Gender differences in the quality of leisure: a cross-national comparison

ABSTRACT

Considerable work-family research has investigated the gendered division of work and care. Gender differences in leisure time have received much less attention from work-family scholars, despite the potential importance of such inequalities for women’s quality of life. Combining key insights from the substantial gendered leisure studies literature with work-family scholarship, the current study examines cross-national variation in gender differences in leisure quality. Using data on 23 countries from the 2007 International Social Survey Program, we expected that women’s leisure quality would be lower than men’s, but the gender gap would be smaller in countries with more gender egalitarian attitudes and divisions of care (via de-familialisation and paternity leave) and where women have more bargaining power. Our results show that these country characteristics moderate the association between gender and the extent to which free time is used to relax and recover. In countries with conservative gender norms, low levels of childcare coverage, limited paternity leave and lower political power for women, women’s leisure quality is lower than men’s. In more egalitarian countries, the gender gap in leisure quality is lower and in some cases, reversed. These results are in line with findings from cross-national research on the gendered division of labor and offer an important contribution to understanding gender differences in leisure quality across countries.

RESUMEN

Numerosos estudios sobre la conciliación de la vida laboral y familiar han investigado la distribución por sexos del trabajo remunerado y las tareas domésticas y de cuidado. Las diferencias de género en tiempo de ocio han recibido mucha menos atención de los investigadores sobre conciliación, pese a la importancia potencial de esas desigualdades para la calidad de vida de las mujeres. Combinando información clave de la extensa literatura de estudios sobre ocio por género con estudios sobre conciliación de la vida laboral y familiar, el estudio analiza las diferencias de género entre países en calidad de ocio. Partiendo de los datos de 24 países del Programa Internacional de Encuestas Sociales 2007, esperábamos que la calidad de ocio de las mujeres fuese inferior a la de los hombres, pero la brecha entre ambos sexos es menor en países que promueven la igualdad de género y el reparto de tareas domésticas y de cuidado (a través de la desfamiliarización y permisos de paternidad) y donde las mujeres tienen mayor poder de negociación. Nuestros resultados muestran que las características de esos países moderan la asociación entre género y el grado en que se utiliza el tiempo libre para relajarse y recuperarse. En países con normas de género más conservadoras, baja cobertura de servicios de guardería, permisos de paternidad limitados y menor poder político para las mujeres, la calidad de ocio de las mujeres es inferior a la de los hombres. En países más igualitarios, la brecha entre sexos en calidad de ocio es inferior y, en algunos casos, se ha invertido. Estos resultados coinciden con las conclusiones de estudios transnacionales sobre la división del trabajo por sexos y contribuyen notablemente a la comprensión de las diferencias de género en calidad de ocio entre países.

KEYWORDS:

• cross-national research
• gender
• leisure
• multilevel analysis
• time pressure

PALABRAS CLAVE:

• Estudios transnacionales
• género
• ocio
• análisis multinivel
• presión de tiempo

Introduction

Considerable work-family and time use research has investigated the gendered division of work and care (Bianchi & Milkie, 2010), but gender differences in leisure time have received much less attention. Within this limited scholarship, gender differences in leisure are seen as indicative of enduring forms of gender inequality across work and care domains, contributing to differences in men and women’s quality of life (Bittman, England, Sayer, Folbre, & Matheson, 2003; Mattingly & Blanchi, 2003). More frequent and higher quality leisure is associated with better and improved quality of life, health, and wellbeing (Bittman et al., 2003; Brajša-Žganec, Merkaš, & Šverko, 2011; Chatzitheochari & Arber, 2012; Gimenez-Nadal & Sevilla-Sanz, 2011; OECD, 2009). In contrast, gender inequalities in access to and experiences of leisure have been a central focus in leisure research. Leisure scholars demonstrate the ways in which gender uniquely constrains men and women’s leisure as well as the way in which leisure is shaped by societal gender relations (Henderson & Gibson, 2013; Henderson & Hickerson, 2007; e.g. Henderson, Hodges, & Kivel, 2002; Shaw, 1994). Drawing on these two literatures, we address an important shortcoming in understanding gender differences in leisure by quantitatively assessing the role of national practices, policies, and norms in explaining differences in the quality of leisure cross-nationally.

Existing studies on gender inequality in leisure highlight differences associated with gender norms and time constraints (Henderson & Gibson, 2013; Henderson & Hickerson, 2007; Henderson et al., 2002; Shaw, 1994). For example, it is argued that women have less and lower quality leisure because their total work load is higher and they have more responsibility for the care of others (Chatzitheochari & Arber, 2012; Gimenez-Nadal & Sevilla-Sanz, 2011; Haller, Hadler, & Kaup, 2013; Henderson & Hickerson, 2007; Nomaguchi & Bianchi, 2004). In addition, research suggests significant cross-country differences exist in the relationship between gender and leisure, for example in time spent in leisure (Craig & Mullan, 2013; Mattingly & Blanchi, 2003; OECD, 2009; Sayer, England, Bittman, & Bianchi, 2009) and how socio-cultural norms shape men and women’s leisure.

Chatzitheochari and Arber (2012, p. 468) argue that measures focused on the amount of leisure time obscure gender differences in the quality of leisure. ‘We find that women’s disadvantaged position is worsened by the quality of their free time, which is lower than men’s due to their ongoing domestic and parental responsibilities.’ This finding highlights an important understudied phenomenon – the gendered nature of leisure quality. The limited work-family and time use research available on this topic suggests that country context is important for men and women’s leisure quality, highlighting the need for more cross-national studies (Bittman et al., 2003; Craig & Mullan, 2013; Haller et al., 2013; Mattingly & Blanchi, 2003). Craig and Mullan (2013) speculate that gender differences in leisure quantity and quality are smaller in countries where men worked shorter hours and governments provided more family support. Yet we are not aware of any studies that quantitatively test the relationship between country-level characteristics and gender patterns in leisure quality.

This study addresses this gap in the literature by analyzing the role of national practices, policies, and norms in relation to the quality of individual leisure in 23 countries. Our central research question is which country characteristics underlie gender differences in leisure quality? Although comparisons of a small number of countries are insightful because they allow for an in-depth examination of the country context (Bittman et al., 2003; Craig & Mullan, 2013), the added value of a multilevel analysis is that it enables us to disentangle the impact of specific country characteristics. Specifically, it allows us to test interrelated effects at the macro (country) and micro (individual) level through cross-level interactions (Ruppanner, 2010). We test whether the impact of gender on leisure quality varies across countries and is moderated by country characteristics, such as childcare coverage. Our study builds on the work-family and time use literature focused on individual-level determinants of leisure quality (Chatzitheochari & Arber, 2012; Gimenez-Nadal & Sevilla-Sanz, 2011) and adds to existing gendered leisure studies (e.g. Henderson & Gibson, 2013) by quantitatively examining whether and how gender gaps in leisure quality are shaped by country context.

Explaining cross-national differences in leisure quality

Leisure is generally presented as an uncontested concept in work-family and time use literature. The underlying assumption is that leisure is positive and that more frequent leisure will lead to better quality of life, health, and wellbeing (Bittman et al., 2003; Brajša-Žganec et al., 2011; Chatzitheochari & Arber, 2012; Gimenez-Nadal & Sevilla-Sanz, 2011; OECD, 2009). Gendered leisure research, in contrast, problematizes the concept, suggesting leisure is neither inherently positive or negative and requires investigations that go beyond just the amount of leisure to examine the quality of leisure (Shaw, 1994). This suggests it is important to look beyond the amount of leisure and also consider the quality of men and women’s leisure experiences. Additionally, these studies provide important insights into how gender differences in leisure shape and are shaped by dominant gender norms in society (Henderson & Gibson, 2013; Henderson & Hickerson, 2007; e.g. Henderson et al., 2002; Shaw, 1994).

The recognition by leisure scholars that societal gender relations shape men and women’s access to and experiences of leisure has only been theorized in work-family research in limited measure. The few cross-national work-family studies on gender differences in leisure available focus on micro-level variations in leisure time without theorizing or testing macro-level explanations for differences in leisure quality (e.g. Bittman et al., 2003; Craig & Mullan, 2013). These studies argue leisure differences are due to variations in men and women’s time availability, time constraints, and norms. It is argued that women face a ‘second shift’ (Hochschild & Machung, 1989) because the time they spend in care and housework adds to their paid work hours leaving them with less time for leisure.

Gender differences in time availability and time constraints may similarly affect differences in leisure quality. The ways in which women spend their available free time is seen to be of lower quality given the frequency of multitasking among women. Women’s leisure is more often bound up with family time. Mothers have less child-free leisure and are more likely than fathers to include children in their leisure activities, in essence ‘contaminating’ women’s leisure (Chatzitheochari & Arber, 2012; Craig & Mullan, 2013; Mattingly & Blanchi, 2003). Moreover, the time women spend in care activities ‘often spill[s] over into free time activities in ways that paid work hours do not’ (Nomaguchi & Bianchi, 2004, p. 417). Another reason why women’s leisure quality is argued to be lower is that their family time and leisure time involve invisible efforts, such as the ‘coordination and planning, emotion and kind work, and the production of intimacy and sociability’ (DeVault, 2000, p. 487). As mothers are more prone to take up these responsibilities, their leisure time may be more demanding and less relaxing than fathers’ leisure (Craig & Mullan, 2013; Nomaguchi & Bianchi, 2004).

Gender differences in leisure quality can also be a reflection of differences in individual norms between men and women. While mothers are less likely to have child-free leisure, when they do participate in leisure without children, modern ‘intensive mothering’ norms can generate feelings of guilt (Craig & Mullan, 2013; Mattingly & Blanchi, 2003; Nomaguchi & Bianchi, 2004). Scholars also suggest that women’s sense of responsibility for others prompts them to adjust their leisure to the needs and preferences of their partner and/or children (Miller & Brown, 2005; Shaw, 1994). As a result, their leisure activities are less in line with their own preferences and thus less enjoyable (Miller & Brown, 2005).

While these theoretical arguments provide interesting explanations for why men and women may differ in their leisure quality at an individual level, they fail to account for possible cross-national differences in leisure quality among men and women. We therefore look to the large body of literature on cross-national differences in the gender division of labor and studies of gender in relation to leisure to explore possible macro-level explanations for differences in leisure quality. We combine these literatures to construct cross-level (macro-micro) hypotheses for our multilevel analyses. Together, these literatures suggest that gendered institutions and divisions of care, power relations and gender role expectations are likely to shape men and women’s experiences of leisure quality, leading to possible gender inequalities.

First, gender differences in leisure quality may be smaller in countries with more gender egalitarian norms. Micro-level explanations for gender differences in leisure find the low quality of women’s leisure is driven by women’s specialized responsibility for care (e.g. Henderson et al., 2002). Country-level norms and policies shape men and women’s behavior through socialization and role expectations (‘doing gender’, e.g. Hook, 2006) even passing on socially constructed norms across generations (Holland, 2013). The gender socialization approach contends that role expectations, which determine who takes primary responsibility for family demands and emotion work, are not stable, but vary across time and countries (Hook, 2006; Miller & Brown, 2005). These role expectations inherent in gender norms shape men and women’s leisure as well, serving to ‘perpetuate gender stereotypes and gender-based inequities, and thus to reinforce structured power relations within society’ (Shaw, 1994). In gender egalitarian countries, such as Sweden and Denmark, couples are more likely to feel jointly responsible for care work, reducing women’s perceived time pressure (Craig & Mullan, 2013). As a result, we expect the gender gap in the quality of leisure to be smaller in countries with more egalitarian gender ideologies (Hypothesis 1).

Second, women’s leisure quality may be higher in ‘de-familialised’ welfare states, including the Scandinavian countries, Slovenia, and the UK. Higher leisure quality for women could result from the fact that in de-familialised welfare states, individuals are less dependent on family relationships for maintaining a socially acceptable standard of living (Hook, 2006; Saraceno & Keck, 2011). Assuming that the decrease in family dependencies is not completely offset by an increase in paid working hours, de-familialised welfare states may provide men and women with more flexibility in their leisure time allocation. Higher leisure quality could also result from the normative function of de-familialised welfare states in shaping men and women’s leisure experiences. In countries where the state shares the responsibility of care, women may feel less responsible for the care of others and this may benefit their leisure quality. Given that women spend more time in the family domain than men, we expect women to benefit more than men in countries that are more de-familialised and where female labor force participation is higher. Namely, if women can share care tasks with the state and/or market, this is likely to decrease women’s time constraints and signal that women are entitled to prioritize leisure time, providing women with more opportunities for high quality leisure. As a result, regardless of women’s own working hours, gender differences in leisure quality will be smaller (Hypothesis 2).

A third theoretical argument prevalent in gendered division of labor studies focuses on women’s relative resources, or bargaining power (Bianchi & Milkie, 2010; Hook, 2006). This approach stresses that women in more egalitarian countries have more bargaining power because they have more resources such as education and income. Similarly, gendered leisure research suggests gendered power relations at the societal level can constrain women’s leisure experiences (Henderson & Gibson, 2013; Henderson & Hickerson, 2007; Henderson et al., 2002; Shaw, 1994). In countries where women’s socio-economic and political power is higher women may be more successful in negotiating leisure time (Fuwa, 2004; Hook, 2006). Consequently, in countries where women’s relative bargaining power is higher (Fuwa, 2004; Ruppanner, 2008), gender differences in leisure quality are likely to be smaller (Hypothesis 3).

Finally, leisure quality may differ across countries depending upon the availability of paternity leave. In countries with generous paternity leave, such as Sweden, Finland or Denmark, men may feel more responsible for care. According to Hook (2010), paternity leave facilitates men’s involvement by providing fathers with solo time to care for their children and enhances care in the long term because fathers develop better care-giving skills and are able to strengthen the father-child attachment. Although this association is not uncontested (Hosking, Whitehouse, & Baxter, 2010) we expect that the provision of paternity leave increases men’s sense of responsibility because it enhances the normative acceptance of men’s involvement in care and signals the expectation that men and women have a joint responsibility for care (Rehel, 2014; Rose, Brady, Yerkes, & Coles, 2015). As a result, care tasks and societal norms about care may contaminate men’s and women’s leisure to a similar extent. Thus, we expect that gender differences in leisure quality will be lower in countries with better availability of paternity leave (Hypothesis 4).

Methods

Data and sample

The 2007 module of the International Social Survey Program (ISSP), ‘Leisure Time and Sports’ provides a unique opportunity to answer our research questions because it was specifically designed to measure leisure quality. The 36 countries covered by the ISSP enable us to employ multi-level analyses and test which country characteristics account for cross-national differences. More recent data on leisure quality for a sufficiently large number of countries that would enable multilevel analysis is, to our knowledge, not available. The 2007 survey was a random sample of just under 52,000 individuals. Response rates, complicated by cross-country differences, vary between 20% and 90%. Key variables were missing for eleven countries; these countries have been excluded from the analysis. Respondents with missing variables were also excluded. Our final sample thus comprised 23 countries¹ and 28,980 respondents. Unfortunately, it is difficult to say how representative the sample is, as the sampling strategies varied across the countries (Scholz & Heller, 2009).

Measurement

We consider two indicators of leisure quality: the level of time pressure during leisure and the extent to which men and women use leisure to relax and recover. It has often been argued that when individuals experience time pressure during leisure, it may not deliver the same benefits compared to individuals who are less time pressured during leisure (Bittman et al., 2003; Craig & Mullan, 2013; Henderson & Hickerson, 2007; Mattingly & Blanchi, 2003). Moreover, when individuals use leisure to relax this suggests agency and indicates individuals are able to disconnect from work and family responsibilities (Haller et al., 2013).²

Dependent variable. Subjective time pressure is our first indicator of leisure quality. Respondents were asked the following question ‘In your free time, how often do you feel rushed’ with answer categories ranging from 1 (very often) to 5 (never).³ Our second indicator captures a positive dimension of leisure. Respondents were asked how often they use their free time to relax and recover. Answer categories ranged from 1 (very often) to 5 (never). Both scores have been reversed for the purposes of our analyses.

Individual-level control variables. We control for several individual characteristics. First and foremost, we consider the role of parenthood. As suggested by some of the literature, different mechanisms may apply to gender differences in leisure quality among parents and non-parents. Parents, and particularly mothers, may have more fragmented leisure and participate in activities that are adapted to the demands of children, lowering their leisure quality (Bittman et al., 2003; Gimenez-Nadal & Sevilla-Sanz, 2011). We control for parenthood using the presence of a child in the household (1 = yes, 0 = no). While we recognize that the age of children has important implications for parents’ leisure time (e.g. Mattingly & Blanchi, 2003), these data are not available in the ISSP. This is an important consideration for future research. The other individual-level variables we control for are marital status (1 = married/cohabiting, 0 = not married or cohabiting), employment status (1 = employed, 0 = unemployed), age (in years), age-squared, educational level (1 = no formal education, 2 = above lowest qualification, 3 = higher secondary completed, 4 = above higher secondary level, 5 = university degree completed), and a continuous employment hours variable (a value of 0 is assigned to the non-employed respondents). These factors have been found to influence the subjective experience of time (Gimenez-Nadal & Sevilla-Sanz, 2011; Haller et al., 2013; Mattingly & Sayer, 2006; Miller & Brown, 2005). Gender is measured by a dummy variable (1 = female, 0 = male). ⁴

Country-level independent variables. The measure of aggregate gender ideologies was based on the 2002 ISSP data. The gender ideology measure was only included in specific ISSP waves, and is not available in 2007. We therefore use the ISSP data from 2002 and reconstruct Fuwa’s (2004) measure of gender ideology. This scale is based on five items ‘A job is all right, but what most women really want is a home and children,’ ‘Being a housewife is just as fulfilling as working for pay,’ ‘A man’s job is to earn money; a woman’s job is to look after the home and family,’ ‘All in all, family life suffers when the woman has a full-time job,’ and ‘A pre- school child is likely to suffer if his or her mother works’. Answer categories ranged from 0 (strongly agree) to 4 = (strongly disagree). We took the mean of the complete responses. Higher values on the final measure reflect a more egalitarian gender ideology.

The percentage of 0 to 2-year-olds in formal childcare (in 2007) was used as an indicator of a country’s level of de-familialisation. We used data from the OECD. We use the ratio of female representatives in parliament as an indicator of women’s status in the public domain (Cooke & Baxter, 2010; Ruppanner, 2010). This measure is based on 2008 UNDP data (see Table 2). We measure availability of paternity leave by distinguishing between countries with no paid ‘use it or leave it’ job-protected leave for fathers (value 0) and countries that do have such leave policies (value 1). The necessary information is derived from Ray, Gornick, and Schmitt (2009), complemented with data from the Multilinks database (2016), and Perez’s (2011) report on Israel. Lastly, we note that whereas the individual-level variables only relate to the region of Flanders, the other country characteristics pertain to the whole of Belgium.

Analytical strategy

We use multi-level analysis in Stata and estimate our analytical models in five steps, first estimating the quality of leisure, followed – in the second step – by the inclusion of individual-level variables. In the third step, we test the assumption that the effect of gender varies across countries by changing the coefficient of gender from fixed to random. We then test the direct effects of country-level characteristics before finally adding cross-level interactions between the four country characteristics outlined above and gender. We repeat these steps for both outcome variables: time pressure during leisure and relaxation and recovery during free time.⁵

Results

Men and women’s leisure quality

Tables 1 and 2 present the descriptive data for the individual and country-level variables. Looking at key variables on leisure quality and country-level measures, women reported significantly higher levels of time pressure during leisure (2.93 vs. 2.70 on a scale of 1 = never to 5 = always). Moreover, they were less likely to indicate that they use free time to relax and recover (3.59 vs. 3.62 on a scale of 1 = never to 5 = always). Our key country-level variables indicate that norms are most egalitarian in Scandinavian countries and that these countries rank highest when it comes to childcare coverage and the proportion of women in parliament.

Table 1. Descriptive statistics of individual-level variables.

	Men (N = 13,267)		Women (N = 16,433)		Men and women (N = 29,700)
	Mean/ Proportion	SD	Mean/ Proportion	SD	Range	p-value t-test/ chi^2-test gender difference
Time pressure during leisure	2.70	1.13	2.93	1.16	1–5	.000
Relax and recover	3.62	0.94	3.59	0.96	1–5	.030
Partner status (1 = partner)	0.70	0.46	0.64	0.48	0–1	.000
Parental status (1 = parent)	0.39	0.49	0.57	0.49	0–1	.000
Age	48.12	17.17	47.50	17.11	15–97	.002
Educational level	2.95	1.39	3.00	1.39	1–5	.001
Working hours	27.72	23.02	19.28	20.20	0–80	.000
Employment status (1 = employed)	0.63	0.48	0.53	0.50	0–1	.000

Source: ISSP 2007.

Table 2. Descriptive statistics of country-level variables.

	N	M Relaxed	M Rushed	Gender ideology^a	Childcare enrolment^b	Ratio female seats in parliament^c	Paternity leave^d
Australia	2239	3.00	3.75	3.07	29.00	0.42	0
Austria	985	2.00	3.72	2.88	12.10	0.36	0
Belgium^e	1078	3.00	3.45	2.98	48.40	0.57	1
Bulgaria	827	4.00	3.38	2.55	14.40	0.28	0
Czech Rep.	1135	2.90	3.60	2.76	2.20	0.24	0
Denmark	1157	2.92	3.62	3.60	66.00	0.61	1
Finland	1151	3.01	3.94	3.20	29.00	0.71	1
France	1719	2.14	3.55	3.22	42.00	0.24	1
Germany	1546	2.45	3.74	3.16	17.80	0.45	1
Ireland	1862	2.77	3.62	3.22	30.80	0.18	0
Israel	1121	3.10	3.19	3.10	23.00	0.17	0
Japan	996	2.55	3.67	2.97	28.30	0.14	0
Latvia	933	3.21	3.63	2.71	16.00	0.25	0
New Zealand	771	2.39	3.58	3.29	55.90	0.64	0
Norway	996	2.78	3.74	3.11	37.90	0.51	1
Poland	1201	3.34	3.72	3.54	51.30	0.57	0
Russia	1854	2.96	3.76	2.75	7.90	0.22	0
Slovakia	1051	2.83	3.34	2.53	20.00	0.13	0
Slovenia	933	3.05	3.52	2.57	3.00	0.24	0
Sweden	1131	3.19	3.15	2.87	33.80	0.11	1
Switzerland	964	2.79	3.64	3.52	46.70	0.89	0
UK	566	1.92	3.81	2.92	12.10	0.37	1
US	1484	2.93	3.65	3.18	40.80	0.19	0

^aAggregate measure based on the 2002 ISSP data (see Fuwa, 2004).

^bPercentage of children aged 0–2 in formal childcare in 2008.

^cRatio of seats held by a respective gender in a lower or single house or an upper house or senate, 2008.

^d0 = no paid ‘use it or leave it’ job protected leave for fathers; 1 = less than four weeks of paid leave; 2 = four weeks or more of paid leave.

^eFlanders.

Sources: International Human Development Indicators, ISSP 2002; OECD.stat; Multilinks, 2016; Perez, 2011; Ray et al., 2009; United Nations Development Programme, Worldbank.

Men and women’s leisure quality across countries

To test the assumption that gender differences in leisure quality are contingent upon the country context, we first estimated whether empty models with random intercepts improve the model significantly (in comparison to no hierarchical structure). In a model with random intercepts the intercepts are allowed to vary across countries. These analyses (not reported) suggest that 10% of the variance in time pressure and 4% of the variance in the rest and recover measure are attributable to country differences. We therefore proceeded to estimate cross-country gender differences in leisure quality by allowing the slope of gender to vary across countries. As the fit improved significantly when allowing for a random slope for gender in these individual-level models, gender differences in leisure quality are dependent upon the country context.

Associations with the individual-level variables. The first column of Tables 3 and 4 present the results from the models with the random intercept, the random slope and the individual-level variables. Estimates of the individual-level variables (results not reported) confirm the descriptive results, with women reporting higher levels of time pressure during leisure. All control variables, with the exception of employment status, were associated with time pressure (Model 1a). Older and higher educated respondents, respondents with children, and those who work longer hours report higher levels of time pressure during leisure. We also see that those with a partner experience more time pressure during leisure, which may suggest that it is difficult to coordinate leisure with a partner.

Table 3. Direct and cross-level interaction effects on leisure quality; how often does the respondent feel rushed during leisure. Coefficients (standard errors between parentheses) (N_countries = 23; N_individuals= 29,700).

	Model 1a	Model 2a	Model 3a	Model 4a	Model 5a	Model 6a
Direct effects
Female (1 = yes)	0.258 (0.02)***	0.258 (0.02)***	0.125 (0.19)	0.221 (0.04)***	0.268 (0.04)***	0.277 (0.02)***
Partner (1 = yes)	0.094 (0.01)***	0.094 (0.01)***	0.094 (0.01)***	0.094 (0.01)***	0.094 (0.01)***	0.095 (0.01)***
Parent (1 = yes)	0.144 (0.01)***	0.145 (0.01)***	0.145 (0.01)***	0.145 (0.01)***	0.144 (0.01)***	0.144 (0.01)***
Age	0.011 (0.00)***	0.011 (0.00)***	0.011 (0.00)***	0.011 (0.00)***	0.011 (0.00)***	0.011 (0.00)***
Age squared	−0.000 (0.00)***	−0.000 (0.00)***	−0.000 (0.00)***	−0.000 (0.00)***	−0.000 (0.00)***	−0.000 (0.00)***
Education	0.033 (0.00)***	0.033 (0.00)***	0.033 (0.00)***	0.033 (0.00)***	0.033 (0.00)***	0.033 (0.00)***
Working hours	0.006 (0.00)***	0.006 (0.00)***	0.006 (0.00)***	0.006 (0.00)***	0.006 (0.00)***	0.006 (0.00)***
Employed (1 = Y)	0.001 (0.03)	0.000 (0.03)	−0.000 (0.03)	−0.000 (0.03)	0.000 (0.03)	0.001 (0.03)
Egalitarian norms		−0.661 (0.41)	−0.752 (0.43)	−0.665 (0.41)	−0.662 (0.41)	−0.653 (0.41)
Childcare		0.017 (0.01)*	0.017 (0.01)*	0.014 (0.01)	0.017 (0.01)*	0.017 (0.01)*
Ratio seats		0.303 (0.44)	0.303 (0.44)	0.298 (0.44)	0.356 (0.48)	0.312 (0.44)
Paternity leave		−0.319 (0.21)	−0.319 (0.21)	−0.307 (0.21)	−0.317 (0.21)	−0.226 (0.23)
Cross-level interactions
Female*
norms			0.044 (0.06)
childcare				0.001 (0.00)
seats					−0.027 (0.09)
paternity leave						−0.052 (0.04)
Intercept	2.532 (0.09)***	4.055 (1.07)***	4.329 (1.14)***	4.144 (1.08)***	4.040 (1.08)***	3.992 (1.07)***
Variance of the intercept	−2.742 (0.25)***	−2.712 (0.25)***	−2.729 (0.25)***	−2.759 (0.26)***	−2.717 (0.25)***	−2.794 (0.27)***
Variance of the gender effect	−1.022 (0.15)***	−1.029 (0.17)***	−1.032 (0.17)***	−1.041 (0.17)***	−1.031 (0.17)***	−1.034 (0.17)***
Covariance intercept-gender	−0.325 (0.29)	−0.699 (0.40)	−0.696 (0.40)	−0.664 (0.40)	−0.693 (0.40)	−0.731 (0.42)
Individual-level variance	0.047 (0.00)***	0.047 (0.00)***	0.047 (0.00)***	0.047 (0.00)***	0.047 (0.00)***	0.047 (0.00)***
ll	−43135.771	−43133.260	−43133.022	−43132.600	−43133.218	−43132.375

Note: The reported models have a random intercept and a random slope for gender.

*p < .05, **p < .01, ***p < .001.

Table 4. Direct and cross-level interaction effects on leisure quality; how often does the respondent use leisure to relax and recover. Coefficients (standard errors between parentheses) (N_countries = 23; N_individuals= 29,700).

	Model 1b	Model 2b	Model 3b	Model 4b	Model 5b	Model 6b
Direct effects
Female (1 = yes)	−0.028 (0.02)	−0.027 (0.02)	−0.438 (0.19)*	−0.105 (0.04)**	−0.084 (0.04)*	−0.059 (0.02)**
Partner (1 = yes)	0.068 (0.01)***	0.068 (0.01)***	0.068 (0.01)***	0.068 (0.01)***	0.068 (0.01)***	0.068 (0.01)***
Parent (1 = yes)	0.106 (0.01)***	0.106 (0.01)***	0.106 (0.01)***	0.106 (0.01)***	0.106 (0.01)***	0.106 (0.01)***
Age	−0.006 (0.00)**	−0.006 (0.00)**	−0.006 (0.00)**	−0.006 (0.00)**	−0.006 (0.00)**	−0.006 (0.00)**
Age squared	0.000 (0.00)	0.000 (0.00)	0.000 (0.00)	0.000 (0.00)	0.000 (0.00)	0.000 (0.00)
Education	0.061 (0.00)***	0.061 (0.00)***	0.061 (0.00)***	0.061 (0.00)***	0.061 (0.00)***	0.061 (0.00)***
Working hours	−0.001 (0.00)*	−0.001 (0.00)*	−0.001 (0.00)*	−0.001 (0.00)*	−0.001 (0.00)*	−0.001 (0.00)*
Employed (1 = yes)	0.123 (0.03)***	0.122 (0.03)***	0.121 (0.03)***	0.121 (0.03)***	0.122 (0.03)***	0.121 (0.03)***
Egalitarian norms		0.441 (0.17)**	0.457 (0.17)**	0.441 (0.17)**	0.440 (0.17)**	0.440v (0.17)**
Childcare		−0.011 (0.00)***	−0.011 (0.00)***	−0.011 (0.00)***	−0.011 (0.00)***	−0.011 (0.00)***
Ratio seats		0.311 (0.18)	0.312 (0.18)	0.315 (0.18)	0.325 (0.18)	0.316 (0.18)
Paternity leave		0.029 (0.09)	0.026 (0.09)	0.024 (0.09)	0.030 (0.09)	0.039 (0.09)
Cross-level interactions
Female *
norms			0.136 (0.06)*
childcare				0.003 (0.00)*
seats					0.162 (0.09)
paternity leave						0.093 (0.04)**
Intercept	3.580 (0.06)***	2.443 (0.44)***	2.395 (0.44)***	2.434 (0.44)***	2.440 (0.44)***	2.442 (0.44)***
Variance of the intercept	−2.543 (0.22)***	−2.558 (0.22)***	−2.708 (0.25)***	−2.766 (0.26)***	−2.661 (0.24)***	−2.780 (0.27)***
Variance of the gender effect	−1.749 (0.16)***	−2.103 (0.17)***	−2.109 (0.17)***	−2.112 (0.17)***	−2.104 (0.17)***	−2.118 (0.17)***
Covariance intercept-gender	0.143 (0.27)	0.261 (0.37)	0.299 (0.36)	0.328 (0.37)	0.278 (0.36)	0.371 (0.38)
Individual-level variance	−0.073 (0.00)***	−0.073 (0.00)***	−0.073 (0.00)***	−0.073 (0.00)***	−0.073 (0.00)***	−0.073 (0.00)***
ll	−39609.938	−39602.551	−39600.273	−39599.642	−39601.068	−39599.680

Note: The reported models have a random intercept and a random slope for gender.

*p < .05, **p < .01, ***p < .001.

The pattern for our second indicator is somewhat similar (Model 1b). Leisure quality is higher for those without a partner and children, whereas the association with age and working hours is negative. The educational gradient is positive, suggesting that even though higher educated respondents experience more time pressure during leisure, they more often use leisure to relax and recover. The results indicate there are no gender differences in leisure quality in terms of using free time to relax and recover.

The associations between the individual level control variables and the outcome variables are generally in line with prior research (e.g. Haller et al., 2013; Sullivan, 2007). Nevertheless, the finding that partnered respondents experience more time pressure during leisure seems to be in contrast with Gimenez-Nadal and Sevilla-Sanz’s (2011) finding that having a partner is positively associated with leisure satisfaction.

Associations with the country-level variables. Models 2a and 2b build on the models with the individual-level variables by including the direct effects of the country characteristics (see Tables 3 and 4). We are primarily interested in cross-level interactions and whether and how country characteristics moderate the effect of gender differences in leisure, so we did not formulate any expectations about direct country effects. We discuss these findings very briefly (see Haller et al., 2013 for an elaborate analysis and discussion of cross-national variations in time pressure during leisure). In countries where childcare coverage is higher, perceived time pressure in leisure is higher. There is also a positive association between childcare coverage and the extent to which respondents use leisure to recover and relax. Respondents in countries with more egalitarian norms more often report they use leisure to relax and recover. This finding could be driven by the Scandinavian countries where childcare coverage is high, but people also experience high levels of time pressure (Drobnič, Beham, & Präg, 2010). This relationship could also reflect the greater time constraints caused by using childcare services, and the stress associated with drop-off and pick-up times.

Cross-level interactions. Turning to our cross-level interactions, we see that our hypotheses are refuted for the first outcome variable: Country characteristics do not moderate the impact of gender on perceived time pressure (Models 3a and 3b). In other words, there is no significant relationship between gender attitudes, childcare coverage, women’s bargaining power, or paternity leave and the impact of gender on perceived time pressure. The cross-level interactions do yield a significant effect in the models predicting the extent to which the respondent uses free time to relax and recover. As predicted, the gender gap is smaller in countries with more egalitarian norms, greater childcare coverage, higher political power for women, and better paternity leave. Nevertheless, if we calculate the differences between the countries who score highest and lowest on these country characteristics, the results show the effects are small. Figure 1(a–c) illustrate the effects by presenting the marginal effects.

Figure 1. Predicted leisure quality (how often does the respondent use leisure to relax and recover) by gender (models 4b–6b).

Figure 1(a) (a graphical representation of Model 4b) shows the gender gap in leisure quality is smaller when norms are more egalitarian. The difference between the least and most egalitarian country (Russia and Norway) is small however. In Russia, the predicted means for men and women are 3.40 and 3.30 respectively; while in Norway the predicted mean is 3.84 for men and 3.88 for women. Thus, whereas men have a small benefit in Russia, women have a small benefit in Norway. This reversal in the effect of gender across countries explains why we did not find a significant gender effect in Models 1b and 2b. For childcare (Figure 1(b), a graphical representation of Model 5b), we see that as the level of childcare coverage increases, leisure quality decreases and the gender gap becomes smaller (and reverses). Women’s bargaining power does not appear to mediate the relationship between gender and leisure quality: There is no direct effect of the percentage of female seats in the parliament. The gender gap in leisure quality is smaller when this percentage is higher (Figure 1(c)). Finally, paternity leave increases women’s leisure quality more than men’s and thereby narrows the gender gap.

We ran additional analyses in which the sample was restricted to certain sub-samples: partnered respondents, parents, non-parents and employed respondents. These models showed similar findings to those reported above (results not shown: available upon request). Moreover, since Gimenez-Nadal and Sevilla-Sanz (2011) found that marriage negatively impacts men’s leisure satisfaction but does not yield an effect for women, we checked if there is a similar effect in our models. Results (available upon request) show that the association between having a partner and time pressure during leisure is slightly stronger for women. In the model explaining the extent to which respondents use leisure to relax, the positive effect of having a partner is stronger for men than for women. Both findings are in line with Gimenez-Nadal and Sevilla-Sanz’s (2011) study. Nevertheless, because including this interaction did not affect the results and we wanted to keep the models parsimonious, we did not include this interaction term in the models shown here.

Discussion

The current study compares and explains gender differences in the quality of leisure across 25 countries, using the 2007 ISSP Leisure Module. This large cross-national comparison enables examination of variations in social expectations and structural inequities relating to gender across countries (Baxter, 1997; Cooke & Baxter, 2010; Haller et al., 2013; Shaw, 1994). We considered two indicators for leisure quality: time pressure during leisure and the extent to which individuals use leisure time to relax and recover. To our knowledge, this study is the first to analyze how and why gender differences in leisure quality vary across countries. We do so using multi-level analysis that estimates whether the gender gap in leisure quality is affected by societal role expectations (measured by gender egalitarian norms), gendered and institutionalized divisions of care (measured by childcare coverage as a measure of de-familialisation; and the availability and conditions of paternity leave), and women’s bargaining power (measured by looking at female political power).

The analyses show cross-national variations in leisure quality exist and that gender differences in leisure quality vary across countries. Our findings offer important contributions to the small but growing work-family and time use literature on gender differences in leisure quality. First and foremost, our results suggest the relationship between country-level factors and gender differences in leisure quality is complex. Women’s ability to use free time to relax and recover is moderated by certain country characteristics. While the effects are small, we find that in countries with conservative gender norms, low levels of childcare coverage, limited paternity leave and lower political power for women, women’s leisure quality is lower than men’s. As countries become more egalitarian, the gender gap decreases and eventually reverses. These findings, despite being small, confirm prior studies, suggesting that societal level norms and institutions can play an important role in shaping men and women’s experiences of leisure. We do note, however, that despite the fact that women clearly experience more time pressure during leisure, a finding in line with prior research (Bittman et al., 2003; Chatzitheochari & Arber, 2012), the country characteristics we studied do not moderate the association between gender and time pressure.

Second, we find that all four country-level factors that we considered (egalitarian norms, childcare coverage, women’s political power, and paternity leave) affect the extent to which men and women use free time to relax and recover across countries. This supports Craig and Mullan’s (2013) suggestion that gender differences in leisure quality are smaller in countries where governments provide more family support. Crucially, while Craig and Mullan’s findings are based on a five-country study, we have been able to test and confirm this idea across 23 countries, using multilevel analysis to focus on these country-level effects. This finding also provides empirical support for the dominant idea in gendered leisure research that societal gender relations shape the extent to which women feel entitled to use free time to relax and recover (e.g. Henderson & Hickerson, 2007; Shaw, 1994). Countries with more egalitarian role expectations, institutionalized egalitarian norms around divisions of care, and higher bargaining power for women in society create a space in which men and women are more equal in their entitlement to use leisure time to relax and recover.

The persistent gender effect on subjective time pressure and strong effects from the individual-level variables suggest, however, that the sources of time pressure during leisure may lie closer to the individual. The small cross-level effects found for free time as a form of relaxation and recovery, in combination with the reversal of the effect across countries, suggest it could be fruitful in future research to look more closely at this relationship within clusters of countries that share similar characteristics. For example, it could be interesting to study whether the impact of individual characteristics other than gender differs across countries. Haller et al. (2013) differentiate between ‘time regimes’ and argue that the social, cultural and political characteristics of specific country clusters shape leisure quality patterns. Such time regimes may offer more robust ways of investigating cross-country differences in leisure than de-familialisation, allowing for a focus on further socio-economic differences, such as those between parents and non-parents and higher and lower educated individuals.

Additional avenues for future research are dependent upon improvements in data. For example, once the sample sizes of cross-national time use datasets are larger, our study can be replicated with time diary data. Time diary data provide the opportunity to analyze more advanced indicators of leisure quality such as leisure voraciousness (Sullivan, 2007), and contamination and fragmentation (Bittman et al., 2003; Chatzitheochari & Arber, 2012). With such data, it would be possible to unpack the complex relationship between the way in which women and men ‘do gender’ (e.g. Hook, 2006; Shaw, 1994) and the moderating effect of egalitarian norms at the macro level, for example. Second, with a large enough cross-national dataset of time diary data, future research could investigate whether the role of country characteristics is different when specific types of activities are considered (e.g. passive versus active leisure). Country characteristics may affect some activities more than others and, as Brajša-Žganec et al. (2011) have shown some activities (such as visiting friends and relatives and going out) matter more for individuals’ quality of life than others.

We note some limitations to our approach. To focus on the country context, we have largely neglected individual-level differences. Prior research on leisure shows it can be important to consider the combination of gender with other defining socio-demographic characteristics, such as class and ethnicity (Cooke & Baxter, 2010; Gimenez-Nadal & Sevilla-Sanz, 2011; Henderson & Gibson, 2013; Shaw, 1994). Variations across the life course may be relevant as well, as the restrictions and opportunities for (high quality) leisure change across the life course. Unfortunately, the ISSP data do not include such detailed information. Four other data-related limitations are important to mention: the country-level response rates ranged from 20% to 90% so data quality varied considerably across countries. Second, more recent cross-country data on leisure quality would be preferable, particularly data collected after the global financial recession, as the rise in second jobs may have changed men and women’s access to leisure time, and cuts to public services may also have reduced leisure opportunities. Third, as the sampling methods varied across the countries, we cannot be certain that the ISSP samples are representative (Scholz & Heller, 2009). We attempted to limit this problem by controlling for basic background characteristics, yet it is possible that our findings apply to a select group of people. Moreover, we were not able to control for the age of the youngest child. Lastly, the small difference in leisure quality between men and women could be a reflection of our measure of leisure quality. As such, it could be interpreted as men having more leisure time, but with men and women having near equivalent leisure quality. In other words, it could be that in more egalitarian countries, the quality of everyone’s leisure is improved. Further research into leisure quality is needed to test this assumption.

Despite these limitations, our study provides key insights into the relationship between gender and leisure quality across countries. Although country characteristics explain part of the gender differences in the quality of leisure, these gender differences are also persistent across countries. These findings have potential important theoretical implications for our understanding of gender differences in leisure. Our cross-country findings confirm the complexity of understanding gender differences in leisure in relation to socio-cultural differences (Henderson & Gibson, 2013). As noted above, at the country level, differences in gendered role expectations, institutionalized norms around men and women’s care responsibilities, and women’s bargaining power matter for the ways in which women and men engage in leisure activities. Despite developments in some countries towards more egalitarian societies in which men and women have more equality in paid work and care work, and thus potentially more space for equality in leisure, gendered differences in leisure quality remain. On the one hand, these differences could reflect a difference in the value that men and women place on their leisure time and activities. Research on gender inequalities in both paid and unpaid work have demonstrated significant gender differences in what men and women value when assessing gender differences as fair or unfair (Thompson, 1991; Yerkes, Martin, Baxter, & Rose, 2017). In relation to leisure, women may place less value on having equal amounts of leisure time but more value on other outcomes, such as having high quality family leisure. On the other hand, the consistency of gender differences in leisure quality across countries could suggest that men and women have, to a greater or lesser extent, internalized gendered leisure expectations. For example, despite the increase in women’s labor market participation and policies promoting egalitarian divisions of care, women continue to take on greater amounts of care responsibility than men (Craig & Mullan, 2010), reflecting gendered expectations that care tasks are women’s responsibility. Similarly, despite the shift to more egalitarian societies, women may continue to feel less entitled to have unencumbered leisure time. These remain important questions for future research.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Mara A. Yerkes is Associate Professor in Interdisciplinary Social Science at Utrecht University. Her research interests include work, care and family, gender, comparative social policy, industrial relations, social inequality and women’s employment.

Anne Roeters is a senior researcher at the Netherlands Institute for Social Research. Her research interests include family sociology, time use research, and work-life balance.

Janeen Baxter is Professor of Sociology and Director of the ARC Centre of Excellence for Children and Families over the Life Course in the Institute for Social Science Research, University of Queensland. Janeen has research interests in inequality, family, gender and the life course. She is a Fellow of the Academy of Social Sciences in Australia.

Notes

1 Australia, Austria, Belgium (Flanders), Bulgaria, Czech Republic, Denmark, Finland, France, Germany, Ireland, Israel, Japan, Latvia, New Zealand, Norway, Poland, Russia, Slovak Republic, Slovenia, Sweden, Switzerland, UK and US.

2 Ideally we would complement these subjective indicators of leisure quality with more objective time use measures. However, the ISSP does not include time diary measures. Although there is some cross-national time use data from other sources, these data do not include measures of time pressure during leisure.

3 The Danish data are missing the answer category of ‘seldom’ (point 4 on the 5-point scale).

4 Preferably we would also include income, however this is not possible given a high number of respondents who did not provide this information.

5 Because the weights that were included in the ISSP pertain to the full sample, and eleven countries are excluded from our analysis, we do not apply weights in our models.