Equal sharing or not at all caring? Ideals about fathers’ family involvement and the prevalence of the second half of the gender revolution in 27 societies

ABSTRACT Using attitude data from the International Social Survey Programme (ISSP) of 2012, we study the prevalence of the second half of the gender revolution – the involvement of men/fathers in care and housework on equal terms with women/mothers. With a focus on the collective consciousness in 27 societies, we (1) map patterns of support for different family model ideals; (2) study the extent to which these ideals are related to national-level indicators of gender equality and modernization; (3) analyse similarities and differences between groups of societies, with a focus on which ideals represent conservative and progressive alternatives in each society; and (4) analyse group differences and the degree to which these ideals are contested within societies. We find that the ideal of a father as provider and a mother as caregiver persists but is challenged in nearly all societies by other alternatives, including: mothers’ part-time work; full-time work for both mothers and fathers; and a dual-earner/dual-carer ideal, with shared responsibilities for paid (part-time) and unpaid work. On the societal level, modernization and gender equality are positively associated with both progressive family ideals and marked group differences, indicating that fathers’ involvement in the family is a contested issue in progressive societies.


Introduction
In this article, we examine the collective consciousnessthat is, attitudes in the general populationabout the gendered division of paid and unpaid work/care in 27 societies.The aim is to delineate the extent to which different countries are heading towards a gender-equal and socially sustainable societyin other words, a realization of the second half of the gender revolution (Goldscheider et al., 2015), which refers to a conception of gender equality that entails not only women's emancipation but also the entrance of men/fathers into the domestic sphere to hold responsibility for both care and housework on equal terms with women/mothers.
Over the latter half of the twentieth century, Western societies underwent a profound change: the mass entrance and establishment of women in the labour force.In tandem with this development, influential research, frequently within the framework of the second demographic transition, recorded several substantial changes in family behaviour: the postponement or abandonment of marriage, declining fertility rates, increasing rates of union dissolution and alternative family forms outside of marriage (Lesthaeghe, 2010;Zaidi & Morgan, 2017).The earlier view of the nuclear family (i.e.married couples with children) as a stable cornerstone of society seemed to lose its validity.To articulate this shift in terms that may be overly dramatic, these demographic changes were perceived as a substantial threat to the survival of societyparticularly as the former 'obligation' or standard way of living in which couples had children was anticipated to lose its importance and become only one option among others for adults' projects of self-actualization (Lesthaeghe, 2010;Goldscheider et al., 2015, p. 215).
However, at the end of the century, accumulating evidence indicated that some of these trends were weakening or had even reversed (most notably fertility), and some researchers started to reconsider the implications of women's increased employment.While acknowledging that the transition of women/mothers into employment initially strained the nuclear family in various ways, the main message from this school of thought is that this shift should rather be understood as the first half of the gender revolution: the entrance of women into a previously typically male sphere of society.The second half of the gender revolution thus involves an opposite change in traditional gender patterns: a fundamental transformation in men's behaviour as men become involved in the family and in the care of childrena change that is presumed to have family-friendly consequences (Goldscheider et al., 2015, p. 215). 1   Together these two master family trends constitute the two halves of the modern gender revolution, as the separate spheres that divided the worlds of adult men and women have been broached, first by women's entry into the public sphere and now by men's entry into the private sphere.(Goldscheider et al., 2014b, p. 996) While these scholars assume that the second half of the gender revolution is not yet far advanced across countries, they draw on research (predominately from Scandinavia and the USA) to suggest that the emerging contours of increasing male involvement in the traditional female sphere signal the dawn of a significant societal change, and that such a development would be beneficial for societies, both socially and economically (European Commission, 2019;Friedman 2015;Goldscheider et al., 2015;Gornick & Meyers, 2008).
Research on the second half of the gender revolution, which has grown over time, uses empirical data on behaviour and/or attitudes and beliefs.Nevertheless, especially when it comes to data on attitudes or beliefs about family and care, systematic large-n crossnational research on men's family involvement is rare.One important reason for this lack is that truly cross-national comparative data treating men and women on equal terms have been scarce.However, thanks to recently designed gender-symmetrical formulations of survey questions, the opportunity to assess the nature and prevalence of this second half of the gender revolution across countries has improved.We believe that such an assessment can serve as an important complement to behavioural data on housework and childcare, which are often limited to samples of appropriate segments of the population (e.g.couples or parents with preschool-aged children).Instead, we use nationally representative data covering the general population.
We focus on normative attitudes regarding how couples with children should organize their work life and family life.In this way, we attempt to measure the prevailing ideals and norms in a societythat is, a country's collective consciousness.This approach makes it possible to study the alternatives people believe to be actually viable, along with the support for these alternatives, both in the population at large and across social groups.As we argue later, group differences can indicate the degree to which gender relations are politicized and salient in a society (Edlund & Lindh, 2015;Kumlin & Svallfors, 2007); to some extent, they can also serve as an analytical tool for predicting societal change.
A country's institutional arrangement and general level of gender equality, measured in terms of women's employment rate and political representation, and the country's modernization, measured as economic affluence, provide a useful fruitful framework for explaining the similarities and differences between countries in terms of the prevalence of the second half of the gender revolution.We employ national representative data on the adult population collected within the International Social Survey Programme (ISSP) of 2012, covering 27 societies (listed in Section 3).

Previous research and our contribution
While country-comparative researchboth cross-sectional and longitudinalon gender and equality is extensive, this research has been dominated to date by a focus on women's/ mothers' gainful employment and the consequences thereofthat is, the first half of the gender revolution (Kunovich & Kunovich, 2008;Perry-Jenkins & Gerstel, 2020;Sullivan et al., 2018).Although many studies concentrate on attitudes towards women's employment (but see Salin et al., 2018), scholars have also examined gendered behaviour related to the distribution of housework among couples.Overall, findings on the distribution of housework show a slight gender convergence over time, mostly due to a reduction in the time women spend on housework and only slightly due to an increase in the time men spend on housework.The national context has an expected, albeit not particularly strong, influence on the division of housework in families (Altintas & Sullivan, 2017).Men do a larger share of the household work in countries where women's labour force participation and general economic and political power are greater (Fuwa, 2004;Hook, 2006).In a similar vein, citizens living in countries with more gender-equal policy contexts and higher employment rates among women have comparatively more positive attitudes towards women's employment (Crompton et al., 2005;Sjöberg, 2004).
This strong emphasis on women and employment is understandable, as the female entry into a traditionally male domain and the consolidation of their position there marked a significant change in the societal structure of Western societies.Lately, however, contemporary debates in many societiesspecifically in rich Western countries are not primarily about whether or not women should be engaged in paid work, but about the role of men in unpaid work and care, and gender equality in both of these spheres (Frejka et al., 2018;Goldscheider et al., 2014a;Saxonberg, 2013).
In line with this growing interest, research on the second half of the gender revolution has expanded over time (Evertsson, 2016;Rush, 2015;Schoppe-Sullivan & Fagan, 2020), and its geographical focus has widened from a focus on the USA and the Nordic countries to include other European countries (Pailhé et al., 2021;Steinbach & Maslauskaitė, 2020;Steinbach & Schulz, 2021).A majority of the studies on men's/fathers' involvement in the family are conducted within a single-country framework or as relatively small-n comparative studies (see Hook, 2006, for a large-n comparison).In general, over the past decades, parents have increased the time they spend on childcare, irrespective of gender.In general, the changes in men's childcare time are greater than the changes in their time spent on housework (Pailhé et al., 2021).In a review covering a substantial number of studies on fathers and fatherhood from several countries across the globe, Seward and Rush (2015) conclude that Scandinavia stands out among Western countries, with Scandinavian fathers taking on more responsibility for their children than fathers in other Western countries.While fathers' involvement is lagging behind in many Western countries, the types of fatherhood practices are fairly similar and vary substantially less compared with those observed among non-Western countries.The observed diversity in fathers' practices and efforts across countries indicates a continued divergence, rather than convergence.
Our contribution builds on and complements previous research.In relation to crossnational research on normative attitudes about gender relations, we broaden the traditional analytical focus on mothers by bringing in both parents on equal terms.Compared with previous research on behaviour, which often relied on specialized samples of parents or couples, we cover the general population in a large number of countries.Although the association between attitudes and behaviour may be less than straightforward, few would agree that they are unrelated.Under certain conditions, the association can even be strong (Ajzen & Fishbein, 2005).Our data do not permit us to assess the strength of this relationship, but we note that Aassve et al. (2014), using European cross-national data, observe that attitudinal dispositions about gender relations are strong predictors of actual gender-related behaviour.
In line with previous research, the theoretical point of departure for this study is that factors at the societal level are relevant for understanding attitudes towards gender equality and men's involvement in the family, assuming that changes occurring at the contextual level will entail changes at the individual level.More specifically, macro-oriented societal and institutional perspectives and feminist theories on the welfare state offer a valuable starting point (Ciccia & Sainsbury, 2018;Rush, 2015).Overall, this framework emphasizes the role of macro-social units in influencing mass attitudes and behaviour, including the ability of social policy to structure and convey normative messages about gender specialization versus dual/overlapping social roles (Bergqvist & Saxonberg, 2017;Daly & Rake, 2003;Lohmann & Zagel, 2016).As systems of rules and regulations, institutions embody national traditions and previous power struggles between social actors.The institutional framework surrounding citizens in a given context is of substantial importance in structuring individuals' behaviours, preferences and perceptions, both through the redistribution of resources and as a carrier of norms (Korpi & Palme, 1998;Mettler & Soss, 2004;Svallfors, 2007).Consequently, through their institutional frameworks, countries facilitate or even promote different 'family models'.Policies can encourage men and women to have double responsibilities as both providers and as carers, or can support differentiated responsibilities for women (caregivers) and men (providers).These two ideal-typical dimensions form the basis for central categorizations of countries into different family policy regimes (e.g.Korpi et al., 2013).The theoretical basis of these classifications usually rests on empirical knowledge of specific countries that are seen as archetypical cases (Rush, 2015).
Another influential tradition of research, which has a more global perspective, suggests that the societal development towards gender equality is part of an ongoing, broader process in society.This process includes a more wide-ranging transformation of values and practices in a post-materialist direction, propelled by increased economic affluencethat is, modernization (Wilensky, 2002) and, on the individual level, by generational replacement and the disproportionate distribution of political resources between lowly and highly educated citizens (Apparala et al., 2003;Inglehart, 2018).
In our view, both of these research traditions emphasize the role of country-context factors in structuring individuals' perceptions of gender relations and attitudes towards the gendered division of paid and unpaid work and care.Nevertheless, it is necessary to qualify this argument.From our perspective, the relationships between macro-level contextual factors (e.g.national institutions) and public opinion should be viewed as mutually dependent over time.In short, while national institutions affect public opinion and the interests of specific social groups, it is equally true that public opinion and group interests affectoften, but not exclusively, via general electionsgovernment behaviour and thereby the structure of national institutions (Brooks & Manza, 2007;Mettler & Soss, 2004;Rothstein, 1998).This latter assumption is integral and forms the rationale for our interest in analysing public opinion, as it is a potential force for societal change or the preservation of the current circumstances. 2  In regard to typologies, our study examines a large number of countries, many of which have not been included in the described country classifications.Therefore, we apply an explorative data-oriented approach in which the classification of countries is not predetermined in existing typologies.
The aim of this study is to explore the prevalence of the second half of the gender revolution by analysing the collective consciousness regarding the gendered division of work, with a special focus on men's involvement in the home, across 27 societies.First, we map patterns of support for different family model ideals.Second, we study the extent to which these ideals are related to national-level indicators of gender equality and modernization.Third, we analyse similarities and differences between groups of countries concerning the support for family model ideals.Here, we particularly focus on which models that represent the more conservative and more progressive alternatives among the public in each country.Finally, we examine differences between social categories and analyse the degree to which ideals on the gendered division of work and care are contested within the studied countries.We particularly concentrate on generational and educational differences.

Data, measures and methodological strategy
The analyses draw on data from the 2012 ISSP module Family and Changing Gender Roles IV.While earlier rounds of this module almost exclusively focused on attitudes towards women's employment, the 2012 module includes the role of men in the domestic sphere and offers a wider focus on gender relations within the household with respect to work and care.The analyses include data from 27 societies: Austria, Australia, Belgium, Bulgaria, Canada, Croatia, Czech Republic, Denmark, Finland, France, Germany-east, Germany-west, Hungary, Iceland, Ireland, Latvia, Lithuania, the Netherlands, Norway, Poland, Portugal, Slovakia, Slovenia, Sweden, Switzerland, the UK and the USA.All samples are nationally representative of the population aged 18-80 years. 3 We focus on attitudes concerning how parents should divide the responsibilities of economic provision and unpaid housework and childcare between them.This focus is captured by two questions asking respondents to indicate how a prototypical family should organize their family and work life.These questions capture the respondents' opinion about both the best option (Q1) and the least desirable option (Q2).
Q1. Consider a family with a child under school age. 4 What, in your opinion, is the best way for them to organize their family and work life?Q2.And, in your opinion, which of these options would be the least desirable?
Alternatives for both questions: 1.The mother stays at home and the father works full-time.
2. The mother works part-time and the father works full-time.
3. Both the mother and the father work full-time.4. Both the mother and the father work part-time.5.The father works part-time and the mother works full-time.6.The father stays at home and the mother works full-time.
These questions cover two dimensions: the first dimension asks about the strength of the respondent's orientation towards paid work, while the other is connected to gender and relates to whether the respondent views the work orientation differently if it concerns the mother or the father.
Recent research has emphasized the multidimensionality of attitudes associated with gender relations (Grunow et al., 2018;Knight & Brinton, 2017;Scarborough et al., 2019).The first part of our analysis aims at distinguishing different types of preference patterns that can be retrieved from the two questions.For this purpose, we will use latent class analysis (LCA), which is a suitable method for identifying qualitatively different configurations of categorical variable responses (Magidson & Vermunt, 2001); that is, it can reveal the patterns of ideals and attitudes that are dominant in different countries.Using this design, the observed patterns across countries are empirically generated, with no prior theoretically based constraints being applied.In more technical terms, LCA examines whether the relationships within a set of observed indicators are explained by a latent variable.The aim is to identify clusters of individuals who share similar preference patterns.If two dominant combinations of preference patterns exist, a two-cluster model will fit the data.If the sample can be divided into three configurations, a three-cluster model will be selected, and so on.Applying different model-fit statistics makes it possible to determine the number of dominant patterns.By including country as a covariate, we specify that (1) the characteristics of each cluster are identical across countries, but that (2) cluster membership probabilities are allowed to vary across countries.We thus ensure that the empirical assessment of the construct is the same in each country.
The second empirical part of our analysis includes two different strategiesone quantitative and one qualitative.The quantitative strategy studies the relationships between the idealtypical family preferences retrieved from the LCA and two well-recognized key indicators that are used in international indices of female empowermentlabour force participation and political representation in parliament (Klasen & Schüler, 2011;Phillips, 1998).Here, we follow the main theoretical assumption that explains the extent to which men engage in domestic duties and care of children as a consequence of women's empowerment (i.e. the process of controlling one's life economically, socially and politically).In relation to the projections of modernization theory, we also include a measure of the countries' wealth (BNI/capita).The country-level data come from the United Nations development programme.
The qualitative strategy builds further on the first classification of countries in the LCA.In this part of the analysis, we classify countries into groups based on the support for the different family models retrieved from the LCA.Hence, our grouping of countries is based on empirical results; not on any a priori defined theoretical categorization.When doing this classification, we focused on which types of family model represent the most conservative and the most progressive options in each country and examined the strength and types of social cleavages within countries associated with different family models.Here, the social cleavages of interest are generation, education and gender, all of which have been demonstrated by previous research to be important in structuring attitudes towards gender equality, albeit to a varying degree.It is expected that women, the young and the highly educated will be in favour of the more progressive gender-equal family models, while men, the elderly and those with low education will prefer models involving a more traditional division of economic provision and unpaid care work between men and women.
Differences between social categories are used, first, as informative descriptions of current circumstances in the countries, and second, as a tool with the potential to indicate the direction in which a society is heading in these matters (Edlund & Öun, 2016).By examining group differences in support of these alternatives, we may be able to predict, at least to some extent, countries' propensity for change towards the second half of the gender revolution, based on the idea of reciprocal associations between public preferences and institutional change.In this case, building on theories on post-materialism and modernization, educational and generational differences are particularly important, as change is driven by generational replacement and the better educated.To perform this analysis, we use LCA to calculate the probability of each individual belonging to each cluster.We then use these probabilities as dependent variables in ordinary least squares (OLS) regressions analysing attitudinal differences between these social categories in each country (i.e.gender (two categories); age (four categories: 18-24, 25-39, 40-64, 65-80) and education (completed number of years in education recoded into quartiles)).For each country group, we have selected one typical example of how social cleavages are structured (see the results in the Appendix, Tables A2-A7).

Part 1: Clusters of different family ideals
How should parents organize their everyday family and work life, who should take on the provider role and who should take care of the home and children?As we will show, for a large majority of respondents, this question is mostly about the degree to which mothers should be employed outside the home (i.e.not at all, work part-time or work full-time).
However, some prefer the idea that mothers and fathers should share the responsibility for economic provision as well as the caregiving role in the family.The LCA demonstrates that six dominant preference patterns exist.Table 1 shows the characteristics of each cluster.The cell entries indicate the probability of supporting each of the options by cluster membership (for model-fit statistics, see the Appendix, Table A1).
Respondents classified into cluster 1 prefer the one-and-a-half-earner family model.For this group, the least desirable option is for both parents to work full-time.Instead, they believe that the best way for a couple with a child to organize their family and work life is for the mother to take on more responsibility for the household and family than the father by working part-time.Almost 25 per cent of the respondents are allocated to this cluster.
The analysis reveals three clusters of preferences in which the mother is mainly understood to be a housewife.However, they differ in certain respects regarding the 'least desirable option'.The respondents in cluster 2 prefer a family model in which the mother stays at home to take care of the house and family, while the father works for pay.The least desirable parental care option in this cluster is a situation with reversed gender rolesthat is, the father stays at home while the mother works full-time.This cluster is therefore named the ideological male breadwinner/female homemaker family model.In cluster 3, the respondents also prefer a family model with a stay-at-home mother and a full-time employed father.However, their least desirable option is that both parents work full-time.We have denoted this cluster as the child-centred male breadwinner/female homemaker family model.Finally, the respondents in cluster 6, the traditional male breadwinner/female homemaker family model, exhibit a pattern similar to that of the respondents in clusters 2 and 3 with respect to strong support for a family model in which the mother stays at home and the father takes on the provider role.However, the least desirable option in cluster 6 is mixed between a situation in which both parents work full-time and a situation in which the father stays at home to care for the children and take care of the household.Taken together, these three clusters constitute almost 50 per cent of the respondents.Cluster 4 contains respondents who prefer a solution with shared responsibility for both paid work and care: the dual-earner/dual-carer family model.The best option for this group is for both parents to work part-time, and their least desirable option is for both parents to work full-time.About 14 per cent of the respondents belong to this cluster.Next, cluster 5 contains individuals with a strong work orientation.These respondents prefer the full-time family model, in which both the mother and the father work full-time.The worst scenario for these respondents is for one parent to stay at home or for both parents to work part-time.This cluster constitutes approximately 13 per cent of the respondents.

Part 2A: The quantitative strategyassociations between macro-level indicators and family model ideals
As the prevalence of the preference patterns captured in the LCA analysis and shown in Table 1 may be influenced by the national context in terms of gender equality in political representation and labour-market participation, as well as the level of modernization, our first analytical strategy is to group the countries along a conservative-progressive scale (CPS) based on their cluster membership probabilities (Table 2). 5In this context, 'conservative' indicates strictly gender-separated spheres of work (men) and home (women), while 'progressive' denotes gender neutrality in these two spheres.The scale (denoted as CPS-1) is based on the support for clusters 1, 4 and 5, where the latter two are considered to be more progressive than the first.CPS-1 is calculated as follows: CPS-1 = (C1*0.50)+ C4 + C5.A high value indicates progressive preferences. 6 To what extent is CPS-1 associated with the selected country-context variables?The correlations between CPS-1 and all three macro-level indicators are strong and positive: gender equality in the spheres of politics (r = 0.79), employment (r = 0.74) and modernization (r = 0.66). 7In conclusion, the higher the gender equality in politics and employment and the higher the degree of modernization, the lower the proportion of citizens in a country that prefer a traditional gendered organization of a family's work and care obligations.

Part 2B: The qualitative strategycountry groups
To display the results from the qualitative classification of the countries based on cluster membership characteristics and social cleavages based on gender, education and generation, we group the countries into six groups, A-F, as shown in Table 2.
As this classification is based on the social support for each cluster across countries, in terms of both levels and existing group differences, we can examine what alternatives people consider to be actually viable and each alternative's support across social groups.It is important to acknowledge that these alternatives may differ in meaning across countries.An alternative that is viewed in one country as the progressive choice (e.g.'mothers take on part-time work and care, while fathers work full-time' vs. 'female homemaker/male breadwinner solution'), might indicate a conservative choice in another country (e.g.'mothers take on part-time work and care, while fathers work full-time' vs. 'both mothers and fathers work part-time with shared care responsibilities').As the family models respectively representing the most conservative and progressive alternatives can vary across countries, it is important to analyse the support for a model in relation to other models in each country.We distinguish six qualitatively different groups of countries with different configurations regarding the levels of support for the family models and the types of family model that represent the most conservative and most progressive option.Group A consists of six countries (Bulgaria, Slovakia, the Czech Republic, Lithuania and Hungary) where the level of support for the different kinds of male breadwinner/female homemaker family model (clusters 2, 3 and 6) is overwhelming.All other clusters receive only marginal support, and no progressive option stands out.Thus, there is continuous support for the male breadwinner/female homemaker model in this group of countries and no indication of support for another family model.The few group differences that exist show that questions concerning the gendered division of labour are not an important political issue in these countries.Slovakia is a typical example country of this group.In Slovakia, the support for two variants of the male breadwinner/female homemaker model dominates public opinion, and differences between social groups are generally small.There are, however, significant differences between educational groups concerning the type of male breadwinner/female homemaker model that is preferred.Individuals in the highest educational group show less support for the most popular ideological breadwinner/homemaker model (cluster 2) and more support for the child-centred breadwinner/homemaker model (cluster 3).
In Group B, a preference for variants of the male breadwinner/female homemaker family model is also widespread.This model constitutes the conservative alternative in this group of countries (Austria, Latvia, the UK, the USA, Portugal, Australia and Ireland).However, in this group, we also find a clear pattern of support for the oneand-a-half-earner family model, which represents the major progressive alternative, although there is also some support for other more progressive alternativesthat is, clusters 4 and 5. Latvia is a typical example country for this group.In Latvia, men's primary responsibility as a provider is not contested.Based on differences in support for the family models, the conflict runs mainly between the one-and-a-half-earner model (cluster 1), which appears as the more progressive or gender-equal alternative, and two conservative ideals (clusters 6 and 2).Highly educated individuals prefer cluster 1 and show less support for cluster 6, while older individuals show less support for cluster 1. Cluster 2 finds more support among men than among women.
Group C, which contains four countries (Croatia, Poland, Canada and Slovenia), is similar to the previous group in the sense that variants of the conservative male breadwinner/female homemaker family model receive strong public support.However, the progressive alternative in this group is shared between the one-and-a-half-earner family model and the full-time family model.This is also the case in Slovenia, which is a typical case in the group.Group differences are pronounced and largely mirror each other when it comes to support for both these alternatives.For the ideological male breadwinner/female homemaker model (cluster 2), support is comparatively higher among individuals with low education, the elderly and men.Conversely, highly educated and younger individuals are more in favour of the full-time and the oneand-a-half-earner models (clusters 5 and 1)the more progressive alternatives.Thus, in Slovenia, we find that a large part of the population supports the conservative male breadwinner/female homemaker model.Those who advocate for change are torn in two different directions: towards the one-and-a-half-earner model, on the one hand, and towards the full-time model, on the other hand.The dual-earner/dual-carer model does not seem to have entered into the collective consciousness of the Slovenian public; rather, the ideological conflict between conservatism and progressivity is about women being or not being in the labour force.
In Group D, which includes three units (Germany-west, Switzerland and the Netherlands), a large part of the population supports the one-and-a-half-earner family model, but there are forces that want to shift society away from this model.Some strive towards the more conservative male breadwinner/female homemaker model, while others want to move towards the dual-earner/dual-carer alternative with shared responsibilities; however, very few support the full-time family model.Switzerland is our example case.Here, we find clear group differences in the expected direction with regard to education and gender for the dual-earner/dual-carer model.Conversely, we find mirroring group differencesalbeit less prominentin support for the more conservative alternatives.
In Group E, the male breadwinner/female homemaker family model is less popular than in the previous groups, but this model still constitutes the conservative alternative for the three countries in the group (Belgium, France and Finland).The progressive alternative is a mixture between the one-and-a-half-earner family model, the full-time family model and the dual-earner/dual-carer family model.France is chosen as an example of this group.In France, the progressive alternative is represented by clusters 4 and 5, while the conservative alternative is made up of clusters 2 and 6.Cluster 1 constitutes a middle alternative.There are clear educational differences in support of both clusters 2 and 6, with individuals with lower education being more in favour and those with higher education less in favour of the male breadwinner/female homemaker family models.The opposite is true for cluster 4: individuals with higher education are more supportive of this cluster than those with low education.There are also clear age differences in the support for clusters 4 and 6 that point in the same direction, with older people being more in favour of the conservative alternative.
Group F (Germany-east, Denmark, Norway, Iceland and Sweden) differs from the other groups in that the male breadwinner/female homemaker family model is discarded, in effect, as an option.Instead, the one-and-a-half-earner family model stands out as the conservative alternative.The progressive alternative is either the full-time model or the dual-earner/dual-carer family model.We also find pronounced differences between social categories in their support for the common family models (except in Germanyeast), indicating that the gendered division of labour may be a salient political issue in these countries.Our example from this group is Sweden, where the main alternatives are the more traditional one-and-a-half-earner model and the more gender-equal dual-earner/dual-carer model.These alternatives are supported by each expected constituency.Women and the highly educated support the progressive alternative, while men and those with lower education support the more traditional family model.However, we do not observe strong generational differences in Sweden.
While it was expected that the gender-equal family models would receive the strongest support from the citizenries in the Nordic countries, the observations for Germany-east deserve a comment, especially as the results are different from those observed for Germany-west.Our explanation points both to the possibility of a lingering impact on public consciousness from the former German Democratic Republic (GDR) institutional setup prior to the reunification (Svallfors, 2010), and to the still-existing institutional and behavioural differences between Germany's two parts.Compared with the Western regions, the Eastern regions of Germany have a higher prevalence of public child care (Schober & Stahl, 2014) and a higher female employment rate (Weinkopf, 2014).
In the above description of the country groups (A-F), social cleavages appear to be less pronounced in the more conservative country groups.Indeed, the correlation between conservative-progressive scale (CPS-1) and the social cleavage scorea summarized measure of all group differences per country 8is rather strong (r = 0.60), suggesting that the higher the proportion of citizens holding progressive preferences, the more pronounced the differences in preferences between social categories.This finding may indicate a higher degree of political salience of these issues in the more progressive countries.At the same time, it is interesting to note that some countries in the middle groups (e.g.Poland, the Netherlands and Switzerland) also show a high social cleavage score.

Conclusions and discussion
Questions of family formation, fertility and care are important components for understanding a society's capacity for survival and wellbeing.Men's increased participation in the domestic sphere is supposed to entail a positive development towards gender equality and a socially and economically sustainable society, as the family becomes a joint project in the dual-earner society.In many countries and in the European Union, the gender-equality debate increasingly involves a discussion on men's responsibility for care and housework, and on men's and women's opportunities to participate equally in work and family, echoing the literature on the second half of the gender revolution.A key question within the field is whether we are presently experiencing a stalled or an ongoingalbeit perhaps slow and unevengender revolution (Friedman 2015;Hook & Paek, 2020).
By studying the attitudes of the general populationthat is, the collective consciousnessregarding the gendered division of work and care, this article attempts to map the prevalence of the second half of the gender revolution across 27 societies.The main conclusion is that this part of the revolution is not widespread.In a large majority of the studied societies, citizens believe that women should be responsible for the home and family, while men's family involvement is seen as less important.However, in all but a handful of societies, this traditional ideal is contested by other alternatives that involve a division of paid work and unpaid work/care.It is only in a group of six Eastern European societies that the male breadwinner/female homemaker ideal stands unchallenged.Across the other societies, we find three different alternatives that represent progressive alternatives to this ideal: women's part-time work; a full-time ideal for both women and men; and a dual-earner/dual-carer family ideal, with shared responsibility for paid work and care (shared part-time) for both mothers and fathers.
What is considered to be a progressive alternative in one country can be considered a conservative alternative in another country.While the one-and-a-half-earner ideal is the progressive alternative in some countries (most strongly in a group of countries made up of Anglophone countries, Austria, Latvia and Portugal), it represents the conservative option in other countries, including four of the Nordic countries, as well as Germanyeast.In the latter societies, the public seems to have left the male breadwinner/female homemaker model behind.Here, the picture of the progressive alternative is divided: the societies in this group lean towards either the full-time family model or the dual-earner/ dual-carer family model.Furthermore, our results show that the degree of modernization and gender equality on the national level is positively associated with the prevalence of progressive gender-neutral family ideals.In line with established research on family policy regimes, this finding suggests that national institutions and the norms that they convey have an impact on the public consciousness of the desirable and appropriate behaviour and responsibilities of men and women at work and in the family.However, given the mutual dependence between institutional developments and peoples' preferences over timein the sense that public opinion may drive political changewe also paid attention to how the extent of the support for different family models was structured by social group belongings.Two of these social cleavages were considered to be particularly important for informing us on future developments of gender relations: education and generation.
The strength of social cleavages tended to be most pronounced in the more progressive countries and least pronounced in the more conservative countries.We believe that a useful explanation for this phenomenon may be found in the institutional setup and political rhetoric surrounding family policy and gender.The most prominent institutional changes over the last decades have occurred in the more progressive countries.Apart from affecting behaviour in these citizenries, there are indications that these developments have been accompanied by salient family-related political rhetoric and public debate.Therefore, in progressive countries, the public is likely to be comparatively more informed of issues about gender and family, and political alternatives (conservative and progressive) for changes in family policy are likely to be more clearly defined.
If it is true that prominent social cleavages in expected directions increase the propensity for change in a country, we cannot expect a convergence across countries in the near future.Rather, we expect that the most progressive countries in our study will continue their development towards a realization of the second half of the gender revolution, thereby increasing the distance between them and the most conservative group of countries, which currently shows very little propensity for change.However, between these two contrasts, we observe a large middle group of countries where the male breadwinner/female homemaker ideal is still perceived as viable yet is contested by a significant segment of the public.Among some of these middle-group countries, we find prominent social cleavages.
Finally, we are aware that a study on the nature and prevalence of the second half of the gender revolution can be approached from different angles, and that a focus on public consciousness from a cross-country perspective is only one of them.It should also be mentioned that our large-n design and the variation of countries included prevented us from using comprehensive institutional data in our analysis.The country-level factors are instead 'output' context measures of gender equality, as institutional data are not available for all countries at present.Nevertheless, it is likely that the prevalence of women's employment and political representation in a country is influenced by national institutions.Moreover, our cross-sectional design limits our ability to observe real developments and change over time, although we make an attempt to discuss the issue in these terms.Our predictions may, however, serve as useful hypotheses that can be subjected to future research.

Notes
1. Studies of gender inequality have skyrocketed over the last decades, from receiving little recognition in mainstream research to a situation in which the field is recognized as a key research area.It is indicative that the United Nations has included gender equality among its 18 goals of sustainability.2. Here, we particularly acknowledge the forceful argumentation and thorough empirical application of the causal force of public opinion on government behaviour by Brooks and Manza (2007).3.In the ISSP, the German data is divided into two strata -Germany-east and Germany-westand, due to the differing historical and social legacy of the two parts (Edlund & Öun, 2016), we treat them as two units in the analysis whenever applicable.4. The exact age of the child is not indicated in the question text, which suggests that people may have slightly different interpretations of the prototypical family.However, one important reason for using 'under school age' is that the years before the child enters compulsory school, it is up to the family to decide on who should take care of the child during daytime; moreover, the age for starting compulsory school varies across countries.5.In this part of the analysis, Germany-east is excluded, as the macro data refer to the period past the reunification of Germany.6.In deciding the key parameters and their weights for the CPS, there is an element of arbitrariness involved.To decrease the risk that the ordering of countries depends on a specific measurement of CPS, we developed four different measures for each country based on its cluster membership probabilities.The second measure is similar to the first (CPS-1), except that cluster 4 is regarded as the most progressive alternative: CPS-2 = (C1*0.50)+ C4 + (C5*0.75).The third measure separates the two most progressive models from the three housewife models and measures the balance between conservatism and progression: CPS-3 = (C4 + C5)−(C2 + C3 + C6).The fourth measure is similar to the second, except that different weights are applied: CPS-4 = (C1*0.33)+ C4 + (C5*0.67).The correlations between these four measures are very strong, with a lowest value of r = 0.99, indicating that they capture the same phenomena.7. We also tested the correlations between the macro-indicators and the other three CPS measures.The results were close to identical.8.The social cleavage score (SCS) is measured as the degree to which the independent variables' regression coefficients deviate from zero in the expected direction, for each of the six clusters in each country.The SCS is the average sum of deviations across clusters.Scores are shown in Table 2.

Table 1 .
Characteristics of six family model ideals.Cell entries are percentages.