The gender gap in welfare state attitudes in Europe: The role of unpaid labour and family policy

Abstract

Previous research has shown a prevailing ‘modern gender gap’ in socio-political attitudes in advanced capitalist economies. While numerous studies have confirmed gender differences in attitudes towards the welfare state in Europe, few have addressed the reason for this rift in men’s and women’s views about the role of government in ensuring the general welfare of citizens. In this article, I examine the relationship between gender equality in unpaid labour, family policy and the gender gap in welfare state attitudes. Based on data from 21 countries participating in the European Social Survey (ESS) Round 4, and using a mix of country- and individual-level regression models and multilevel models, I find that there is a clear relationship between country-level gender equality in unpaid labour and gender differences in support of an encompassing welfare state. A more equal distribution of unpaid care and domestic work correlates with women being increasingly supportive of a large and encompassing welfare state, in comparison with men. This pattern holds when controlling for individual-level economic risk and resources, cultural factors such as trust and social values traditionally related to the support of an encompassing welfare state, and beliefs about welfare state efficiency and consequences for society in general. This pattern is evident for countries with a low level of familistic policies, while no distinguishable pattern is discernible for highly familistic countries. These findings have implications for the perception of gender as an emergent social cleavage with respect to welfare state attitudes. The results are discussed in the light of institutional theories on policy feedback, familism, social role theory and previous findings relating to modernization theory and ‘gender realignment’.

Keywords

Attitudes comparative research division of labour family policy gender gap gender roles welfare state

Introduction

Over the last two decades, evidence has mounted in support of a ‘modern gender gap’ in political attitudes. Women in advanced capitalist economies are generally more supportive of leftist parties and policies than are men; this includes support for an encompassing role of the state in the redistribution of resources and provision of welfare (e.g. Inglehart, 2018; Inglehart and Norris, 2003). Systematic gender differences have been found to affect the overall support of policies and to be decisive in the outcome of elections (e.g. Abendschön and Steinmetz, 2014; Diekman and Schneider, 2010; Inglehart and Norris, 2003; Iversen and Rosenbluth, 2006). While most research has confirmed a gender gap in attitudes towards the welfare state in Europe, few studies go beyond treating gender as a covariate to statistical analysis.

One possible explanation, offered by social role theory, is that lingering inequalities in the division of unpaid labour maintain gender differences in the life patterns, norms and experiences that shape attitudes (e.g. Eagly et al., 2004). However, previous research within the modernization literature on ‘gender realignment’ has demonstrated a widening gender gap in ideology in Europe since the 1990s (Inglehart and Norris, 2003), even though women have been performing less unpaid labour during the same period (Hook, 2010). To add further complexity, comparative research has shown that any relationship between the gender division of labour and gendered attitudinal outcomes may be subject to moderation by the totality of the institutional context (Edlund and Öun, 2016).

In this article, I draw upon social role theory, modernization theory and institutional theory to explore the possible connection between unpaid labour, family policy and gender differences in welfare state attitudes. What is the nature of the relationship between the division of unpaid labour and the gender gap in welfare state attitudes? What role, if any, does institutional context in the form of family policy play in shaping this relationship? The analysis will reveal whether the gender gap in welfare state attitudes varies between European countries, if such a variation relates to the countries’ division of unpaid labour, and whether such a relationship is moderated by the institutional context. Using ordinary least squares (OLS) and multilevel regression, I examine the gender gap in overall preference for state responsibility in providing for jobs, healthcare, standard of living for the old and unemployed, childcare and paid leave for the care of sick family members, among 21 countries participating in the European Social Survey (ESS), Round 4.

Next, I offer a literature review structured around the relationship between the division of unpaid labour and welfare state attitudes, and the potential moderating effect of family policy. I then state the hypotheses and disclose the full analytical strategy before finally reporting and discussing the results.

Socio-political attitudes and the gender division of unpaid labour

Previous research has shown that women in advanced capitalist societies are more supportive than their male peers of an encompassing role of the state in the redistribution and provision of welfare services (e.g. Inglehart, 2018; Inglehart and Norris, 2003). Nearly all studies utilizing gender as a covariate have shown a positive association between gender and the support of state responsibility in welfare provision in Europe. A study of 20 countries participating in the International Social Survey Programme (ISSP) found women to be more likely than men to believe that the government should be responsible for social support, and that the government, rather than the family, should carry the costs of caregiving (Mair et al., 2016). A similar study of 24 countries found women to be more supportive than men of welfare policies benefitting those who are unemployed, sick or old (Blekesaune and Quadango, 2003). A study exploring attitudes towards welfare services and social security benefits in Finland found that gender as a social cleavage matched the effects of traditional explanatory factors such as socio-economic status or education (Muuri, 2010).

One general approach to gender differences in socio-political attitudes is to explain them by means of internalized gender roles, rooted in the gender division of labour in society (Diekman and Schneider, 2010; Wood and Eagly, 2012). Role theory stipulates that the observations of men and women performing different everyday tasks are generally attributed to innate differences aligned with the broad grouping of traits into communal (e.g. warm, nurturing and relation-oriented) and agentic (e.g. instrumental, assertive and achievement-oriented) domains. These attributions then make up broad societal expectations about how most men and women usually behave and should behave. This leads to the internalization and integration of such attributions into identity and behaviour through psychological and biosocial processes such as self-stereotyping and the situational elicitation of hormones (Wood and Eagly, 2012). Correspondingly, findings within the field of political psychology generally identify gender differences in attitudes in egalitarian or collective-oriented policies, versus policies rooted in system justification that legitimize social inequalities (e.g. Eagly et al., 2004; Sidianus et al., 2004).

As gender roles are argued to be influential above and beyond specific roles (e.g. occupational or parental roles), internalized expectations guiding socio-political preferences such as welfare state attitudes are thought to be observable even when controlling for gender differences in occupational class, parental status, or associated values and beliefs (Diekman and Schneider, 2010). That is, when comparing women and men working the same jobs or having the same family situation, a gender effect should still be observable if trans-situational gender norms prevail. However, if the gender division of labour were to become more equal, thereby lessening the basis for gender-differentiated expectations and undermining the basis for gendered personalities, a gender convergence in socio-political attitudes would be expected to result (Diekman and Schneider, 2010). For the purpose of this article, I term this expectation the gender convergence hypothesis.

The social role perspective has shown merit in explaining longitudinal changes in gender stereotypes, with masculine and feminine traits showing signs of convergence (Donnelly and Twenge, 2017; reviews in Diekman and Schneider, 2010; Wood and Eagly, 2012). However, in a study in the United States using repeated cross-sectional data from 1973 until 1998, the gender gap in socio-political attitudes proved to be surprisingly stable (Eagly et al., 2004), despite major progress in both the labour market and civil rights. Eagly et al. (2004) argued that this was due to a lack of corresponding change in the division of unpaid labour, and that the goals inherent to domestic responsibilities – such as promoting the welfare of children and families – continue to exert influence on women’s socio-political attitudes.

Another perspective on attitudinal gender differences is the theorization on ‘gender realignment’ found within modernization research. This perspective builds on the observation that women have moved from a morally conservative, right-leaning position before the 1970s to a progressive position to the left of men from circa 1990 and onwards (Abendschön and Steinmetz, 2014; Inglehart, 2018; Inglehart and Norris, 2003; Iversen and Rosenbluth, 2006). Meanwhile, men’s ideology demonstrates trendless fluctuations (Inglehart and Norris, 2003). This ‘modern gender gap’ in political attitudes is thought to be driven by the psychological and economic autonomy of women from men, brought about by the erosion of the nuclear family, the entry of women into the labour force, changed views on gender roles and feminist political mobilization (Inglehart and Norris, 2003).

It is difficult to disentangle causality regarding gender egalitarianism and gender-differentiated attitudes. For example, views on gender roles change among both men and women, but more evidently among women (Inglehart and Norris, 2003), thus acting as both cause and symptom of the general gender gap in socio-political attitudes. In sum, women’s left-leaning political orientation occurs in lockstep with progression in gender equality, following a process of modernization. The reason why increased autonomy leads to women holding left-leaning political views, and not to a different result, is argued by Inglehart and Norris (2003) as being partly due to the fact that left-wing and right-wing politics often articulate different views on gender roles (with conservatives often being slightly more in favour of gender specialization). Furthermore, the large welfare state favoured by leftist parties plays a role in imbuing women with social and political rights (Inglehart and Norris, 2003). To this, I would like to add the argument that women – in comparison with men – have more to lose from welfare state retrenchment in modern economies, where all adults are expected to participate in paid work (e.g. Daly, 2011).

Although the processes of modernization contribute to greater gender equality (or at least changed ideals) overall (Inglehart and Norris, 2003), changes in the division of unpaid labour are largely due to women doing less unpaid work (Hook, 2010). The latter is most evident in encompassing welfare states, where it is partly due to some of the care responsibilities being shifted to the state (Hook, 2010). Due to lingering gender norms and economic inequalities, a rollback of existing welfare services would likely affect women’s access to means of redistribution more severely than men’s access (e.g. Korpi et al., 2013), especially within heterosexual families with young children. As such, I argue that the division of unpaid labour may serve as a proxy for the work–family balance in women’s everyday lives while also resulting in salient gender differences in terms of reliance on welfare state arrangements.

Thus, within an institutional perspective on public opinion (e.g. Svallfors, 2007), the division of unpaid labour can be understood as a gender-specific measure of welfare state performance. At the individual level, differences in welfare state attitudes between men and women may partly reflect (attentiveness to) differences in collective self-interest, as well as differences in identification and solidarity with welfare recipients. Previous research has demonstrated that support for encompassing welfare policies is higher among those who view welfare recipients as deserving – a view that is generally stronger among those who see themselves as likely to become reliant upon welfare services (Van Oorschot et al., 2017). Thus, the overall expectation is that equality in the division of unpaid labour is positively associated with the gender gap in attitudes, with women showing stronger support for an encompassing welfare state, in comparison with men. I term this the gender realignment hypothesis.

Policy, familism and variation across institutional settings

For some time, advanced capitalist economies have experienced a loss of industrial-era family wages, which has rapidly eroded the basis for a self-sustaining single-earner family model (e.g. Ciccia and Bleijenbergh, 2014; Daly, 2011; Oesch, 2013). The family policies adopted by states to tackle this loss have been shown to affect dependencies on a spouse for access to resources and claimant rights (e.g. Ciccia and Bleijenbergh, 2014; Daly, 2011; Korpi et al., 2013). Policy context has also been shown to affect the gender gap in attitudes. For example, the European countries with the lowest levels of support for parental responsibilities simultaneously display the largest gender polarization in policy preferences (Banducci et al., 2016). More importantly, other comparative research has shown that while the gender division of labour relates to differences in experiences of work–family conflict and attitudes towards gender roles, this is in turn moderated by family policy context (e.g. Edlund and Öun, 2016; Öun, 2013; Sjöberg, 2004). To be specific, in countries where family policies were modelled on gender specialization, there was relatively little attitudinal divergence around issues of gender roles and the division of labour. The opposite was true in contexts modelled on overlapping gender roles, where both men and women acted as workers and caregivers. The finding that large inequalities are not always viewed as problematic may be partially explained by the existence of policies that affect how citizens think about social and political issues (e.g. Svallfors, 2007). Enacted policies may well shape the perception of which social issues are deemed important (and which are not), as well as providing a mental framework for how state performance is to be evaluated in terms of what type of society is seen as ideal (or achievable).

In specific regards to family policy, scholars have conceptualized policies aimed at placing the responsibility for caregiving on the family – as opposed to on the state (or on market solutions) – as familistic (e.g. Leitner, 2003). Familism in this regard entails policies that promote intrafamilial dependencies (economic and other), often by the subsidization of single-earner families through tax credits and allowances, and by the promotion of gender specialization (e.g. through long maternity leave at low replacement levels) (Leitner, 2003; see review in Lohmann and Zagel, 2016). Thus, while familism certainly includes a normative component in its rationale, it goes beyond an aggregate gender egalitarianism/traditionalism approach by including policy restraints that enable or impair the realization of specific gender ideologies in everyday practice (e.g. Daly, 2011).

Another feature of familistic policies is that they make the family – rather than the individual – the primary unit for access to resources and welfare benefits (Daly, 2011). I argue that this feature makes welfare state attitudes less likely to diverge as a result of unequal power relations within the household, such as an unequal division of unpaid labour between men and women. Drawing on arguments about institutional feedback effects (e.g. Svallfors, 2007), I also propose that familistic policies help establish a mental frame in which an unequal division of unpaid labour is less likely to be perceived as problematic, thus making it a less evident source of attitudinal divergence.

Other policy strategies that signify defamilization/individualization or market-orientation – such as targeting fathers as caregivers, or pushing families to find market solutions for child care arrangements (Daly, 2011; Korpi et al., 2013; Leitner, 2003) – have the potential to spark tensions around the division of unpaid labour. That is, if parents have the option to fully share parental leave or if mothers are not subsidized for caregiving at home, the division of unpaid labour is more likely to generate discussion. Moreover, when claims on the welfare state are made by individuals rather than by family members, the possibility of a divergence of interests is more evident, especially to the extent that men and women experience differences in reliance on welfare arrangements for the functioning of everyday life. Thus, I argue that the relationship between the division of unpaid labour and the gender gap in welfare state attitudes should be most pronounced where familism is low, and least evident in highly familistic countries. I term this the moderation hypothesis.

Summary of hypotheses

Given that social role theory expects greater gender similarity in life patterns and experiences to be associated with less gender difference in socio-political attitudes, I formulate a first hypothesis as follows:

H1. High gender equality in unpaid labour is associated with small gender differences in support for an encompassing welfare state.

Conversely, combining modernization theories on gender realignment with institutional theories on public opinion, the expectation is that the greater autonomy of women in gender-equal societies – together with equality of unpaid labour being a gendered measure of welfare state performance – corresponds to a larger gender gap in welfare state attitudes. I pose the competing hypothesis:

H2. High gender equality in unpaid labour is associated with large gender differences in support for the welfare state, with women supporting an encompassing welfare state more than men do.

Finally, arguing that familistic policies reduce the potential of the division of unpaid labour as an area of conflict spurring attitudinal differences, I formulate the third hypothesis as follows:

H3. The relationship between gender equality in unpaid labour and the gender gap in support of the welfare state is most pronounced for countries characterized by low levels of familism and is least pronounced in highly familistic countries.

Data and method

I used individual-level data from a rotating module on welfare state attitudes included in the ESS, Round 4, collected in 2008/2009 (ESS Round 4: European Social Survey, 2018; ESS Round 4: European Social Survey Round 4 Data, 2008). The ESS is a biannually distributed, multi-country survey aimed at gauging values and attitudes among European populations. The data gathering was conducted through face-to-face standardized interviews, with strict random probability sampling and rigorous translation protocols. The response rates varied from 43 percent (Germany) to 81 percent (Cyprus). The achieved sample used in this study consisted of 38,319 individual respondents from 21 European countries.¹ I weighted the responses using post-stratification weights (and population size weights for any statistics based on the pooled sample, except for multilevel models). I also used indicators for policy and country-level outcomes from the European Institute for Gender Equality (EIGE, 2010), the Multilinks database on intergenerational policy and the Organisation for Economic Co-operation and Development (OECD).

Analytical strategy

The overall aim of this study was to gauge (1) how gender equality in the division of unpaid labour correlates with the gender gap in support of the welfare state at the country level and (2) whether any moderating effect on this relationship was posed by different levels of familism at the contextual level. I first constructed a set of single-country OLS regression models in which I saved the mean difference in support of the welfare state between women compared with men in each country, with and without individual-level covariates. I then ran OLS regression models using these country-level gender-gap variables as a dependent variable, and the country scores on division of unpaid labour as an independent variable. I ran three such models: one with a pooled sample consisting of all countries, and two with countries separated by level of familism (low/high). I then plotted the results using a simple scatter plot in order to illustrate each country’s position on the gender gap in welfare state support (WSS) by gender equality in division of unpaid labour.

In order to assess whether men and women were differentially responsive to variation in the equality of unpaid labour, I also ran a set of linear multilevel models with a random slope and intercept for gender, and a three-way interaction term between gender, country-level gender equality in division of unpaid labour, and level of familism (e.g. Hox, 2010).² Besides determining the simple slopes for men and women within different contexts, the aim of these models was to estimate whether there was sufficient reduction in the variance of the random slope for gender between countries. Missing values³ were dealt with using a multiple imputation approach (e.g. Van Buuren, 2011).⁴

Dependent variable

To construct the dependent variable, WSS, I followed the suggested guidelines in the ESS4 Welfare State module final template. The index measured attitudes towards government scope and responsibilities and was composed of the following statement and six questions:

People have different views on what the responsibilities of governments should or should not be. For each of the tasks I read out, please tell me on a score of 0–10 how much responsibility you think governments should have. 0 means it should not be the governments’ responsibility at all, and 10 means it should be entirely the governments’ responsibility. Firstly, to [. . .] ensure a job for everyone that wants one? [. . .] ensure adequate healthcare for the sick? [. . .] ensure a reasonable standard of living for the old? [. . .] ensure a reasonable standard of living for the unemployed? [. . .] ensure sufficient child care services for working parents? [. . .] provide paid leave from work temporarily for those who have to care for sick family members?

I scaled the index to range from 0 to 100, with the score representing the arithmetic mean of all items. A score of ‘100’ meant that all of the listed policy areas should be entirely the responsibility of governments, indicating support of an encompassing and far-ranging welfare state; ‘0’ indicated the opposite. For the grand mean, reliability (Cronbach’s α) and other measures, see Table 1.

Table 1.

Descriptive statistics.

Individual-level categorical variables			Individual-level continuous variables
Variable	Group	% (valid)		WSS	INT	IPT	Auth	Egal	GEgal	MC	PR	TP	SD	O/U	EC	MSC		Age	Edu	Woman
Gender	Women	54.2	(unweighted) n	36,862	36,769	37,892	37,692	37,306	37,497	35,364	25,796	33,398	36,741	33,028	33,822	35,318		38,319	38,319	38,311
	Men	45.8	Min	0	0	0	0	0	0	0	0	0	0	0	0	0		20	0	0
Class and labour market status	High	28.0	Max	100	10	10	10	10	10	10	10	10	10	10	10	10		99	30	1
	Intermediate	20.7	Mean	77.371	4.856	4.913	7.658	5.377	5.731	5.179	3.095	4.149	5.098	5.731	5.258	5.325		49.810	12.210	−
	Unemployed/housework	14.4	Std. dev.	15.598	2.275	2.045	1.907	1.999	2.551	2.225	2.209	1.647	2.222	1.689	1.555	1.866		17.323	4.218	−
	(Other)	5.0	Skew.	−0.609	−0.234	−0.237	−0.786	0.027	−0.234	−0.262	0.588	−0.092	−0.127	−0.054	−0.038	0		0.168	−0.057	−
	Low	31.9	SE skew.	0.013	0.013	0.013	0.013	0.013	0.013	0.013	0.015	0.013	0.013	0.013	0.013	0.013		0.013	0.013	−
Family situation	Single	34.2	No. of items	6	3	3	2	2	2	4	2	4	2	4	4	4		−	−	−
	Single parent	9.5	Cronbach’s α	0.836	0.800	0.833	−	−	−	0.557	−	0.704	−	0.551	0.402	0.666		−	−	−
	Cohabiting	26.0	Correlation with WSS	1	−0.098**	−0.094**	0.158**	0.258**	−0.070**	−0.01	0.108**	−0.236**	−0.071**	0.052**	0.103**	0.183**		0.013*	−0.171**	0.076**
	Cohabiting parent	30.3
Country-level variables												Country-level correlations
		N	Min	Max	Mean		Std. dev.	Skew		SE skew.		WSS gender gap, no controls					WSS gender gap, controlled for all individual-level covariates
Gender equality, division of unpaid labour 2007 (0–100)		21	34.18	84.59	65.786		13.04	−0.676		0.501		0.436*					0.436*
Familism, average 2004–2009 (0–100)		21	12.18	86.37	46.618		19.628	0.280		0.501		0.179					0.178
Low familism (0–49.9)		14	0	1	–		−	–		–		–					–
High familism (50–100)		7	0	1	–		−	–		–		–					–

WSS: welfare state support; INT: institutional trust; IPT: interpersonal trust; Egal: egalitarianism; Auth: authoritarianism; GEgal: gender egalitarianism; MC: multiculturalism; PR: perceived risk of welfare dependency; TP: task performance; SD: service delivery; O/U: overuse/underuse perceptions; EC: economic consequences; MSC: moral/social consequences; Edu: years of education; Std. dev.: standard deviation; SE: standard error; skew: skewness.

All except correlations are original data. Correlations (Pearson’s r) are based on multiple imputations and weighted with post-stratification weights and for population size.

p < 0.01; *p < 0.05.

Country-level variables

To measure gender equality in the division of unpaid labour, I used a sub-index taken from the EIGE that measured the relative gender gap in the performance of care and domestic work in the year 2007. The index was based on the difference between the percentage of men and women, respectively, self-reporting daily involvement (occurrence) in caring for and/or educating their children or grandchildren, caring for elderly and disabled people, and cooking and housework outside of paid work.⁵ The index ranged from 0 to 100, with higher values indicating greater gender equality in the relative division of unpaid labour between men and women. Because the index did not include actual levels of unpaid work (e.g. fixed time units), it was insensitive to between-country differences in the total amount of unpaid work performed. Thus, it was suitable for cross-country comparison without further adjustment.

To measure the level of familism, I followed the guidelines available in the work of Lohmann and Zagel (2016) to construct an index based on information on allowance and taxes,⁶ parental leave⁷ and support of the elderly.⁸ These indicators were based on an extensive review of the empirical and theoretical literature on the subject, including a comparison of previously used indices. I constructed the scale to range from 0 to 100, with higher scores indicating a greater degree of familism. Finally, I dichotomized the index to measure high and low levels of familism, respectively. The cut-off point was at the theoretical middle of the scale (again following the procedures in Lohmann and Zagel, 2016). All indicators were taken from the Multilinks database on intergenerational policy, and averaged for the years 2004 and 2009. For the binary variables used to construct the index, the latest available measurement point was used. The reason for averaging years was to include as many cases as possible, as several countries were missing information for one or more indicators for one of the two measurement points.

Individual-level covariates

To capture variation in risk and resources, I used information on current or last held occupation to construct a combined class and labour market variable based on the three-category version (high–intermediate–low) of the European socio-economic classification (ESeC 2008 version). Unemployed respondents and those listed as performing housework (main activity in the last 7 days) formed a separate category, in order to explore the effect of being outside the labour market. The category ‘other’ was composed of all other situations with no known occupational history. I also constructed a family situation variable by combining information on cohabitation and having children living at home. Furthermore, I included a set of variables measuring trust and social values. These were indices (0–10) measuring interpersonal and institutional trust, egalitarianism, authoritarianism, gender egalitarianism and multiculturalism. These are all included in the ESS4 Welfare State module and have been specifically proposed as individual predispositions that are explanatory of support of the welfare state.

Finally, I constructed five indices measuring beliefs about welfare state efficiency: task performance, service delivery, overuse/underuse, economic consequences and moral/social consequences.⁹ I constructed each index to range from 0 to 10, with ‘10’ indicating fair and efficient workings of the welfare state along with overall positive consequences for society and ‘0’ indicating the opposite. The number of sub-items and the reliability analysis (Cronbach’s α) are reported in Table 1. All individual-level continuous covariates were centred on the country mean in order to ease interpretation of the intercept.

Results

First, among the 21 studied countries, women are more in favour of an encompassing welfare state than men are. The average mean difference between women and men is equal to 2.07 points on the index for WSS (Table 2, M2:0). Furthermore, the null model in Table 2 indicates that the between-country variation represented by the random coefficient for gender (‘woman’) is statistically significant (p < 0.05), which warrants further investigation of the matter.

Table 2.

Welfare state support 0–100.

		M2:0	M2:1	M2:2	M2:3	M2:4	M2:5	M2:6
Individual-level fixed effects
Intercept		76.17*** (1.20)	89.75*** (5.85)	89.95*** (5.86)	76.93*** (1.47)	76.92*** (1.48)	92.18*** (5.70)	92.91*** (5.68)
Woman		2.07*** (0.24)	2.07*** (0.24)	−0.68 (1.11)	2.07*** (0.24)	2.09*** (.30)	−0.85 (1.24)	−1.31 (1.10)
Country-level fixed effects
Division of unpaid labour			−0.20* (0.09)	−0.18* (0.09)			−0.24** (.09)	−0.23* (0.09)
Familism: High					−2.29 (2.54)	−2.25 (2.56)	−40.38* (17.33)	−40.15* (17.24)
Familism: Low (ref.)					−	−	−	−
Cross-level interactions
Woman*	Division of unpaid labour			0.04* (0.02)			0.05* (0.02)	0.04* (0.02)
	Familism: High					−0.06 (0.52)	1.44 (3.88)	1.23 (3.46)
	Familism: Low (ref.)						−	−
Familism*	Division of unpaid labour						0.57* (0.25)	0.57* (0.25)
FamilismWoman	Division of unpaid labour						−0.02 (0.06)	−0.02 (0.05)
Random parameters
Variance	Within groups	211.05** (1.56)*	211.05** (1.56)*	211.05** (1.56)*	211.05** (1.56)*	211.05** (1.56)*	211.05** (1.56)*	185.74** (1.40)*
Variance	Between groups	30.03* (9.56)*	25.91* (8.53)*	25.77* (8.45)*	30.34* (9.92)*	30.37* (9.94)*	21.66* (7.52)*	21.47* (7.44)*
Random coefficient	Woman	0.72 (0.38)*	0.72 (0.38)*	0.47 (0.31)	0.72 (0.38)*	0.78 (0.41)*	0.57 (0.36)	0.39 (0.26)
Model information	−2LL	313,785.84	313,785.17	313,785.78	313,781.31	313,780.69	313,780.74	307,276.74
n		38,311	38,311	38,311	38,311	38,311	38,311	38,311

Individual- and country-level fixed effects and cross-level interactions with random intercept and slopes.

M2:6 includes all individual-level covariates. Parameters are pooled averages of multiple imputed data. Model information criterion is based on the last iteration. All individual-level continuous variables are centred on the country mean.

***

p < 0.001, **p < 0.01, *p < 0.05 two-tailed (sig. for variance components are one-tailed).

I begin by illustrating the relationship between the gender gap in WSS (controlled for all individual-level covariates) and gender equality in the division of unpaid labour, in countries separated by their levels of familism (high/low). These relationships are illustrated in the scatter plot in Figure 1. A positive difference indicates that women score higher on WSS than men do. Within the group of countries with low levels of familistic policies, about 44.3 percent (r² = 0.443) of the variation in the gender gap in support of the welfare state is explained by variation in equality in the division of unpaid labour. The more equal the division, the larger the gender gap. Conversely, among the countries with high levels of familistic policies in place, the variation in the gender gap in support of the welfare state is not explained by equality in the division of unpaid labour (r² = 0.007). Thus far, these results are in favour of hypotheses H2 and H3, and in negation of hypothesis H1.

Figure 1.

Scatter plot showing predicted mean gender difference in support of the welfare state by gender equality in division of unpaid labour (controlled for all individual-level covariates). Positive scores indicate that women score higher than men do. Pooled sample (21 countries): R = 0.436; R² = 0.190; b = 0.032; sig. = 0.048; Low familism (14 countries): R = 0.665; R² = 0.443; b = 0.037; sig. = 0.009; high familism (seven countries): R = 0.084; R² = 0.007; b = 0.013; sig. = 0.858.

In Figure 2, I have plotted the estimated marginal means for men and women in support of the welfare state by gender equality in unpaid labour, clustered within countries based on the level of familism. The illustrated simple slopes are based on the multilevel random intercept-random slope model (M2:6, Table 2), which includes a three-way cross-level interaction term between gender (‘woman’), country level of gender equality in unpaid labour and level of familism (high/low).¹⁰ First, looking only at countries characterized by low levels of familistic policies, there is an overall negative association between gender equality in unpaid labour and support of the welfare state (b = −0.23; p < 0.05). This is illustrated by the negative direction for the simple slopes for men and women (Figure 2, left column). However, the negative association is weaker for women, resulting in a positive difference (b = 0.04; p < 0.05) when compared with men. At the minimum observed value of gender equality in unpaid labour (corresponding to Greece), there is virtually no difference between men and women in the estimated marginal mean support of the welfare state. At the maximum observed value for equality in division of unpaid labour (corresponding to Sweden), women are significantly more positive when compared with men (by an estimated 3.17-point difference).

Figure 2.

Support of the welfare state by gender equality in division of unpaid labour. Estimated marginal means, simple slopes for gender grouped within country level of familism. All individual-level covariates are centred on the country mean. Data points represent valid country scores for division of unpaid labour. Plotted relationships are based on the 10th imputation of missing values. Error bars are 95 percent CI.

Conversely, looking at countries with a high level of familistic policies (Figure 2, right column), there is a positive association between gender equality in unpaid labour and support of the welfare state. However, the difference in slopes between men and women is very small and is not statistically significant. That is, while women are overall more positive towards an encompassing welfare state, there is no observable variation in the difference between men and women that is associated with the division of unpaid labour.

Finally, there is a significant reduction in the variance between countries in the random coefficient for gender (‘woman’) (from 0.72 to 0.39) between the null model in M2:0 and the three-way interaction model (M2:6). This means that the interplay between gender equality in the division of unpaid labour and contextual level familism contributes to explaining the between-country differences in the gender gap in support of the welfare state.

In summary, I find that the greater equality there is in the gender division of unpaid labour in a country, the greater the gender gap in support of the welfare state will be, with women being more positive than men. However, this relationship is most pronounced for countries with low levels of familistic policies, while this relationship is not evident in countries with high levels of familistic policies. Thus, H2 and H3 are confirmed, and H1 is rejected.

Robustness checks and sensitivity analysis

To safeguard against a spurious relationship, I opted to investigate both labour market characteristics and outcomes, as well as welfare state/economic factors, as possible explanatory factors of the attitudinal gender gap. Overall, the labour market¹¹ and welfare state/economic¹² factors do not contribute to explaining more of the gender gap in support of the welfare state than the gender division of labour does. Although some measures correlate moderately with the gender gap in welfare state attitudes, these results are not statistically significant.¹³

As an alternative contextual analysis, I grouped countries by the tax rate for second earners compared with singles.¹⁴ In the group of countries with a tax ratio below 1.2 (corresponding to low familism), there is a large and positive correlation (r = 0.546) between the division of unpaid labour and the gender gap in welfare state attitudes. The correlation is weaker (r = 0.214) for the group of countries with a tax ratio of 1.3 or above. However, none of the results are statistically significant. I also opted to group countries by the level of gender egalitarianism.¹⁵ Within both highly gender-egalitarian countries (theoretically corresponding to low familism) and countries with low levels of gender egalitarianism, there is a positive correlation between gender equality in the division of unpaid labour. The correlation is slightly weaker (r = 0.398) within highly gender-egalitarian countries compared with countries with lower levels of gender egalitarianism (r = 0.663). However, none of the results are statistically significant. As for the relationship between familism and gender egalitarianism, this correlation is weak (r = 0.106) and not statistically significant, implying that these variables do not measure the same phenomenon.

To check for the influence of outliers, I re-ran all the country-level models while excluding countries one at a time. The correlations between the division of unpaid labour and the gender gap in welfare state attitudes within a low familism context vary between r = 0.588 (sig. = 0.035)¹⁶ and r = 0.785 (sig. = 0.001),¹⁷ which confirms the results for the group of countries marked by low familism. Similarly, the correlations in a high familism context vary between r = −0.183 (sig. = 0.729)¹⁸ and r = 0.216 (sig. = 0.681).¹⁹ In sum, these results do not alter the overall conclusions from the models based on the full sample of countries.

Discussion and conclusions

In this research, I set out to explore the gender gap in support of the welfare state in 21 European countries. In particular, I expected gender equality in the division of unpaid labour to be associated with gender differences in support of the welfare state. I further expected this relationship to be moderated by levels of familism at the contextual level. First, this study showed that in general, women support an encompassing welfare state to a greater degree than men do (b = 2.07, M2:0), and by levels that well match the levels of traditionally explored factors such as class and labour market status. This finding is in line with previous findings within the literature on welfare state attitudes (e.g. Blekesaune and Quadango, 2003).

At the country level, role theory predicted that gender equality in unpaid labour would be associated with gender similarity in socio-political attitudes, due to an overlap of life patterns and role expectations (Diekman and Schneider, 2010). However, the present findings show exactly the opposite. Gender equality in unpaid labour is positively associated with gender differences in support of the welfare state. One possible explanation is that variation in the division of unpaid labour is due to women doing less (Hook, 2010) and thus is not challenging gender roles to the extent that new norms are formed and internalized into personality. However, while such reasoning might explain a status quo scenario, it provides fewer clues to the positive association between the division of unpaid labour and the gender gap in support of an encompassing welfare state.

Instead, these results align better with the predictions that were made by combining modernization theory and institutional theory on public opinion (e.g. Inglehart and Norris, 2003; Svallfors, 2007; Van Oorschot et al., 2017). This perspective holds that the positive impact on women’s everyday lives that results from a more equal division of unpaid labour should correspond to women being more supportive of an encompassing welfare state, in comparison with men – especially if part of the reason for this progress is a shift of some of the care responsibilities to the state, rather than altered gender roles for men (e.g. Hook, 2010). This reasoning makes sense, supposing that a rollback of welfare services would affect women more negatively than men (e.g. Korpi et al., 2013), as a risk of welfare dependency is known to prompt feelings of solidarity with vulnerable groups and increase support of an encompassing welfare state (e.g. Van et al., 2017). While the specific mechanisms pertaining to collective self-interest, identification and solidarity with welfare recipients remain to be properly operationalized, this prediction turned out to be accurate.

Regarding the institutional context, I argued that familistic policies (e.g. Lohmann and Zagel, 2016) downplay the division of unpaid labour as a source of attitudinal divergence along gender lines, in particular by tying access to benefits and resources to the family, rather than to individuals, and by using tax credits and allowances to subsidize a single-earner family model. Furthermore, the emphasis on gender specialization that is conveyed by, for example, long maternal leave at low replacement levels, promotes a certain mental framework for the evaluation of welfare state performance (e.g. Svallfors, 2007). In other words, an unequal division of unpaid labour is not necessarily seen as drifting from the type of society that is being conveyed as ideal or achievable. Conversely, I argued that both dual-earner policies, which force families to seek market solutions for care, and dual-carer policies, which target the role of both parents in caregiving (e.g. Korpi et al., 2013), highlight the gender division of labour as a potential area of conflict.

Having observed no specific pattern for familistic countries, this study shows that the positive association between equality in the division of unpaid labour and gender differences in support of the welfare state is only valid for countries that do not pursue familistic policies. This finding is similar to previous research that has found that the relationship between gender attitudinal divergence and the division of labour is weaker when family policy is modelled on gender specialization and stronger when the state promotes dual or overlapping roles for men and women (e.g. Edlund and Öun, 2016; Öun, 2013).

On a final note, it is inherently problematic to speak of causality within the framework of a cross-sectional study. Thus, it is only possible to speak of the evident association between gender equality in unpaid labour and the gender gap in support of the welfare state. It remains to be determined whether the division of unpaid labour causes attitudinal divergence between men and women – perhaps as a case of spiralling causality. In order to gauge causality, I suggest that future research should focus on longitudinal studies of countries with a history of progression in the division of unpaid labour.

Footnotes

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This work was supported by the Swedish Research Council for Health, Working Life and Welfare (grant number 2016-00255).

Notes

References

Abendschön

Steinmetz

(2014) ‘The Gender Gap in Voting Revisited: Women’s Party Preferences in a European Context’, Social Politics 21(2): 315–44.

Banducci

Elder

Greene

Stevens

(2016) ‘Parenthood and the Polarisation of Political Attitudes in Europe’, European Journal of Political Research 55: 745–66.

Blekesaune

Quadango

(2003) ‘Public Attitudes toward Welfare State Policies: A Comparative Analysis of 24 Nations’, European Sociological Review 19(5): 415–27.

Ciccia

Bleijenbergh

(2014) ‘After the Male Breadwinner Model? Childcare Services and the Division of Labor in European Countries’, Social Politics: International Studies in Gender, State & Society 21(1): 50–79.

Daly

(2011) ‘What Adult Worker Model? A Critical Look at Recent Social Policy Reform in Europe from a Gender and Family Perspective’, Social Politics: International Studies in Gender, State & Society 18(1): 1–23.

Diekman

A.B.

Schneider

M.C.

(2010) ‘A Social Role Theory Perspective on Gender Gaps in Political Attitudes’, Psychology of Women Quarterly 34: 486–97.

Donnelly

Twenge

J.M.

(2017) ‘Masculine and Feminine Traits on the Bem Sex-Role Inventory, 1993–2012: A Cross-Temporal Meta-Analysis’, Sex Roles: A Journal of Research 76(9–10): 556–65.

Eagly

A.H.

Diekman

A.B.

Johannesen-Schmidt

M.C.

Koenig

A.M.

(2004) ‘Gender Gaps in Socio-Political Attitudes: A Social Psychological Analysis’, Journal of Personality and Social Psychology 87(6): 796–816.

Edlund

Öun

(2016) ‘Who Should Work and Who Should Care? Attitudes towards the Desirable Division of Labour between Mothers and Fathers in Five European Countries’, Acta Sociologica 59(2): 151–69.

10.

ESS Round 4: European Social Survey (2018) ESS-4 2008 Documentation Report. Edition 5.5. Bergen, European Social Survey Data Archive, NSD – Norwegian Centre for Research Data for ESS ERIC.

11.

ESS Round 4: European Social Survey Round 4 Data (2008) Data File Edition 4.5. NSD – Norwegian Centre for Research Data, Norway – Data Archive and Distributor of ESS Data for ESS ERIC.

12.

European Institute for Gender Equality (EIGE). (2010) Gender Equality Index 2010, available at https://eige.europa.eu/gender-equality-index/2010/domain/time.

13.

Hook

J.L.

(2010) ‘Gender Inequality in the Welfare State: Sex Segregation in Housework, 1965–2003’, American Journal of Sociology 115(5): 1480–523.

14.

Hox

J.J.

(2010) Multilevel Analysis, Techniques and Applications. New York: Routledge.

15.

Inglehart

R.F.

(2018) Cultural Revolution: People’s Motivation Are Changing and Reshaping the World. Cambridge: Cambridge University Press.

16.

Inglehart

R.F.

Norris

(2003) Rising Tide: Gender Equality and Cultural Change around the World. Cambridge: Cambridge University Press.

17.

Iversen

Rosenbluth

(2006) ‘The Political Economy of Gender: Explaining Cross-National Variation in the Gender Division of Labor and the Gender Voting Gap’, American Journal of Political Science 50(1): 1–19.

18.

Korpi

Ferrarini

Englund

(2013) ‘Women’s Opportunities under Different Family Policy Constellations: Gender, Class, and Inequality Tradeoffs in Western Countries Re-examined’, Social Politics: International Studies in Gender, State & Society 20(1): 1–40.

19.

Leitner

(2003) ‘Varieties of Familialism: The Caring Function of the Family in Comparative Perspective’, European Societies 5(4): 353–75.

20.

Lohmann

Zagel

(2016) ‘Family Policy in Comparative Perspective: The Concepts and Measurement of Familization and Defamilization’, Journal of European Social Policy 26(1): 48–65.

21.

Mair

C.A.

Chen

Liu

Brauer

J.R.

(2016) ‘Who in the World Cares? Gender Gaps in Attitudes Support for Older Adults in 20 Nations’, Social Forces 95(1): 411–38.

22.

Muuri

(2010) ‘The Impact of the Use of the Social Welfare Services or Social Security Benefits on Attitudes to Social Welfare Policies’, International Journal of Social Welfare 19: 182–93.

23.

Nikolka

(2016) ‘Taxation and Female Labor Supply in OECD Countries’, Journal of Institutional Comparisons 14(2): 55–8.

24.

Oesch

(2013) Occupational Change in Europe: How Technology and Education Transform the Job Structure. Oxford: Oxford University Press.

25.

Öun

(2013) ‘Is It Fair to Share? Perceptions of Fairness in the Division of Housework among Couples in 22 Countries’, Social Justice Research 26(4): 400–21.

26.

Sidianus

Pratto

Van Laar

Levin

(2004) ‘Social Dominance Theory: Its Agenda and Method’, Political Psychology 25(6): 845–80.

27.

Sjöberg

(2004) ‘The Role of Family Policy Institutions in Explaining Gender-Role Attitudes: A Comparative Multilevel Analysis of Thirteen Industrialized Countries’, Journal of European Social Policy 14(2): 107–23.

28.

Svallfors

(2007) The Political Sociology of the Welfare State. Palo Alto, CA: Stanford University Press.

29.

Van Buuren

. (2011) ‘Multiple Imputation of Multilevel Data’, in Hox

J.J.

Roberts

(eds) Handbook of Advanced Multilevel Analysis, pp. 173–96. New York: Routledge.

30.

Van Oorschot

Roosma

Meuleman

Reeskens

. (2017) The Social Legitimacy of Target-ed Welfare: Attitudes to Welfare Deservingness. Cheltenham: Edward Elgar.

31.

Wood

Eagly

A.H.

(2012) ‘Biosocial Construction of Sex Differences and Similarities in Behavior’, in Olson

J.M.

Zanna

M.P.

(eds) Advances in Experimental Social Psychology, Vol. 46, pp. 55–123. Burlington, NJ: Academic Press.