion of women from this domain of the public sphere, at least on an equal footing with men. This implies that even if the dual-earner category comprises the prevalent type among European couples, it is far from being a cross-cutting reality.
42 Going into the specific characteristics of the clusters, we may say that what sets the first three apart are the types underrepresented in each one. Indeed, in cluster 1 the types that are underrepresented are the one-earner and the half-earner/not employed couples, both representing some kind of exclusion from the labour market. Therefore, this cluster highlights the central place of paid work in men's and women's lives, in line with a gender-equal entitlement to the "adult worker model" (Lewis and Giullari, 2005). This is, in fact, the cluster where housewives are underrepresented. Clusters 2 and 3 follow to some extent the same principles, also revealing the underrepresentation of types where partners hold unequal situations (one-and-a-half earner and one-earner); however, these couples are less protected from more precarious situations than those from the previous cluster, and housewives are overrepresented in cluster 3, suggesting a certain polarization of gender role practices. 43 Of the other clusters, cluster 4 offers a portrait of the precariousness of the labour market, xii with the overrepresentation of couples where one or both partners are not employed (or both employed part-time), the one-earner and the half-earner/not employed couples. In cluster 5 there is an overrepresentation of one-earner and one-and-a-half earner couples that underpins gender-unequal practices in paid work based on the structural exclusion or lesser inclusion of women in the labour market. Finally, cluster 6 combines the main traits of the previous two. On the one hand, it reveals the massive exclusion of women from the labour market, since it records the lowest proportion of dual-earner couples and the highest proportion of housewives (three times more than the average). On the other hand, there is weak protection of full-time employment even for men, xiii as revealed by the overrepresentation of half-earner/not employed couples, thereby suggesting that the entitlement to paid work is neither universal nor guaranteed.
44 Figure 4 shows gender role attitudes in the clusters by drawing on the two attitudinal indexes: gender division of paid and unpaid work (index 1) and employment-motherhood conflict (index 2). In both the scores range from 1 to 5, going from the most traditional to the most modern. 46 Within the clusters we find different attitudinal patterns, with cluster 1 and cluster 4 representing opposite poles, as they contain the most and the least modern gender role attitudes. Indeed, cluster 1 stands out, by revealing high scores in both indexes; and the distance in relation to all the others clusters is particularly big in Index 1, thus disclosing a strong commitment to pro-egalitarian attitudes in relation to paid work. Beyond this cluster, only cluster 3 also records scores above the average, meaning that modern attitudes are fairly prevalent, particularly in view of the employment-motherhood conflict (index 2). However, unlike the previous cluster, the index 1 score is lower, suggesting some ambivalence with regard to women's role in paid and unpaid work. Holding more traditional gender role attitudes, below the average, we find cluster 5 and, above all, clusters 2 and 4. But while cluster 2 records the lowest index 1 score, supporting unequal gender roles in paid and unpaid work, cluster 4 records the lowest score in index 2, stressing the employment-motherhood conflict that underpins an essentialist outlook on women's roles. Finally, cluster 6 presents a pattern in line with the general attitudinal climate.
47 Now comparing these attitudes and the overall practices of gender division of labour in the clusters (Figures 2 and 3), we may say that some hold more consistent patterns (clusters 1, 4 and 5), while others contain inner contradictions. In cluster 1, the low genderunequal with high commitment to paid work, and the prevalence of dual-earner couples, are in tune with gender-equal attitudes on paid and unpaid work and the devaluation of women's role conflict between employment and motherhood. In clusters 4 and 5, the moderate or high gender-unequal and the low incidence of dual-earner couples reflect more traditional gender role attitudes. Two other clusters (2 and 6), however, display some contradictions. Cluster 2 reveals a noticeable gap between one of the most traditional gender role attitudes and practices characterized by a moderate gender-unequal division of labour with commitment to paid work, with one of the highest prevalence of dual-earner couples, while cluster 6 embodies the contradiction between average gender role attitudes and the more traditional gender role practices (at least for women), which are moulded by a high gender-unequal division of labour with high commitment to care work, and the lowest participation of women in the labour market, with an unparalleled proportion of housewives. Finally, cluster 3 shows an interesting correlation between fairly modern attitudes, and a moderate gender-unequal division of labour with commitment to care work, with a high incidence of dual-earner couples, but also a high proportion of housewives. This paradox suggests a certain polarization of women with regard to the labour market, reflecting either the principle of choice, or the impact of social inequalities on the way couples cope with work-life balance, both structural traits of countries with liberal welfare regimes.

Final remarks
48 With the aim of identifying and characterizing current patterns of gender division of labour in European families with children, a cross-national analysis of time use data from ISSP 2012 "Family and Changing Gender Roles" was carried out. The allocation of time to paid work, household work and care work was at the heart of the cluster analysis, which drew on a subsample of partnered men and women of working ages and with dependent children from 18 selected EU countries. Six patterns of gender division of labour were identified: low gender-unequal with high commitment to paid work (cluster 1-SE, DK, FR, PT); The gender(ed) division of labour in Europe: patterns of practices in 18 EU c...
Sociologia, Problemas e Práticas, 90 | 2019 moderate gender-unequal with commitment to paid work (cluster 2-LV, SK, SI); moderate gender-unequal with commitment to care work (cluster 3-GB, FI); moderate gender-unequal with female commitment to care work and household work (cluster 4-ES, HR, PL, LI); high genderunequal with male high commitment to paid work (cluster 5-CZ, BG, DE, AT); high genderunequal with high commitment to care work (cluster 6-IE). These patterns were briefly portrayed using indicators of gender role practices (the conjugal division of paid work and the proportion of housewives) and attitudes (indexes of division of paid and unpaid work and of employment-motherhood conflict).
49 First, findings confirmed what is widely acknowledged, that the division of labour in Europe is structurally gender-unequal in a cross-cutting way, sparing no country from this status quo. This is true even for countries that have endorsed the dual-earner/dual carer model through explicit degenderizing policies (Saxonberg, 2013), like the Scandinavian ones. Indeed, the persistence of sex asymmetries in the allocation of time to paid and unpaid work according to prescriptive gender roles comprises factual evidence that European societies, albeit embedded in changing gender cultures (Aboim, 2010;Pfau-Effinger, 1998), unyieldingly attach men and women to essentialist responsibilities in the division of labour . 50 Likewise, findings from the cluster analysis confirmed, to a certain extent, the influence of welfare regimes, with their enabling conditions and gender culture models for worklife balance (Aboim, 2010;Esping-Andersen, 2009;Pfau-Effinger, 1998) on current patterns of gender division of labour (hypothesis 1). In fact, it is not difficult to identify the gender-equal defamilializing script from the social democratic regime in cluster 1, the individual preference script from the liberal regime in cluster 3, and the gender-unequal mother-centred conservative script in clusters 5 and 6. However, the predictive scope of welfare regimes does not fully explain the rationale of these clusters, some of them hosting a puzzling combination of countries that challenges our knowledge, since these countries are not usually grouped together in other cross-national analyses where the starting point is the examination of public policies or gender role attitudes (Aboim, 2007(Aboim, , 2010Crompton, Lewis and Lyonette, 2007;Esping-Andersen, 1990, 2009Ferrera, 1999;Knight and Brinton, 2017;Saxonberg, 2013;Wall, 2007). 51 In fact, taking into account gender role practices in the allocation of time to paid work, care work and household work, as well as the resulting sex asymmetries, brought challenging findings which justify some further remarks.
52 First, looking at gender role practices from a cross-national perspective, and the way they bring together or separate countries regardless of spatial, cultural and attitudinal affinities, we obtain enlightening insights into how practices build the gendered daily existence of couples across Europe. Countries which are usually seen as holding modern gender role attitudes (Aboim, 2010;Jansen et al., 2016) may have unequal practices, falling into a high gender-unequal cluster, as in the case of Germany, or into a moderate genderunequal cluster, as in the case of Finland, while other countries that are usually seen as holding traditional gender role attitudes may have more equal practices, falling into a low gender-unequal cluster, as in the case of Portugal. These inconsistencies between attitudes and practices are most likely the reason why Portugal has always had a puzzling position in the conceptualization of a Southern welfare regime, owing to a singular historical pathway that pushed women very early on into paid work -particularly under the political commitment to gender equality in civil rights and paid work that stemmed from the Revolution of 1974 -even if gender role attitudes did not evolve at the same pace  Aboim, 2010;Wall, 2010;Torres et al., 2005;Cunha and Marinho, 2018). 53 A second remark is related to the heuristic potentialities of going beyond the dichotomy between paid and unpaid work to apprehend current patterns of gender division of labour. If the binary perspective was probably the most suitable for understanding the division of labour in the golden years of the male breadwinner/female homemaker model, this is no longer true. As we have argued above, by disentangling care work and household work (Sullivan, 2013), it becomes clear not only that the major trade-off in gender role practices takes place between paid work and care work, but also that household work comprises an autonomous domain of family life ruled by cultural standards of domesticity that are quite distinctive across Europe. In fact, time-consuming patterns in household work may be present whether women are mostly committed to paid work, as in cluster 2, or mostly committed to care work, as in clusters 4 and 5, but they are always linked to greater sex asymmetries in this domain than when the standards of domesticity are lower, as in clusters 1, 3 and even 6. This finding suggests that Esping-Andersen is somehow right when he claims that what is stalling the gender revolution is the fact that women continue to have less bargaining power in the couple, which is reflected in the persistent overload in household work, and therefore, that the "main source of gender equalization comes from women's sharp reduction in home production" (2009: 34). Another finding is that even if major differences in the gender division of labour in Europe still come from women's practices much more than men's (Aboim, 2010;Knudsen and Waerness, 2008), with the latter recording substantially lower cross-national divergences than the former, when we analyse the two sides of the coin of unpaid work, it becomes clear that care work is the domain of work-life balance where men have more diverse practices, revealing that changing masculinities are more closely linked to caring roles (Elliott, 2016; than roles in housework. Indeed, in clusters 1, 3 and 6, men allocate substantially more time to care work than their counterparts from the other clusters. And, finally, by going beyond the dichotomy, the patterns of gender division of labour displayed the different contribution of the three domains to the characterization of each pattern: while the dispersion of clusters in sex asymmetries defined the degree of gender inequality in the patterns, the volume of time in paid work, care work and household work in each cluster revealed the domain(s) of family life in which men's and/or women's commitment stands out.
54 A third remark concerns gender role attitudes. As expected, clusters contain noticeable differences in attitudinal patterns -the support of traditionalism or egalitarianism in paid and unpaid work (index 1), and the stronger or weaker gender essentialism that mediates women's roles as mothers and professionals (index 2) (Knight and Brinton, 2017) -according to their degree of commitment to gender-equal practices (hypothesis 2). However, clusters 2 and 6 stand out for their inconsistencies between attitudes and practices, with the former supporting more traditional attitudes and the latter displaying more traditional practices. This is a challenging finding which draws attention to the different dynamics that underlie the stalling of the gender revolution (England, 2010): on one side, the attitudes that persistently support gender-unequal roles, i.e. male primacy in paid work and female primacy in (or the ability to choose) household work and care work; on the other side, the restraining influence of a rigid institutional context which does not keep up with attitudinal change at societal level. This seems to be the case of cluster 6, in which gender role practices are much more unequal than attitudes. Finally, we should not neglect the impact of the economic downturn, which affected Europe when ISSP 2012 was carried out, on the patterns of gender division of labour identified in this article. In fact, the crisis affected national economies of different sizes and, as we have seen, clusters 4 and 6 were those that most felt its impact on the precarious situations of couples in the labour market, defying male primacy in paid work. How these developments may affect social and gender inequalities across Europe, nurturing new masculinities and femininities, and creating new meanings and practices in family divisions of labour (Bianchi et al., 2012), is a key topic for future research.

BIBLIOGRAPHY ENDNOTES
i. If we look at the Portuguese parental leave policy introduced in 2009, we may observe that the advanced principles written into the law -combating gender inequalities and promoting a work-life balance based on shared responsibilities between men and women in professional and family life -fell short in practice, due to the concurrent essentialist perspective that mothers should be the main target of leave policy, in primacy and in the amount of time available Ramos, Atalaia and Cunha, 2016;Wall et al., 2017).  iv. Even though we opted to investigate the gender division of labour among partnered men and women, it is clearly crucial to investigate these issues in other types of family arrangements with children where work-life balance and the negotiation of gender roles might be even more demanding, namely same-sex couples. However, there are some limitations in the ISSP data on household composition.
v. "How many hours, on average, do you usually work for pay in a normal week, including overtime?"; On average, how "many hours a week do you spend looking after family members (e.g. children, elderly, ill or disabled family members)?"; "On average, how many hours a week do you personally spend on household work, not including child care and leisure time activities?" vi. In fact, there are more missing data for the weekly hours allocated by the partner to paid work, care work and household work, which would have an impact on the total number of valid cases.
vii. There were two questions in the survey, one for each domain (see previous note). An exploratory attempt to estimate respondents' overall allocation of time to unpaid work (to contrast with the allocation of time to paid work) proved to be spurious, as it resulted in an unrealistic volume of hours, over 168 per week, confirming the eminently subjective nature of time and its limitations for the analysis.
viii. The inclusion of zero hours in the analysis obviously decreases the average in countries that have faced increased unemployment as a result of the crisis.
ix. It is interesting to note that the countries where men allocate more time to care work are those where fertility levels are higher, suggesting that men are being called on to do more childcare (or become more responsive) when there are several children in the household.
x. As we have already mentioned, we do not have information on the sex of the partner, so we have estimated the proportion of housewives taking into account only women's self-reporting.
xi. In the analysis of the conjugal division of paid and unpaid work of couples in ISSP 2002, Aboim (2010) came up with 22.8% of one and a half earner/female carer and only 13.4% of male earner/ female carer couples. Even if we take into account methodological differences in the construction of the indicator, the significant divergence with regard to the current data suggests that the increase in couples where one of the partners is out of labour market may be due to unemployment.
xii. Although we did not carry out the analysis of the impact of unemployment on the types of conjugal division of paid work, 7.9% of the total sample stated they were unemployed. This situation is only overrepresented in cluster 4, where the proportion of declared unemployment rises to13.4%. In fact, Spain was an exemplary case of the dramatic rise in unemployment during the crisis: in 2012, the proportion of unemployment in the active population was 24.6% for men and 25.1% for women (source: Eurostat, LFS, consulted in March 2018).

ABSTRACTS
Drawing on cross-national data from ISSP 2012, this paper aims to identify and characterize the patterns of gender division of labour in European families with children. Analysis focuses on gender role practices at country level, assuming that welfare regimes frame the gendered allocation of time to paid work, care work and household work. From a cluster analysis based on time-use (weekly hours and sex asymmetries) in 18 EU countries, six patterns are identified. Findings disclose the relevance of disentangling care work from household work in order to understand in more detail the diversity of patterns across Europe. Este artículo tiene por objeto identificar y caracterizar los patrones de división sexual del trabajo en las familias con niños en 18 países de la UE (ISSP 2012). A partir del supuesto de que los regímenes de protección social estructuran la asignación de tiempo en el trabajo pagado, de cuidar y doméstico de hombres y mujeres, el estudio aquí presentado se basa en la comparación de las prácticas instituidas en cada país. El análisis de clusters permitió identificar seis patrones de uso del tiempo (horas semanales y asimetrías entre hombres y mujeres) distintos. Los resultados evidencian la importancia de separar el trabajo doméstico del trabajo de cuidar, para comprender mejor la diversidad encontrada.
Keywords: cross-national analysis, gender division of labour, families with children, time use.