Introduction The study of the gender division
The study of the gender division of household labor is an area of research social scientists
frequently investigate, but the underlying social mechanisms that continue to maintain this
division are still not fully understood. Contemporary societies have experienced substantial
developments leading to more gender equality in the past several decades, and the equality
between men and women has been considered desirable in the domain of paid labor in most
industrial countries. Yet, unpaid labor in the domestic sphere is one of the most obstinate realms
of gender inequality and is less affected by such a social change. Despite considerable movement
toward greater gender equality in paid work during the last few decades, women still have a
majority of responsibility for domestic labor, and the types of housework are also “gendered”
regardless of individual characteristics such as income, education, and ideology (Bianchi et al.
2000; Coltrane 2000; Geist and Cohen 2011; Hook 2006, 2010; Inglehart and Norris 2003).
Given this, scholars argue that the family, as a socially constructed institution, is one of central
places where gendered practices occur in everyday life (Berk 1985; Ferree 2010), and that
unpacking the mechanisms of the gendered divisions of household labor is essential to
understanding the continued generation and perpetuation of gender inequality in society (Bittman
et al. 2003; Coltrane and Adams 2008).
Multiple theoretical perspectives (e.g., relative resources, time availability, and gender
role ideology) have been proposed to account for why the gendered division of household labor
persists and how individual factors are associated with the unequal distribution of housework
between couples (Baxter 2000, Baxter and Western 1998; Bianchi et al. 2000; Brines 1994;
Coltrane and Ishii-Kuntz 1992; Crosby 1976; Kamo 1988; Sorensen and McLanahan 1987).
While most studies through the late 1990s focused on the influences of individual-level
characteristics on the division of household labor, the effects of cross-national contexts have
received much attention since 2000, with developments of statistical techniques and a growing
availability of cross-national data sets. Comparative scholars have found that unequal divisions
of household labor and housework-related family life are situated in various national contexts,
which can be explored through variables such as the economic development level, female labor
force participation, divorce and cohabitation propensity, and social policy (Batalova and Cohen
2002; Braun et al. 2008; Davis and Greenstein 2004; Fuwa 2004; Fuwa and Cohen 2007; Geist
2005; Greenstein 2009; Hook 2006; Knudsen and Wærness 2008; Ruppanner 2012; Yodanis
2005). In particular, both individual- and national-level gender equality, in most studies, are
consistently observed as significant factors that affect housework distribution between couples as
well as cross-national variations in the division of household labor. This suggests that gender
equality at the micro- and macro-levels plays a key role in explaining the persistence of gender
division of household labor.
Previous comparative research provides valuable insights into the mechanism of
household labor across countries, but there are still some gaps to be addressed. First, the
influences of macro-level variables in terms of gender equality are well documented, but the
effects of micro-level variables, which might depend upon gender equality at the national level,
have been investigated less often. This calls attention to the complexity of cross-level
interactions which situate macro-level gender equality in the context of various individual
characteristics and social status. Second, while many have documented actual and perceived
inequalities in household labor, cross-national research examining the link between the gender
division of household labor and work-family conflict has been scant. Work-family conflict has
become an important issue for both researchers and policy makers as the dominant family model
largely shifted to the dual-earner family in most industrial countries. This suggests the need for
further comparative research examining antecedents and consequences of work-family conflict,
in terms of the gender division of household labor and related-family life. Lastly, many possible
predictors have not yet been examined empirically even though they are likely to link to the
division of household labor theoretically. Specifically, the attitudes toward work-family
arrangements are likely to be associated with the actual division of household labor, and in turn,
national contexts may also affect the construction of individuals’ preferences regarding
workfamily arrangement. However, partly because of the limitations of data sets, almost no
attention has been paid to the role of personal preferences with regard to work-family
arrangement.
In this dissertation, I attempt to address the above three issues by conducting multi-level
analyses using data from the 2012 International Social Survey Programme, a continuing annual
program of cross-national collaboration involving 40 countries (ISSP Research Group 2016).
The 2012 module “Family and Changing Gender Roles IV” provides a variety of measures for
the current study including gender division of time spent on paid and unpaid work, attitudes
toward gender role ideology, preferences on work-family arrangement, work-family conflict, and
family life satisfaction. Building on prior theoretical and empirical literature on social changes in
gender roles and changing gender relations within the family, I aim to examine the relationships
between the division of household labor, work-family conflict, and family life satisfaction across
countries, with a particular emphasis on the national context of gender equality.
As a result, the purpose of this dissertation is to contribute to expanding our knowledge
about how gendered power relations are (re)produced and legitimated in domestic life and how
national contexts influence that relationship. The important connections my work draws between
national contexts and individual interactions support the development of family policies that can
encourage individuals to move toward gender equality within the family. Without considering
national contexts, some family policies aiming to dissipate women’s double burden in paid and
unpaid labor can lead to unintended results. For instance, prior research shows that a policy
allowing a long-term parental leave led to women’s long-term career interruptions, and the
effects of that policy also vary according to the national contexts of gender equality and labor
market conditions in a given country (Fagnani 1999; Fuwa and Cohen 2007). By shedding light
on the cross-national variations in the relationship between housework, family life, and gender
inequality, this study will contribute to a better understanding of the complexities of outcomes of
family policies across countries.
1.2. Structure of the Dissertation
I organize this dissertation as follows. Following this introductory chapter (Chapter 1), I
review prior literature on the division of household labor and its relationships with perceived
fairness of that division, work-family conflict, and family life satisfaction (Chapter 2). By doing
this, I build theoretical frameworks for the current study, and then propose hypotheses that I test
throughout my dissertation. Next, I introduce the data set, measures, and analytical strategies in
the methodology section (Chapter 3).
Chapter 4 is the results section that presents the findings of the analyses. First, I examine
the cross-national variations in the division of household labor, with a particular emphasis on the
role of individuals’ preferences on work-family arrangement (Chapter 4.1). The division of
housework is likely to be associated with an individual’s preference on work-family
arrangement, and such a relationship may also be affected by cultural, structural, and economic
aspects of national context. Yet, the empirical investigation of this relationship across countries
are lacking. I address this void by using the 2012 ISSP data with a new question about
preferences on work-family arrangement. Next, I explore the extent to which and how the
unequal division of household labor and perceived unfairness of that division are associated with
individual-level work-family conflict from a comparative perspective, by focusing on gender
differences (Chapter 4.2). Lastly, I examine the association between the division of housework,
work-family conflict, and family life satisfaction across countries by considering the possible
moderating effect of country-level gender equality (Chapter 4.3).
In the final chapter (Chapter 5), I conclude this dissertation by summarizing the key
findings of the analyses and by discussing implications and contributions of this study.
Chapter 2. Theoretical Frameworks and Literature Review
In this chapter, I begin with an overview of social changes related to changing gender
roles during the past several decades. It is typically expected that such trends in changing gender
roles in paid work lead to changes in gender relations in families. However, women still bear a
disproportionate share of family responsibilities at home, and traditional patterns in the division
of household labor remain prevalent in most industrial societies. Given this, I review theories
that explain the persistence of gender inequalities in household labor at the individual-level.
Then I highlight the role of national contexts by reviewing previous comparative research on
cross-national variations in the division of household labor and its influence on houseworkrelated
family life. In the following three subchapters, I discuss three main topics of this dissertation—
the division of household labor, work-family conflict, and family life satisfaction, respectively.
Finally, I propose hypotheses for the current study.
2.1. Social Changes, Changing Gender Roles, and National Context
2.1.1. Social Changes and Changing Gender Roles
The past several decades have witnessed considerable social changes in most industrial
societies which are closely related to changes in gender roles. First, the increasing participation
of women in paid work is one of the major driving forces of changes in gender relations in
families. The concept of the “modern nuclear family” consisting of a breadwinner-husband, a
homemaker-wife, and their children has been dominant in most advanced countries during the
20th century (Cornell 1990). The nuclear family model assigns responsibilities of paid work to
men while responsibilities of domestic labor to women based on the belief in gendered separate
spheres (South and Spitze 1994; West and Zimmerman 1987). Given this family model, men
should work in the paid labor market to support their family, while women should stay at home
to take responsibility for childrearing and housework, and this form of division of labor has been
considered most desirable. Meanwhile, economic systems that allow men to earn enough money
to support their families have legitimated the ideal of a modern nuclear family since the postwar
era (Fraser 1994; Glickman 1999). Yet, economic changes and growing during the past few
decades no longer allow men to earn a “family wage.” For example, the inflation-adjusted
median hourly wage for American men was peaked in 1974 and has not increased since then
(Johnson 1997). Earnings inequality has also rapidly spread throughout the United States around
the same time. For the first three decades after World War II, pretax incomes grew at roughly the
same rate for households at all rungs of the income ladder. Since the mid-1970s, however, the
lion’s share of real income growth has been concentrated among top earners in the group, while
those at the bottom of the top quartile have seen little real income growth (Frank 2018; Saez
2018). In the context of the sharp rise in income inequality and the virtual stagnation of the
average real wage, the dual-earner family is more likely to be preferred than the
malebreadwinner model for the greater financial contribution to the household. Indeed, real
median household incomes in the United States have grown by roughly 15% since the mid-
1970s, and this is primarily because of large increases in female labor force participation (Frank
2018). With this background, the share of dual-income households and the labor force
participation rates of women, especially mothers, had both increased substantially in many
industrialized societies over the last few decades (Gornick and Meyers 2003), even though the
labor force participation rates for both women and men have somewhat decreased over the last
two decades at global level (52.4% to 49.6% for women, 79.9% to 76.1% for men, 1995-2015)
(International Labour Organization 2016). The shift to the service economy, in turn, has played a
role in widening the opportunities of women’s participation in the labor market with the
expansion of jobs in the service sector along with increasing numbers of part-time jobs (Lewis
2009). As a result, dualearner households have become commonplace today, and women are
more likely to be expected to work outside the home.
Second, changes in gender ideology toward greater gender equality also made a
significant contribution to changes in gender roles in both public and domestic dimensions.
Above all, the women’s liberation movement that emerged in the 1960s and led to the second
wave of the feminist movement until the early 1980s (Tong 2013) has played a significant role in
constructing a contemporary society favorable to egalitarianism. Not only did the feminist
movement expand understandings of the concept of gender equity, but it also had a major
influence on the development of state policies aimed to enhance gender equality in the public
sphere (McAdams et al. 1996). In addition to the role of the feminist movement, other
mechanisms and factors have also influenced the general trend toward a more egalitarian beliefs
about gendered behaviors at the both micro- and macro-levels. For instance, by focusing on the
micro-level dynamics underlying the trend toward a more egalitarian gender ideology,
Bolzendahl and Myers (2004) argue mechanisms producing feminist opinion using two sets of
approaches based on interests-based explanations and exposure-based explanations. The
interestbased explanations posit that individuals are likely to favor egalitarianism when their
interests benefit from gender equality, while the exposure-based explanations emphasize the
effects of exposure to gender egalitarian situations and contexts (Bolzendahl and Myers 2004).
These approaches provide valuable insights into understanding variations in changes in
egalitarian attitudes at the individual level, while showing how changing gender ideology
interplays with other social, cultural, and political contexts at the macro-level. Given those two
clusters of approaches, Davis and Greenstein (2009) categorize social and demographic
predictors of gender ideology, including individuals’ social location, parents’ gender ideologies,
racial differences intersecting with social stratification, religion, educational attainments,
employment status, occupation, and entrance into gendered relationships of marriage and
parenthood. In doing so, they point out the complex nature of the construction of gender
ideology (Davis and Greenstein 2009). On the other hand, recent research has found that the
changes in population structure on the macro-level, which result from births and deaths, play an
important role in explaining the general trend toward a more egalitarian gender ideology
(Bolzendahl and Myers 2004; Brewster and Padavic 2000; Ciabattari 2001; Davis and
Greenstein 2009; Mason and Lu 1988).
Conducting a regression decomposition analysis using data from the General Social Surveys
(1977 to 1996), Brewster and Padavic (2000) illuminate that demographic processes of
population turnover and micro-level attitude change caused by cohort replacement are both
important in understanding changes in attitude trends toward more egalitarian gender ideology.
In the research on the period and cohort effects on changes in men’s attitudes toward women’s
roles, Ciabattari (2001) argues that the overall increase in men’s egalitarian attitudes results from
two main processes: individual-level changes (e.g., the improvements of women’s education and
labor force participation, and the expansion of the feminist movement) and the cohort
replacement effect. In other words, she finds that the oldest cohort is more traditional than the
more recent cohort, suggesting that men’s conservative attitudes have overall decreased as the
size and influence of this oldest and most conservative cohort has reduced over time (Ciabattiari
2001).
Finally, a third major social change related to changing gender roles is recent
transformations in family and living arrangements that occurred in most industrialized countries
during the latter half of the 20th century. Despite considerable variations in the degrees and
patterns of changes in family life and behaviors within and between countries, today people are
more likely to live longer, have fewer children, and get married later or not at all. Divorce rates
are high (although they have recently leveled off in many countries), and various forms of
alternative family, such as cohabitation, nonmarital childbearing, gay marriage, childlessness, or
single-parent families, are more socially acceptable in many industrialized societies (Amato et al.
2007; Casper and Bianchi 2002; Cherlin 2010; Coontz 2004, 2015; Hiekel et al. 2014; Kasearu
and Kutsar 2011; Kiernan 2001; Raymo et al. 2015; Reher 1998). As family life and behaviors
have dramatically changed, the function and meaning of family have also changed, although the
direction of the causality is not always clear. The traditional “modern nuclear family” has lost its
normative place, while many alternatives to marriage have emerged (Cherlin 2010; Cooke and
Baxter 2010; Coontz 2015; Raymo et al. 2015; Reher 1998; Surkyn and Lesthaeghe 2004). In
turn, all these changes have led to increasing numbers of various family structures and
relationships. The traditional gender roles based on the modern nuclear family consisting of the
male-breadwinner and female-homemaker are no longer adequate to cope with recent changes in
families. Accordingly, families need to reorganize their roles to be flexible in response to recent
changes in family structures and living arrangements.
In sum, the literature provides evidence that there have been considerable social changes
that are related to changes in gender relations within families in most industrial countries over
the last decades. In this regard, changing gender roles and their impacts on family life have
received much attention from scholars and policymakers, and examining the ramifications of
such changes in gender roles has become a more important research agenda for family scholars.
Given the considerable improvement of women’s participation to paid labor, the spread of
egalitarianism, and transformations of family structures, family scholars have sought to examine
how such changes influence gender dynamics in the domestic sphere (Sullivan 2004; Davis and
Greenstein 2009). In particular, a growing body of literature has focused on explaining how the
traditional division of household labor is redistributed between couples (e.g., Hochschild 1989),
how dual-earner couples reorganize work and family life in the boundaries of roles between
providers and caregivers (e.g., Gornick and Meyers 2003; Padavic and Reskin 2002), and how
such experiences of changes in family life are associated with individuals’ well-being and
subjective perceptions (e.g., Greenstein 1996; Minnotte et al. 2010; Sanchez and Kane 1996).
2.1.2. Theories Explaining Unequal Gender Relations in Families at the Individual-Level
Up until the late 1990s, scholarship explaining domestic gender relations focused
primarily on the influences of individual-level characteristics, mostly because of the lack of
available cross-national data sets. The various theoretical frameworks are proposed to account
for why unequal gender relations in families persist despite the substantial improvements in
women’s roles in the public sphere. As a result, both economic approaches and feminist
perspectives have provided important conceptual and theoretical frameworks for family
scholarship.
The time availability approach from Becker’s (1965) theory of the allocation of time is
one of the most influential theories explaining the division of household labor from an economic
point of view. This theory focuses on rational time allocation between paid labor market and
housework by postulating that the time spent in household labor is negatively associated with the
time spent in the paid labor market. In light of this framework, the time availability perspective
assumes that if one spouse works longer outside of the home, he or she does less housework.
This model was particularly well supported in the context of the modern nuclear family model in
which the male breadwinner specializes his role in the labor market while the female homemaker
specializes her role in terms of domestic labor. On the other hand, the relative resources approach
based on exchange theory from Blood and Wolfe’s (1960) bargaining model posits that
individuals’ resources (e.g., income and education) can be exchanged for household labor, and
they use their resources to negotiate the allocation of housework with their partners. According
to this perspective, one spouse who has a higher level of resources than his/her partner is likely
to do less housework. In sum, both models address the gender division of labor from an
economic perspective, and many studies offer empirical evidence to support them (Bianchi et al.
2000; Blair and Lichter 1991; Brines 1994; Coverman 1985; Davis & Greenstein 2004; England
& Farkas 1986; Greenstein 2000; Kamo 1988; Ross 1987, South & Spitze 1994). Yet, these
economic-based approaches do not provide a thorough explanation for disproportionate shares of
domestic labor performed by women regardless of their economic resources in many countries.
Meanwhile, feminist scholars have argued that gender itself has a crucial effect on the
mechanisms in gender practices within the domestic sphere. Most notably, West and Zimmerman
(1987) provide the concept of “doing gender” that highlights that gender is socially constructed
by displaying proper gender identity and by doing appropriate gender behaviors based on
gendered social orders. Given this conceptualization of doing gender, men and women are
socialized in accordance with their assigned gender. In this regard, gender ideology and gender
role expectations, which individuals internalize through their daily life, play a significant role in
explaining the division of labor between men and women. Furthermore, in light of the concept of
family as the basic unit of society where daily practices and interactions occur, scholars argue
that the family becomes a “gender factory” that creates both the material (i.e., housework) and
symbolic (i.e., gender) productions of the households (Berk 1985). This claim provides
important insights into understanding how gender is (re)constructed as a routine accomplishment
embedded in everyday life within families. As a result, these feminist conceptual frameworks
have considerably influenced family scholarship. Empirical evidence has also been well
documented (Blair and Lichter 1991; Brines 1994; Davis & Greenstein 2009; Kamo 1988, 1994;
Pittman and Blanchard 1996; Presser 1994; Shelton & John 1996). Nowadays, most research on
examining gender-based inequalities in families builds upon both approaches.
2.1.3. Cross-national Differences in Changing Gender Relations and the Importance of
National Context
While the impact of individual-level factors on gender relations in families is well
documented, the importance of national contexts has received increasing scholarly attention in
that the patterns of domestic gender relations and its effects on individuals’ family life vary
across countries. Previous studies have found that individual-level processes of domestic gender
relations operate differently in different national contexts, which can be explored through
variables such as the level of economic development, female labor force participation, welfare
regimes, social policies, divorce and cohabitation propensity, and the degree of gender equality
typically measured by the Gender Empowerment Measure (GEM) and aggregate gender role
ideology in countries (Batalova and Cohen 2002; Braun et al. 2008; Davis and Greenstein 2004;
Fuwa 2004; Fuwa and Cohen 2007; Geist 2005; Geist and Cohen 2011; Greenstein 1996, 2009;
Hook 2006, 2010; Knudsen and Wærness 2008; Ruppanner 2010a, 2010b; Stier et al. 2001;
Yodanis 2005).
Modernization and concomitant industrialization are associated with changes in family
life and behavior. Although the relationship between modernization and gender equality in the
domestic sphere is not very clear (for example, see Jackson 1998; Mies 2014), their development
in social, economic, and political areas might affect changing dynamics of domestic gender
relations. On the one hand, structural and institutional developments in modern societies provide
greater opportunities for women in education and employment, which causes the necessity for
women to reduce their load of unpaid work, while the increasing demand for men to participate
in domestic labor (Hareven 1976). On the other hand, the development of the service economy
and technology provides the new possibility of outsourcing housework to the market (Cohen
1998). In this regard, the degree of modernization and industrialization, which is typically
measured by a country’s gross domestic product (GDP) and female labor force participation, is
likely to relate to changes in gender relations in families.
Social policy also appears to have an important impact on individuals’ processes of
organizing family life between couples (Fuwa and Cohen 2007; Hook 2006, 2010; Stier et al.
2001). For example, social policy that inhibits gender discrimination in the labor market (e.g.,
prohibitions to employment discrimination against women) and that encourages women to
participate in the labor market (e.g., affirmative action) can lead to the necessities of reorganizing
the traditional divisions of labor between men and women. On the contrary, social policy
assuming a sole earner in the household (e.g., conservative taxation, and limited or absence of
parental leave) might strengthen traditional gendered relations in families by discouraging
women from labor market participation. At the same time, other social policies, such as national
employment hours, and the presence and availability of public childcare services, may also affect
organizing individuals’ processes of domestic gender relations.
In a similar vein, cross-national variations in welfare regimes are associated with gender
relations in families. The classical welfare state classification has been given by the Esping-
Andersen’s (1990) categorization into three welfare regimes, labeled as liberal (e.g., the United
States), conservative (e.g., Germany), and social democratic (e.g., Sweden), respectively.
Empirical findings have shown that in countries with a conservative regime, couples are less
likely to share housework equally in general (Fuwa 2004; Geist 2005). In this sense, it may be
the case that the negative effects of unequal domestic labor on individuals’ well-being and family
life may be weaker in less egalitarian counties because the traditional gender norms and beliefs
embedded in social systems and orders may justify the unequal gender relations in the domestic
sphere.
In addition, there is evidence of the effects of cross-national differences in “divorce and
cohabitation cultures” on gender relations between married couples. Yodanis (2005) finds that a
“divorce culture,” which is a composite measure of attitudes toward divorce and divorce rates on
the national level, appears to contribute to more equal marital relationships between men and
women. On the other hand, Batalova and Cohen (2002) document that couples’ premarital
cohabitation experience is positively associated with husbands’ housework participation. These
results support a hypothesis that in a country where alternatives to marriage are widely available,
women may have greater bargaining power to demand more gender-equitable relationships.
Meanwhile, in a society with limited alternatives to marriage, women may have less power to
negotiate the domestic division of labor with their partners because women are most likely to
stay in the current marital relationship regardless of how dissatisfied they are with their unequal
status in marriage (Batalova and Cohen 2002; Yodanis 2005).
Especially given the gendered nature of the family, it is not surprising that country-level
gender equality and gender role ideology, in most studies, are consistently observed as a
significant factor that influences the effects of individual-level factors on domestic gender
relations. In a society characterized by structural, institutional, and cultural factors that promote
and favor gender equality, women tend to share household labor more equally with their partners
in a given national context that views gender equality as more preferable and desirable.
Therefore, women who live in more egalitarian countries may perceive the unequal division of
labor as more unfair because they can easily find a comparable couple who shares domestic labor
more equally, which makes their unequal situation more unjust (Braun et al. 2008; Davis 2010;
Greenstein 2009; Jansen et al. 2016; Öun 2013; Ruppanner 2008). Likewise, when it comes to
the effects of the unequal division of household labor on work-family conflict and subjective
well-being, the negative effects of the unequal divisions of housework may be stronger in more
egalitarian countries through this comparison process (Greenstein 2009; Ruppanner 2010).
Meanwhile, it is important to differentiate country-level gender equality from
countrylevel gender ideology. The country-level gender equality refers to the degree of national-
level gender equality in social, political and economic areas. This is typically measured by the
Gender Empowerment Measure (GEM) score, gender wage gap, female labor force participation
rates, and sociopolitical structures of support for gender equality in relation to the type of welfare
regimes and the effect of social policy. The country-level gender ideology refers to how people
think about the proper roles for men and women. This measurement is typically constructed by
aggregating the responses to a set of statements about individuals’ attitudes toward gender roles,
such as “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 job is all right, but what
most women really want is a home and children.” Although both are frequently used to measure
the degree of gender equality in a given society, it is important to note that the underlying nature
of those two measures are different and that there might be a discrepancy between them
depending on individual and national contexts. For example, there may be considerable
differences in gender role ideology between men and women at the country-level even though
both men and women live under the same social, political, and economic settings in a given
society. It is also possible that the direction and magnitude of correlation between them may
differ across countries.
Thus far, I have broadly dealt with the main drivers for the recent changes in gender
relations within families, and how individual and national contexts are associated with such
changes, with an emphasis on a gender perspective and cross-national differences. In the
following sections, I will focus on three specific aspects of domestic gender relations, which
have been key issues of comparative family scholarship in the context of social changes and
changing gender roles over the past decades: the division of household labor, work-family
conflict, and family life satisfaction.
2.2. Stalled Revolution? Continuity and Changes in the Division of Household Labor
Women’s participation in paid work has been expected to lead to men’s increasing
participation in unpaid labor. However, unlike anticipation, men have not yet become
comparably involved in domestic labor, whereas female labor force participation has gradually
increased over the past several decades in most industrial countries. Women still perform a larger
share of domestic labor than their male partners over the globe and even in the most industrial
countries (Bianchi et al., 2000). This phenomenon is often described by the well-known term
“stalled revolution,” coined by Hochschild (1989). This term indicates that women remain
largely responsible for household labor and care work at home despite substantial social changes
related to changing gender roles. To examine factors that are associated with this enduring
continuity of the gendered division of household labor, a substantial body of theoretical
approaches—such as relative resources, time availability, and gender ideology—have been
proposed, and much of them are well supported empirically by individual-level variables
including earnings, paid work hours, and gender role attitudes (Bianchi et al. 2000; Brines 1994;
Coltrane and Ishii-Kuntz 1992; Kamo 1988; Sorensen and McLanahan 1987; South and Spitze
1994).
While individual-level factors influencing the existence and persistence of gendered
division of housework are well documented, comparative scholars highlight the importance of
national contexts that have differing effects for individuals’ processes in the distribution of
housework between partners (Cooke and Baxter 2010). For example, Fuwa (2004) finds a
significant association between the division of housework and the various macro-level factors,
such as gender inequality, economic development, female-labor force participation, gender
norms, and welfare regimes. Geist (2005) shows the important link between welfare state
policies and the domestic division of labor. Meanwhile, Hook (2010) finds that public policies
that are particularly influential to women’s employment and individuals’ work-family life affect
the unpaid work behaviors of men. Yodanis (2005) finds that a strong divorce culture in a society
is related to a more equal division of housework, and Batalova and Cohen (2002) reveal the
association between premarital cohabitation experiences and the equal sharing of household
labor along with equalizing effects of national cohabitation rates.
More recently, scholarly attention has been given to exploring how national patterns in
the division of household labor change over time. In research that analyzes time use survey
between 1965 and 2003 from multiple countries, Hook (2006, 2010) finds that cross-national
differences in time spent on housework between men and women have decreased over time.
Furthermore, Geist and Cohen (2011), using data from 13 countries between 1994 and 2002,
conclude that cross-national differences in housework patterns have converged over time toward
greater gender equality. But, at the same time, they emphasize that it does not mean that such
convergence will lead to complete equality in the future because this is mostly attributable to the
shift of housework patterns in most traditional countries, and even in most egalitarian countries
women still do the majority of housework (Geist and Cohen 2011: 842-843).
Overall, previous research on the division of household labor provides valuable insight
into the mechanism underlying household gender inequality in terms of both individual- and
societal-level characteristics. However, partly because of the lack of datasets, almost no attention
has been paid to the impact of personal preferences with regard to the family distribution of paid
and unpaid labor on the actual divisions of housework. The attitudes toward preferred
workfamily arrangement between partners are likely to be associated with the actual divisions of
domestic labor. For instance, women who prefer the traditional divisions of labor may participate
less in paid work, do more housework, and be less egalitarian. On the contrary, women who do
not favor the traditional division of labor may participate more in the labor market, perform less
housework, and be more egalitarian. Disentangling the influence of preferences on work-family
arrangement on the actual division of household labor can contribute to the previous literature on
the division of household labor by testing the association of housework divisions with previously
unexplored characteristics.
In this regard, preference theory elaborated by Hakim (2000, 2003) provides a useful
framework for the relationship between personal preferences on the division of labor and the
actual distribution of housework. Preference theory posits that women are heterogeneous in their
preferences and priorities for work-family life, and they have genuine choices about how to
shape their work and family lives on the basis of three different lifestyles: adoptive,
workcentered, or home-centered (Hakim 2003: 355, 358-359). The distinctive feature of
preference theory involves the causal explanation between personal preferences and behavior.
Hakim (2003) argues that personal preferences are defined as causal factors in modern affluent
societies, and in broad terms preferences can predict outcomes (Hakim 2003: 355, 364).
Although there are ongoing debates over the aptness of preference theory for explaining
women’s lives to combine work and family life (for example, see Leahy and Doughney 2006),
some ideas of preference theory—such as the growing importance of personal preferences as an
influential factor for work and family life in modern societies, and the causal explanations
centered on women’s preferences on their work-family choices—can help us deepen our
understanding of the underlying mechanisms of the domestic division of labor.
Given the link between personal preferences and individuals’ choices for work and family
life, the impact of preferences on the division of household labor is likely to be stronger in
modern societies, in which individuals gain more freedoms and options for their family and
personal lives with the spread of individualization. Sociological theories of late modernity argue
that unlike the past, in which family and personal life was largely regulated by the social norms,
the power of social norms and institutions has weakened in contemporary societies, while the
role of personal choice to make a decision about individuals’ lives has become important (Beck
and Beck-Gernsheim 1995, 2002; Giddens 1991, 1992). In his work that examines the changes in
the meaning of marriage during the past few decades, Cherlin argues that marriage in the United
States (and possibly in other Western modernized countries also) has undergone a process of
deinstitutionalization along with the weakening of social norms and the expanding roles of
personal choice (Cherlin 2004: 848). These works suggest the importance of personal choice for
family life in modern societies on the one hand and the growing roles of personal preferences in
relation to their decision-making on the other.
In order to examine the influence of preferences on work-family arrangement on the
actual divisions of housework, it is important to consider national contexts because
individuallevel gender relations as well as expectations for work and family life are socially
constructed (West and Zimmerman 1987; Ferree et al. 1999). Country-level gender equality is
particularly important to understanding cross-national variations in the division of household
labor. Previous comparative studies have shown that housework arrangements between women
and men considerably vary depending on country-level gender equality (Batalova and Cohen
2002;
Knudsen and Wærness 2008; Fuwa 2004; Davis and Greenstein 2004; Geist and Cohen 2011;
Greenstein 2009; Ruppanner 2010a; Treas and Tai 2016). Specifically, individuals in egalitarian
countries tend to divide household labor with their partner more equally. Although work-family
arrangement has not yet been explicitly examined in a cross-national setting, it is reasonably
expected that individuals in egalitarian countries may prefer to organize their work and family
life in an egalitarian way.
Another important structural condition related to preferences on work-family
arrangement and the divisions of household labor is women’s labor force participation in each
country. With women’s increased participation in the labor market, the dominant family model
has changed from the male-breadwinner/female-homemaker to the dual-earner model in most
industrial societies, which necessarily leads to the need of changes in work-family arrangements.
Hence, in countries with high rates of female labor force participation, both women and men are
likely to prefer dual-earner family models (either part-time or full-time) than a traditional male
breadwinner model. Nonetheless, women’s increased labor force participation has not yet
comparably led to men’s increasing involvement in domestic labor (Bianchi et al. 2000; Coltrane
2000; Hochschild 1989). Given little changes in men’s housework participation, reducing the
amount of time spent in domestic labor would be the best way for women to deal with their
double burden in both paid work and domestic labor in a society where women are expected to
participate more in paid work as an important income earner for households. Thus, it is plausible
that women in countries with higher rates of female labor force participation may do less
housework, while men’s household labor may be relatively unaffected by country’s female labor
force participation.
Along with country’s gender equal climate and female labor force participation,
economic development is often regarded as one of the key national contexts that have a major
influence on changes in traditional gender roles and in the way that people organize their work
and family lives due to social changes caused by economic development (Fuwa 2004; Jansen et
al., 2016; Knudsen and Wærness 2008; Greenstein 2009; Hu and Yucel 2018). Researchers also
find that cross-national differences in state regimes have implications for the division of
household labor in terms of gender inequality at home (Davis and Greenstein 2004; Geist 2005;
Kornrich and Eger 2016).
Thus far, I have reviewed prior studies about the division of household labor by putting
emphases on the importance of comparative perspectives and on the void that has received
limited attention from family scholarship. Previous comparative research on housework has
revealed that national contexts play a significant role in understanding the gender division of
household labor. Yet, there is a lack of studies which examine the effects of preferences on work-
family arrangement regarding the actual divisions of housework from a comparative perspective,
despite the expanding roles of personal preferences and choices for their work and family lives in
modern societies. This calls for the further cross-national research on the division of household
labor with a consideration of the explanatory power of personal preferences, which
I address in Chapter 4.1 of my dissertation.
2.3. Work-Family Conflict and Gender Inequality in Household Labor
Work-family conflict has been the subject of extensive research in the past few decades
(Bianchi and Milkie 2010; Eby et al. 2005). With the considerable advancement of women’s
labor force participation, the dominant family model has been largely shifted to the dual-earner
families from the breadwinner model in most industrial countries. Unlike couples in the
traditional family model, in which men take primary responsibility for earning in the labor
market while women play a major role in household and care work at home, couples in the
dualearner household need to deal with the (re)distribution of paid and unpaid work between
partners. As a consequence, work-family tensions and/or incompatibilities has been growing for
both women and men (Nomaguchi 2009; Winslow 2005). Previous findings show that
workfamily conflict has the deleterious effects on individuals’ lives in general. Work-family
conflict is associated with a number of negative consequences, including psychological distress,
reduced satisfaction with job and family, less occupational commitment, increased marital
tensions, and poorer performance of the parenting role (for a review, see Eby et al. 2005; Frone
2003; Greenhaus and Parasuraman 1999). Given this, examining factors relating to work-family
conflict has become an important issue of social science scholarship and public policy (Allen et
al. 2000).
Work and family researchers have attempted to identify the different forms and types of
work-family conflict (Netemeyer et al. 1996). Notably, Greenhaus and Beutell (1985) define
work-family conflict as “a form of interrole conflict in which the role pressures from the work
and family domains are mutually incompatible in some respect” (p. 77). This definition is based
on theories of role conflict and interrole conflict (Kahn et al. 1964). Greenhaus and Beutell
(1985) assume that the demands in the workplace, which require a certain amount of time and
energy for individuals, make them difficult to fulfill the demands at home and vice versa. Given
this, they propose three forms of work-family conflict: (a) time-based conflict, (b) strain-based
conflict, and (c) behavioral-based conflict. More specifically, time-based conflict occurs when
time spent on requirements within one role makes it difficult to fulfill requirements of another,
strain-based conflict occurs when the strain from activities within one role makes it difficult to
comply with the demands of another, and behavioral-based conflict occurs when specific
behavior in one role is not compatible with expected behaviors in another role (pp. 77-82).
Nowadays, the definition of work-family conflict proposed by Greenhaus and Beutell’s (1985) is
most widely used in work and family research.
While earlier studies contributed to distinguishing the different forms of work-family
conflict, the direction of work-family conflict has received attention in recent years (Duxbury et
al. 1994; Frone et al. 1992; Gutek et al. 1991). Although work-family conflict occurs when role
requirements from the work and family domains are not compatible with each other, its features
and aspects might vary depending on whether such conflict results from work interfering with
family (WIF) or family interfering with work (FIW). For example, Gutek et al. (1991) find that
conflict arising from family responsibilities (FIW) operates differently than conflict arising from
paid work duties (WIF) using two separate samples of psychologists from the American
Psychological Association and senior managers in a university-based executive education
program. Given these findings, they suggest that WIF and FIW are separable and relatively
independent of each other. Additionally, Carlson and her colleagues (2000) further a
multidimensional measure of work-family conflict by combining three forms (time, strain, and
behavior) and two directions (WIF and FIW) of conflict. After performing a set of analyses with
those items, they conclude that each of the scales is differentially related to consequences of
work-family conflict. These findings indicate the importance of considering both forms and
direction of work-family conflict in a given context.
Given the salient adverse effect of work-family conflict and the increasing sense of
workfamily conflict across the globe over the past decades, a large literature has examined
factors that contribute to the change in work-family conflict (Allen et al. 2012; Batt and Valcour
2003; Dex and Bond 2005; Grzywacz and Marks 2000; Kinnunen and Mauno 1998; Kinnunen et
al. 2004; Maume and Houston 2001; Voydanoff 2007; White et al. 2003). Based on the nature of
workfamily conflict that makes role pressures in the work and family domains mutually
incompatible, most studies have been done on two main axes: One is work characteristics that
make it difficult to fulfill the family demands, and the other is family characteristics that make it
difficult to fulfill the work demands. With respect to work characteristics, prior research finds a
strong relationship between job characteristics, workplace conditions, and work-family conflict
(Batt and Valcour
2003; Grzywacz and Marks 2000; Maume and Houston 2001; White et al. 2003). Grzywacz and
Marks (2000) find that work and family factors that facilitated development (e.g., decision
latitude and family support) are associated with less negative and more positive spillover
between work and family, while work and family barriers (e.g., job pressure and family
disagreements) are associated with more negative spillover and less positive spillover between
work and family. Using a sample of white-collar workers, Maume and Houston (2001) find that
work hours are positively associated with work-family conflict. White et al. (2003) demonstrate
that alongside long work hours, employer’s ‘high-performance’ practices (e.g., appraisal
systems, group-working practices, and performance-related pay) have a negative impact on
work-family conflict. These results indicate that job characteristics play a significant role in
producing work-family conflict.
Previous studies also show that variables related to the roles and demands within families
are significantly associated with work-family conflict (Dex and Bond 2005; Kinnunen and
Mauno 1998; Kinnunen et al. 2004; Maume and Houston 2001; Voydanoff 2007). For instance,
the presence of children is strongly associated with work-family conflict. Having a child not only
requires more time and effort in terms of child care but also increases overall time spent on
housework. This makes it difficult to fulfill the demands of other roles at work, and thus people
who have children are more likely to experience work-family conflict (Voydanoff 2007). Gender
also appears to be one of most powerful determinants of who experience more or less
workfamily conflict. Even though the dominant family model has been largely shifted to the
dualearner families from the breadwinner model in most industrial societies, women still do the
majority of household labor in a context of the stalled revolution (Bianchi et al. 2000; Coltrane
2000; Hochschild 1989). The disproportionate family distribution of paid and unpaid labor may
create more role pressures and interrole conflict for women, which results in gender differences
in work-family conflict (Cinamon and Rich 2002; Duxbury and Higgins 1991; Duxbury et al.
1994). In a similar vein, housework divisions may also have uneven impacts on men’s and
women’s work-family conflict. Given the gendered nature of housework, women are expected to
be more responsible for domestic labor than men, regardless of their socioeconomic status.
Therefore, even if men and women spend the same amount of time on their work and family
lives, women likely feel more pressures and strains for their family responsibilities than men
because of gendered social norms and role expectations that impose the primary family
responsibilities for women (Milky and Peltola 1999). Such a gendered nature of household labor
may lead to the stronger effects of unequal divisions of housework on work-family conflict for
women than men.
Aside from individual-level characteristics, country-level characteristics play a crucial
role in people’s experiences of work-family conflict (Allen et al. 2015; Ollier-Malaterre et al.
2013). Earlier comparative research on work-family conflict utilized a small number of countries
largely centered on Europe (e.g., Crompton and Lyonette 2006; Gornick and Meyers 2003;
Hochschild 1997; Nordenmark 2004; Ruppanner 2008; Strandh and Nordenmark 2006).
However, with the increased availability of cross-national datasets, growing attention has been
paid to cross-national variations in work-family conflict by considering various macro-level
contextual factors (Edlund 2007; Stier et al. 2012). Using the sample of working couples in 29
countries, Edlund (2007) shows that the likelihood of achieving work-family balance varies
systematically according to the level of modernization (i.e., GDP per capita) and types of family
regimes (i.e., the market-oriented, familialist, and de-familialist regime). He further finds that
although the overall probability of work-family balance increases with the wealth of a country,
women’s disadvantage of work-family conflict persist because modernization transforms the
female disadvantage from a dual work-overload situation in poor countries to an occupational
work-overload in the rich countries (Edlund 2007). On the other hand, Stier and colleagues’
(2012) study of the relationship between social policies and work-family conflict in 27 countries
shows that women on average report higher levels of work-family conflict than men, and that
working long hours, having young children, and holding more demanding jobs all contribute to
increased work-family conflict at the individual-level, while the state policy providing day care
centers for young children allow parents to better balance their work and family demands at the
country-level (Stier et al. 2012). To sum up, the recent cross-national research on work-family
conflict underscores the important role of national contexts played in individuals’ work-family
conflict, as well as gender differences in work-family conflict.
Among macro-level contextual factors, the significant role of country-level gender
equality in family and housework are well documented, but the effect of country-level gender
equality on work-family conflict in terms of the gender division of household labor has not been
explicitly examined. Nonetheless, previous cross-national studies provide important insight for
the possible influence of country-level gender equality on work-family conflict. Findings show
that country-level gender equality—typically measured by country-level gender equity or
country’s average gender ideology—plays a significant role in family processes, such as the
divisions of household labor, perceptions of fairness of that division, satisfaction with family life,
and marital quality (Bataloa and Cohen 2002; Braun et al. 2008; Davis 2010; Forste and Fox
2012; Fuwa 2004; Geist and Cohen 2011; Greenstein 2009; Hu and Yucel 2018; Jansen et al.
2016; Kornrich and Eger 2016; Knudsen and Wærness 2008; Öun 2013; Ruppanner 2010a,
2010b; Treas and Tai 2016). Women in more egalitarian countries not only share housework with
their partners more equally but also benefit more from their individual-level assets in the
negotiation over housework with their partners (Batalova and Cohen 2002; Knudsen and
Wærness 2008; Fuwa 2004; Greenstein 2009; Ruppanner 2010; Treas and Tai 2016).
Furthermore, people living in countries with a higher level of gender equality tend to strongly
perceive a large share of housework as unfair, and the effect of the actual division of housework
on perceived fairness is stronger in those countries (Braun et al. 2008, Greenstein 2009; Jansen et
al. 2016; Öun 2013). Similarly, country-level gender equality appears to be a significant
predictor of women’s family life satisfaction, and the negative effect of perceived unfairness on
family life satisfaction is also stronger in countries with a higher level of gender equality
(Greenstein 2009; Hu and Yucel 2018; Knudsen and Wærness 2008; Kornrich and Eger 2016s).
This line of research underscores that mechanisms of the division of household labor and
relatedfamily processes are highly associated with country-level gender equality.
Considering the close link between the division of household labor and work-family
conflict, the similar process can be applied to explaining cross-national variations in work-family
conflict. Previous comparative studies to date have shown that housework arrangements between
women and men considerably vary depending on country-level gender equality, and people in
egalitarian countries tend to share household labor more equally with their partner (Fuwa 2004;
Geist and Cohen 2011; Treas and Tai 2016). Therefore, individuals—particularly women—in
more egalitarian countries may experience less work-family conflict due to the relatively equal
divisions of household labor. On the other hand, it might also be possible that people in
egalitarian countries may report more work-family conflict because they are likely to be more
sensitive and/or have higher expectations to gender equality within the family, and thus they may
perceive gendered responsibilities of household labor as more unfair.
Although a particular direction of the effect of country-level gender inequality on
workfamily conflict still needs to be tested, the recent meta-analytic research provides useful
insights for the relationship between country-level gender equality and work-family conflict. By
metaanalytically examining mean differences in work-family conflict across multiple macro-
level factors, Allen and colleagues (2015) conclude that work-family conflict is higher in
countries with a higher gender gap. In other word, individuals who live in countries with a
higher, rather than lower, level of gender equality experience less work-family conflict. They
also find that among various macro-level factors, country-level gender equality—measured by
the gender gap index—appears to be the most powerful differentiator (Allen et al. 2015). Based
on these findings, it is expected that individuals in countries with greater gender equality may
experience less work-family conflict, compared to their counterparts living in countries with a
lower level of gender equality.
Meanwhile, considering work-family conflict as a gendered process, the influence of
country-level gender equality on work-family conflict may differ by gender. While much of
comparative research investigating the role of country-level gender equality in terms of the
division of housework has be done with the female sample (i.e., Fuwa 2004, Greenstein 2009,
Braun et al. 2008) or without considering the interactive effect of gender (i.e., Forste and Fox
2012), recent studies examining both female and male samples find gender differences in the
influence of country-level gender equality. Jansen and colleagues’ (2016) results show that
country-level gender role attitudes significantly affect perceived fairness of the division of
housework only for women. In the study examining individual- and national-level factors that
affect spouses’ absolute and relative housework, Knudsen and Wærness (2008) find that wives’
absolute amount of housework are primarily affected by GEM, while husbands’ are affected
more by GDP. On the other hand, Ruppanner’s (2010a) research on conflict over the divisions of
housework shows that national contexts, like societal gender equality and full-time female labor
force participation, affect women’s and men’s housework conflict in the same way. These mixed
results with respect to the role of country-level gender equality on family processes might be
partly due to differences in main variables of interest, methods, as well as data. Examining these
gaps can contribute to extending our knowledge in the cross-national research on work-family
conflict and the division of household labor as well. This suggests that cross-national research
examining work-family conflict should consider the possible gender differences in the influence
of national contexts.
Although important, comparative studies of work-family conflict are relatively scant
compared to other family-related research, such as the division of household labor, perceptions
of fairness of that division, and individuals’ well-being in family life. Examining work-family
conflict with respect to unequal gender relations from a comparative perspective has received
insufficient attention from family scholars. In the area of the gender division of household labor,
housework has been more often addressed as an outcome (Kluwer et al. 1996, 1997) or mainly
discussed as a predictor in a context of perceptions of fairness or subjective well-being. In the
area of work-family conflict, more attention has been paid to consequences associated with
work-family conflict (Allen et al. 2000). To my knowledge, the relationship between
individuallevel work-family conflict and country-level gender equality in a context of the gender
division of household labor has not been explicitly addressed. There is a notable exception
examining housework conflict with an emphasis on both individual-level housework and
national-level gender equality from a comparative perspective (Ruppanner 2010a). However,
strictly speaking, this research focuses on conflict over the division of housework (i.e., the extent
to which individuals have conflict with their partner about how to divide the housework), not
work-family conflict (i.e., interrole conflict that occurs due to the competing roles of the work
and family domains). This study also includes only European countries, so we cannot observe
cross-national differences from a broader international perspective. This suggests the need for
further comparative research on the association between work-family conflict, country-level
gender equality, and the division of housework using the dataset with a wider set of countries. I
seek to address this void in Chapter 4.2 of my dissertation.
2.4. What Makes People Satisfied with Family Life? The Role of Division of Household
Labor, Perceptions of Fairness, and Work-Family Conflict in Family Life Satisfaction
As stated, despite considerable improvements in women’s educational attainment and
labor force participation, women still take primary responsibility for housework and childcare in
most societies, regardless of the individual- and national-level characteristics (Hook 2006; Geist
and Cohen 2011). Given disproportionate domestic responsibilities between women and men, a
number of studies have examined how the gendered division of household labor affects
individuals’ subjective well-being, such as marital satisfaction, life happiness, and psychological
distress (Baxter 2000; Baxter and Western 1998; Bird 1999; Erickson 1993; Greenstein 1996;
Hawkins et al 1995; Kluwer et al. 2002; Lye and Biblarz 1993; MacDonald et al. 2005; Major
1993; Piña and Bengston 1993; Roxburgh 2002, 2004; Sanchez 1994; Voydanoff and Donnelly
1999; Yogev and Brett 1985). Overall, prior studies demonstrate that the unequal division of
housework has a significant negative effect on subjective well-being, particularly for women.
Among a set of well-being indicators, family life satisfaction has received continued
interest. Satisfaction with family life is not only a good indicator of assessing a level of one’s
well-being within the family processes but also a significant factor to be linked to mental and
behavioral health problems, such as depression, anxiety disorder, suicidality, and substance
dependence (Headey et al. 1993, Fergusson et al 2015). Given the persistence of the gender
division of household labor, growing attention has been paid to individuals’––largely women’s––
satisfaction with family life as the outcome of gender inequality at home. Because men do
substantially less housework than women in most societies, women’s disadvantage of family life
satisfaction with respect to the unequal divisions of household labor is reasonably expected. In
this regard, previous literature has focused much more on women’s satisfaction than men’s
(Kornrich and Eger 2016) and considered a mediating or intervening factor on such a negative
association between the unequal division of household labor and family life satisfaction
(Coltrane 2000).
Notably, most studies find that individuals’ gender role ideology is extremely important
to the relationship between the housework divisions and family life satisfaction (Baxter 2000;
Baxter and Western 1998; Greenstein 1996; Kluwer et al. 2002; Major 1993; Roxburgh 2004;
Sanchez 1994; Voydanoff and Donnelly 1999). Gender ideology matters because people’s
lifequality assessment largely depends on how they judge their division of labor as fair or unfair,
and such a judgement is closely linked to gender role expectations that they hold. Even though
women do the lion’s share of housework, they may perceive current unbalanced housework
arrangements as fair if they strongly agree with the traditional division of labor based on
stereotypical gender role expectations (Benin and Agostinelli 1988; Berk 1985; Greenstein 1996;
Pleck 1985; Yogev 1981). Accordingly, the negative effect of unequal division of housework on
their satisfaction will decrease, compared to their egalitarian counterparts (Greenstein 1996).
This suggests that the association between the housework divisions and family life satisfaction is
not straightforward, rather an interaction involving gender ideology and housework division
should be taken into account as a key mechanism of this assessment process regarding family life
satisfaction. This notion is well supported by an extensive body of literature showing that the
antecedents, consequences, and interpretations of family processes are conditional upon
individuals’ gender role ideology (Benin and Agostinelli 1988; Berk 1985; Lavee and Katz 2002;
Mikula 1998; Pleck 1985).
While previous household labor research before the 1990s focused more on the actual
division of household labor (i.e., one’s weekly hours spent on housework or one’s share of
housework), growing attention has been paid to perceptions of fairness of that division (Blair and
Johnson 1992; Greenstein 1996; Hawkins, Marshall, and Meiners 1995; Sanchez and Kane 1996;
see also Coltrane 2000). It is typically expected that individuals—mostly women—who do a
larger share of housework are more likely to perceive their share of housework as unfair.
However, despite unbalanced distribution of household labor between women and men, many
women and men feel such unequal distributions of housework as fair (Mikula 1998; Lennon and
Rosenfeld 1994; Greenstein 1996). This suggests that the actual division of household labor does
not directly determine perceptions of fairness of that division, and other factors may play a role
in shaping perceptions of fairness with respect to the actual division of housework.
Understanding such a discordance between actual and perceived inequality in the division of
housework is important and perceptions of fairness play a significant role in the link between the
actual division of household labor and subjective satisfaction with family life (Braun et al. 2008;
Greenstein 1996, 2009; Hu and Yucel 2018; Lavee and Katz 2002; Lennon and Rosenfield
1994). In this regard, researchers attempted to find factors affecting fairness evaluations about
housework. Specifically, findings show that women who are less educated, who are not
employed in the labor market, and who hold more traditional gender role ideology perceive the
current unequal housework divisions to be fair (Blair and Johnson 1992; DeMaris and Longmore
1996; Greenstein 1996; Lennon and Rosenfield 1994; Robinson and Spitze 1992; Sanchez and
Kane 1996). Among these factors, gender ideology appears to play a crucial role, as expected
given the gendered nature of the division of household labor. Gender ideology significantly
moderates not only the effect of housework on perceived fairness, but also the effect of perceived
fairness on women’s marital quality (Greenstein 1996). These findings underline the importance
of perceived fairness of the housework division to life-quality assessment on the one hand and
the role of gender ideology as a key moderator in that process on the other.
Another key factor affecting family life satisfaction is work-family conflict. During the
past decades, there have been substantial changes in gender roles, particularly in paid work.
Women’s continuing integration into the labor market has led to a considerable increase in
dualearner couples who need to cope with their dual roles as both economic and care providers
of the household, which results in growing work-family tensions and/or incompatibilities for
both women and men. Therefore, examining antecedents and consequences of work-family
conflict has become an important issues for social science scholars and policy makers (Allen et
al. 2000). Examining the role of work-family conflict in the association between the divisions of
housework and family life satisfaction is especially important in light of the consistently adverse
effects of work-family conflict. Evidence shows that work-family conflict has significant,
deleterious impacts on individuals’ lives in general (for a review, see Eby et al. 2005; Frone
2003; Greenhaus and Parasuraman 1999). Work-family conflict is associated with a number of
negative consequences, including psychological distress, reduced job satisfaction, less
occupational commitment, increased marital tensions, and poorer performance of parenting roles.
Likewise, work-family conflict negatively affects individuals’ satisfaction with family life,
regardless of gender as well as the type of work-family conflict (e.g., work interfering with
family or family interfering with work). Given existing findings of significant, adverse effects of
work-family conflict, it is reasonably expected that work-family conflict may have a negative
influence on women’s and men’s family life satisfaction, along with the unequal division of
housework and perceptions of unfairness.
In addition to individual-level factors, national contexts also play a significant role in the
process of shaping women’s and men’s satisfaction. Recent comparative studies reveal that the
division of household labor and its relationship with family life are situated in cultural,
institutional, and economic aspects of country-level characteristics (Batalova and Cohen 2002;
Baxter 1997; Baxter and Tai 2016; Braun et al. 2008; Davis and Greenstein 2004; Fuwa 2004,
2014; Fuwa and Cohen 2007; Geist 2005; Greenstein 2009; Hori and Kamo 2014; Hook 2006,
2010; Jansen et al., 2016; Knudsen and Wærness 2008; Öun 2013; Ruppanner 2010a, 2010b,
2012; Stier et al. 2001; Yodanis 2005). In particular, country-level gender equality—which is
typically measured by GEM scores, gender wage gap, or aggregate gender role ideology—is
consistently observed as an important factor affecting the division of household labor as well as
housework-related family life (Batalova and Cohen 2002; Braun et al. 2008; Fuwa 2004;
Knudsen and Wærness 2008; Kornrich and Eger 2016; Greenstein 2009; Geist and Cohen 2011;
Hori and Kamo 2014; Hu and Yucel 2018; Öun 2013; Ruppanner 2010a, 2010b; Jansen et al.,
2016; Treas and Tai 2016). The division of household labor is embedded in social norms and role
expectations of the time. Thus, gender equal atmosphere in a given society may function as a key
contextual variable influencing the division of housework, as well as its relationship with family
processes. Economic development is often regarded as one of the key national contexts that have
a major influence on changes in traditional gender roles and in the way that people organize their
work and family lives due to social changes caused by economic development
(Fuwa 2004; Jansen et al., 2016; Knudsen and Wærness 2008; Greenstein 2009; Hu and Yucel
2018). Researchers also find that cross-national differences in state regimes have implications for
the division of household labor in terms of gender inequality (Davis and Greenstein 2004; Geist
2005; Kornrich and Eger 2016).
Comparative studies examining the relationship between the gender division of
housework and family life satisfaction is still scant, but recent cross-national studies shed light
on the importance of country-level gender equality on that association. (Greenstein 2009;
Kornrich and Eger 2016; Hu and Yucel 2019). Building on relative deprivation theory,
Greenstein (2009) finds that country-level gender equality moderates not only the association
between the actual division of housework and perceptions of fairness but also the association
between perceptions of fairness and family life satisfaction, for women. Kornrich and Eger
(2016) extend the scope of comparative research on family life satisfaction by considering men’s
family life satisfaction and perceptions of fairness, which is determined by different processes
from women’s. Conducting separate multilevel models for women and men, they find that
relative deprivation theory does not explain men’s family life satisfaction, while women’s
satisfaction is well explained by relative deprivation framework. Hu and Yucel (2018)
underscore distinct dimensions of women’s fairness comparison. They distinguish two major
dimensions of women’s fairness comparison (i.e., inter-gender relational comparison between
partners and intra-gender referential comparison within country) and conclude that not only do
two dimensions of fairness comparison operate independently of each other, but also
countrylevel gender equality plays out differently for each dimension. These studies provide
valuable insights for cross-national research on family life satisfaction, with respect to the
division of household labor and perceptions of fairness. Yet, the role of work-family conflict,
which has been consistently found as a significant factor affecting individuals’ family life
satisfaction, has received insufficient attention in this context. Given prior findings of the strong
negative effect of work-family conflict on individuals’ satisfaction, it is reasonably expected that
work-family conflict is negatively associated with family life satisfaction for both women and
men. Moreover, in light of the different nature between work-family conflict and gender division
of housework, it is expected that the possible negative effect of work-family conflict on family
life satisfaction would be largely independent of the negative effects of unequal housework
division as well as unfairness perceptions of that division.
Although there are few studies examining the association between work-family conflict
and family life satisfaction cross-nationally, the recent meta-analytic research on work-family
conflict offer insights into the important role of country-level gender equality in that association.
Allen and colleagues (2015) investigate cultural, institutional, and economic aspects of national
contexts in explaining mean-differences in work-family conflict across countries and find that
the country’s gender gap index is the most powerful differentiator. Work-family conflict appears
to be higher in countries with a higher gender gap. In other words, individuals in lower gender
gap countries (i.e., countries with a higher level of gender equality) report less work-family
conflict than their counterparts living in higher gender gap countries (i.e., countries with a lower
level of gender equality). The negative association between work-family conflict and greater
gender equality at the county-level might be because that individuals in more egalitarian
countries divide household labor more equally with their partner. On the other hand, more
egalitarian countries may likely have more developed social policies and infrastructure systems
to mitigate individuals’ work-family incompatibilities, which might help reduce the average
levels of work-family conflict in those countries. The link between work-family conflict and
country-level gender equality has an important implication for the moderating role of
countrylevel gender equality in family life satisfaction. According to relative deprivation theory,
individuals judge outcomes based on a comparative referent, and the choice of such a referent is
situated by national contexts (Crosby 1976). From this framework, the possible negative
association between work-family conflict and family life satisfaction would be stronger in
countries with a higher level of gender equality, since individuals who experience higher levels
of work-family conflict in more egalitarian countries would be more relatively deprived. Further,
there might be gender differences in the moderating effect of country-level gender equality, as
the recent empirical research finds that relative deprivation framework does not fit well in
explaining men’s satisfaction (Kornrich and Eger 2016). Taken together, this suggests that the
effect of work-family conflict on family life satisfaction may vary by country-level gender
equality, with the possible differential effect of country-level gender equality on women’s and
men’s satisfaction.
Researchers have examined the relationship between the unequal division of housework,
perceptions of unfairness of that division, and family life satisfaction across countries. To my
knowledge, however, the role of work-family conflict in the association between the division of
housework, perceived fairness, and family life satisfaction in a cross-national setting has not ever
been addressed, even though there is solid evidence of the strong adverse effect of work-family
conflict on the family processes. In Chapter 4.3, I expand on previous cross-national research on
family life satisfaction by disentangling the potential, negative effect of work-family conflict. I
also test the possible moderating role of country-level gender equality by investigating whether
and how the effect of work-family conflict varies by country-level gender equality, as well as
whether the moderating effect of country-level gender equality varies by gender.
2.5. Research Objectives and Hypotheses
In light of the literature I have reviewed so far, I examine three major topics of
comparative family scholarship with the following hypotheses in the rest of my dissertation.
In Chapter 4.1, I examine the relationship between individuals’ preferences on
workfamily arrangement and the actual division of household labor from a comparative
perspective. In a context of the growing individualization of family and personal lives in modern,
affluent societies, the actual division of housework is likely to be affected by individuals’
preferences on how they want to organize their work and family life with their partner. Thus, I
expect that personal preferences on work-family arrangement are associated with the actual
division of household labor for both women and men. Further, a growing number of comparative
research shows that national contexts play a significant role in explaining the division of
housework across countries. Building on previous findings, I expect that cross-national
variations in the division of housework in terms of preferences on work-family arrangement can
be explained by the macro-level differences in cultural, structural, and economic aspects of
context, which are measured by country’s egalitarian gender ideology, welfare state regimes,
female labor force participation, and GDP. Finally, in light of the nature of the division of
housework as the highly gendered process, I expect that there are gender differences in the
effects of individual- and country-level variables. As a result, I test the following three
hypotheses in Chapter 4.1:
Hypothesis 1: Individuals’ preferences on work-family arrangement are associated with the
actual division of household labor for both women and men.
Hypothesis 2: Individuals’ share of household labor is conditional upon national contexts, which
are measured by egalitarian gender ideology, female labor force participation, GDP, and
welfare state regimes.
Hypothesis 3: There are gender differences in the effects of individual- and country-level
variables on the division of household labor.
In Chapter 4.2, I address the association between country-level gender equality and
individual-level work-family conflict in a context of the gender division of household labor.
With the considerable improvement of women’s labor force participation, the dominant family
model has been largely shifted to dual-earner families in which couples struggle to balance their
work and family lives. Yet, women are still expected to be more devoted than men to housework
and child care in most societies. In this regard, I expect that women experience more workfamily
conflict than men, and the effect of household labor on work-family conflict are stronger for
women than for men. Moreover, consistent with previous research that find strong associations
between national contexts and the family processes, I expect that country-level gender equality
plays a significant role in explaining work-family conflict. Specifically, in light of recent meta-
analytic research showing the negative link between work-family conflict and greater gender
equality at the country-level (Allen et al. 2015), I expect that individuals living in countries with
a higher level of gender equality may experience less work-family conflict. Finally, given
inconsistent findings about gender differences in the effects of national contexts on the family
processes, I explore whether and how the effect of country-level gender equality on work-family
conflict differs for men and women, by testing two competing hypotheses. To sum up, I posit the
following hypotheses in Chapter 4.2:
Hypothesis 4: Women experience more work-family conflict than men.
Hypothesis 5: The effect of household labor on work-family conflict are stronger for women than
for men.
Hypothesis 6: individuals may experience less work-family conflict in countries with a higher
level of gender equality.
Hypothesis 7a: The effect of country-level gender equality on work-family conflict varies by
gender.
Hypothesis 7b: The effect of country-level gender equality on work-family conflict does not vary
by gender.
In Chapter 4.3, I investigate the association between work-family conflict and family life
satisfaction across countries. Recent cross-national research has revealed that the unequal
division of household labor and unfairness perceptions of that division are negatively associated
with family life satisfaction. However, the influence of work-family conflict on that association
has received insufficient attention. Given existing evidence showing the salient adverse effect of
work-family conflict on the family processes, I expect that work-family conflict is negatively
associated with family life satisfaction for both women and men, net of the housework division,
perceptions of fairness, and other individual controls. Moreover, based on the link between work-
family conflict and country-level gender equality, as well as building on the relative deprivation
framework, I expect that the possible negative association between work-family conflict and
family life satisfaction is stronger in countries with a higher level of gender equality. Finally, in
light of recent empirical finding showing that relative deprivation framework does not fit well in
explaining men’s satisfaction (Kornrich and Eger 2016), I expect that the moderating effect of
country-level gender equality on the association between work-family conflict and family life
satisfaction differs by gender. As a result, I test the following hypotheses in Chapter
4.3:
Hypothesis 8: Work-family conflict is negatively associated with family life satisfaction for both
women and men, net of individual controls.
Hypothesis 9: The negative association between work-family conflict and family life satisfaction
is stronger in countries with a higher level of gender equality.
Hypothesis 10: The moderating effect of country-level gender equality on the association
between work-family conflict and family life satisfaction differs by gender.
Chapter 3. Methodology
3.1. Data
I utilize data from the 2012 International Social Survey Programme (ISSP), a continuing
annual program of cross-national collaboration over 40 countries (ISSP Research Group 2016).
Given the purpose of this study aiming to analyze the cross-national variations in the division of
household labor, work-family conflict, and family life satisfaction, cross-national data sets
composed of a large number of countries are necessary. The ISSP is one of the few international
data sets that provide a broad array of cross-sectional surveys of nationally representative
samples of adults (Haller et al. 2009). Moreover, the 2012 ISSP is the most recent module on
family and changing gender roles, which provides a wealth of information about housework and
related aspects of family life across countries. In particular, the 2012 module contains measures
about the division of household labor, perceptions of fairness of housework division, workfamily
conflict, and family life satisfaction, along with a variety of measures about individuals’
socioeconomic and demographic characteristics. The 2012 ISSP module also includes a new
question about preferences on work-family arrangement, which is one of the key measures of
this study. As of 2014, the ISSP has 49 members, and the following 41 countries participated in
the 2012 module: Argentina, Australia, Austria, Belgium, Bulgaria, Canada, Chile, China,
Croatia, Czech Republic, Denmark, Finland, France, Germany, Great Britain, Hungary, Iceland,
India, Ireland, Israel, Japan, Korea (South), Latvia, Lithuania, Mexico, the Netherland, Norway,
the Philippines, Poland, Portugal, Russia, Slovakia, Slovenia, South Africa, Spain, Sweden,
Switzerland, Taiwan, Turkey, the United States, and Venezuela.
Among the countries participating in the 2012 ISSP module, I have excluded some from
the analyses. Turkey was excluded because of the lack of data on the number of children, which
is one of the key predictors affecting the amount of housework and related aspects of family life.
China and India were excluded due to a shortage on information on the number of hours spent on
paid work by the partner. China also has an unreliable measure for the education variable, which
is measured every three years until twelve years of education, and capped at 16 years. Finally, I
consider only persons who are at least 18 years old throughout the study.
3.2. Measures
3.2.1. Individual-Level Measures
Hours spent on household labor. The amount of time respondents spend on housework is
measured by the question, “On average, how many hours a week do you personally spend on
household work, not including childcare and leisure time activities?” Respondents are also asked
about the weekly housework hours performed by their spouse/partner, which measured by the
question “And what about your spouse/partner? On average, how many hours a week does
she/he personally spend on household work, not including childcare and leisure time activities?”
As such, this variable indicates the exact hours of household labor a respondent or their
spouse/partner performed, respectively, and both measures are only based on the respondent’s
report, not their actual spouse/partner’s report.
Respondents’ perceived share of household labor. Based on the above two measures, the
relative measure of household labor is calculated by dividing respondents’ hours spent on
housework by the sum of both respondents’ and their spouse/partner’s housework hours.
Although some households may outsource domestic labor by purchasing services or may receive
support from other family members (e.g., their parents), such information is not available in the
2012 ISSP data. Thus, this measure reflects the respondent’s relative share of the couple’s
housework time, without considering housework done by a third person. I express the
respondent’s relative share of housework as a percentage by multiplying by 100. As a result, the
measure ranges between 0 and 100, with 0 indicating that the respondent performs none of the
housework and 100 indicating that the respondent performs all of the housework.
Perceptions of fairness of the division of household labor. Perceptions of fairness of the
division of household labor are measured by the question, “Which of the following best applies
to the sharing of household work between you and your spouse/partner?” The response
categories are (a) “I do much more than my fair share of the household work,” (b) “I do a bit
more than my fair share of the household work,” (c) “I do roughly my fair share of the household
work,” (d) “I do a bit less than my fair share of the household work,” and (e) “I do much less
than my fair share of the household work.” Although perceived fairness of the division of
housework has been increasingly used in comparative housework research, this measure was
operationalized inconsistently. For example, Greenstein (2009) measured it as a continuous
variable, coded 1 = “I do much more than my fair share of the household work” through 5 = “I
do much less than my fair share of the household work,” so that “higher scores indicate higher
levels of perceived fairness of the division of household labor” (p. 1043). On the other hand, Öun
(2013) recoded this variable with three levels of categories, “where the highest level means that
the respondent sees the division of household work as fair. The middle level indicates that the
respondent or partner perceives that he or she does a bit less than his or her fair share of the
household work, and the lowest level means that the respondent or partner perceives that he or
she does much less than his or her fair share” (p.407). Finally, Jansen et al. (2016) measured this
as a multinomial variable by collapsing the original variable into three categories indicating
doing one’s fair share, doing more than one’s fair share, and doing less than one’s fair share
(p.58). Following Jansen and colleagues’ work (2016), I measured perceived fairness as a set of
two binary variables indicating that a respondent perceives it as “doing more than one’s fair
share of housework” or “doing less than one’s fair share of housework.” The reference category
is doing roughly one’s fair share of housework. As an additional robustness check, I estimated
the same regression models including four dummy variables and found almost identical
estimates.
Preferences on work-family arrangement. Preferences on work-family arrangement are
measured by the question, “Consider a family with a child under school age. What, in your
opinion, is the best way for them to organize their family and work life?” Response categories
are “The mother stays at home and the father works full-time,” “The mother works part-time and
the father works full-time,” “Both the mother and the father work full-time,” “Both the mother
and the father work part-time,” “The father works part-time and the mother works full-time,” and
“The father stays at home and the mother works full-time.” Because a small proportion of
responses often causes reliability issues, I excluded the last two responses from analyses (e.g.,
less than 1% for both responses). Then, I created a set of three binary variables, indicating
whether a respondent prefers (a) male full-time/female part-time, (b) both full-time workers, or
(c) both part-time workers model. The reference category is the male breadwinner/female
homemaker model.
Work-family conflict. Work-family conflict is measured by the responses to four
statements, which respectively indicate four different types of incompatibility of work and family
demands: “I have come home from work too tired to do the chores which need to be done,” “It
has been difficult for me to fulfil my family responsibilities because of the amount of time I
spent on my job,” “I have arrived at work too tired to function well because of the household
work I had done,” and “I have found it difficult to concentrate at work because of my family
responsibilities.” With regard to these four statements, respondents are asked “How often has
each of the following happened to you during the past three months?” The response categories
are, “Several times a week,” “Several times a month,” “Once or twice,” or “Never.”
For analytic purposes, responses were coded from “1 = Never” to “4 = Several times a week.”
Meanwhile, note that the first two questions are related to WIF (i.e., work interference with
family), while the last two are linked to FIW (i.e. family interference with work). In light of the
literature on the different nature between WIF and FIW, I conducted a factor analysis for these
four statements. The results showed that all four statements retain one factor with a high level of
reliability (alpha = .74). This is also consistent with the previous study that used the same
workfamily conflict items but in the different wave of 2002 ISSP data (Stier et al. 2012).
Therefore, I use the single factor structure of work-family conflict regardless of the direction of
work-family conflict, while keeping the different nature of WIF and FIW in mind.
Family life satisfaction. Respondents are asked, “All things considered, how satisfied are
you with your family life?” The response categories are “Completely satisfied,” “Very satisfied,”
“Fairly satisfied,” “Neither satisfied nor dissatisfied,” “Fairly dissatisfied,” “Very dissatisfied,”
and “Completely dissatisfied.” For analytic purposes, and consistent with previous studies
(Greenstein 2009; Kornrich and Eger 2016), I coded these responses ranging from “1 =
Completely dissatisfied” to “7 = Completely satisfied.” As a results, higher scores indicate
higher levels of family life satisfaction.
Egalitarian gender ideology. Individual-level egalitarian gender ideology is constructed
from the responses to seven statements about attitudes toward family and gender roles: (a) A
working mother can establish just as warm and secure a relationship with her children as a
mother who does not work, (b) A pre-school child is likely to suffer if his or her mother works,
(c) All in all, family life suffers when the woman has a full-time job, (d) A job is all right, but
what most women really want is a home and children, (e) Being a housewife is just as fulfilling
as working for pay, (f) Both the man and woman should contribute to the household income, and
(g) A man’s job is to earn money; a woman’s job is to look after the home and family. The
response categories range from “1 = Strongly agree” through “5 = Strongly disagree.” The first
and sixth items were reversed, so that higher scores reflect more egalitarian gender role attitudes.
I then conducted a factor analysis, and the results showed that all seven statements retain one
factor. I thus produced the egalitarian gender ideology index by summing those items and
dividing them by the number of completed responses, resulting in a relative high level of
reliability (Cronbach’s alpha = .75). This measure has been widely used in a number of
comparative studies (Braun et al. 2008; Fuwa 2004; Geist and Cohen 2011; Greenstein 2009; Hu
and Yucel 2018; Jansen et al. 2016).
Hours spent on paid work. The surveys include questions about the number of weekly
hours of paid work for both respondents and their spouses/partners. Both measures are used
when the analyses build on the time availability perspective.
Respondents’ share of paid work. Based on the above two measures, the relative measure
of respondents’ share of paid work is calculated by dividing respondents’ working hours by the
sum of working hours for both a respondent and his/her partner.
Relative income. Guided by the relative resources theory, respondent’s relative income is
considered. Respondents are asked the following question, “Considering all sources of income,
between you and your spouse/partner, who has the higher income?” The seven response
categories are “My spouse/partner has no income,” “I have a much higher income,” “I have a
higher income,” “We have about the same income,” “My spouse/partner has a higher income,”
“My spouse/partner has a much higher income,” and “I have no income.” This measure is coded
from 1 = “I have no income” to 7 = “My spouse/partner has no income.”
Age. Consistent with prior research (Greenstein 2009), age is measured in years and
modeled as a quadratic function to account for a commonly-observed non-linear relationship
between age and household labor.
Marital status. Marital status is measured as a dummy variable indicating a respondent is
married or in civic relationship.
Educational attainment. Due to considerable variations in education system across
countries, years of education are standardized by country using its mean and standard deviation
for each country.
Presence of children. Previous literature has shown that children in the household affect
not only the total amount of housework but also its relationship with family life (Bianchi et al.
2000; Jansen et al. 2016; South and Spitze 1994; Stier et al. 2012). The children variable is
measured by two dummy variables indicating whether a respondent has one child or two or more
children under 18 years of age in the household. The reference category is having no child.
3.2.1. Country-Level Measures
Country-level gender equality. Two different measures of country-level gender equality
are used in the analyses. The first is the Gender Inequality Index (GII) by the United Nations
Development Programme (UNDP). The GII is an inequality index, which measures gender
inequalities in three important aspects of human development—reproductive health, measured by
maternal mortality ratio and adolescent birth rates; empowerment, measured with proportions of
parliamentary seats occupied by females and males and proportions of females and males aged
25 years and older with at least some secondary education; and economic status, expressed as
labor market participation and calculated from labor force participation rates of female and male
populations aged 15 years and older (UNDP 2018). Although the UNDP introduced the
Genderrelated Development Index (GDI) and Gender Empowerment Measure (GEM) in 1995,
both measures had faced a lot of criticism because those measures largely depend on country’s
income level calculated from per capital GDP. As a result, in order to replace the GDI and GEM,
the UNDP has created the GII in 2010, which reflects gender-based disadvantage in three
dimensions—reproductive health, empowerment, and the labor market—for as many countries as
data of reasonable quality allow. The GII ranges from 0, where women and men fare equally, to
1, where one gender fares as poorly as possible in all measured dimensions (UNDP 2018),
indicating that the higher the GII value the more disparities between females and males and the
more loss to human development. The second measure of country-level gender equality is
aggregated egalitarian gender ideology, which is constructed by aggregating individuals’ gender
ideology for each country. To capture the overall atmosphere toward gender ideology, this
aggregated measure was calculated before sampling restrictions as well as listwise deletion so
that this measure included all the responses available in a given country.
Although the GII would be one of the most relevant measures reflecting country-level
gender inequality nowadays, given its definition and methodology, the GII already encompasses
country’s labor force participation of women. Thus, when I include the GII and female labor
force participation together in the analysis, I am not free from multicollinearity. Given this,
depending on the analytical strategy and hypothesis, I use the aggregated measure of egalitarian
gender ideology as an indicator of country-level gender equality, instead of the GII, along with
female labor force participation. Specifically, both aggregate egalitarian gender ideology and
female labor force participation are used in Chapter 4.1, because I expect that those two country
level measures may play a different role in explaining the association between preferences on
work-family arrangement and the division of housework. On the other hand, I focus more on the
effect of country-level gender equality as a key national context on individuals’ work-family
conflict (in Chapter 4.2) and on the association between work-family conflict and family life
satisfaction (in Chapter 4.3), given a lack of comparable cross-national research. Thus, I include
only the GII in Chapters 4.2 and 4.3.
Female labor force participation. This variable reflects the percentage of females among
the economically active population in a given country. The 2012 female labor force participation
data were drawn from the World Bank (2017).
GDP. Consistent with previous housework research (Fuwa 2004; Geist and Cohen 2011;
Kornrich and Eger 2015; Hu and Yucel 2018), I include GDP per capita as an indicator of
country-level economic development. The 2012 country’s gross domestic product (GDP) per
capita was drawn from UNDP data (UNDP 2014) and expressed in thousands of U.S. dollars.
Welfare regimes. Building on the two main classical categorizations of welfare regimes
(Blossfeld and Hakim 1997; Esping-Anderson 1999), I classified countries into five regimes with
four dummy variables—familialist (Austria, Belgium, France, Germany, Israel, Japan, Korea, the
Netherlands, Portugal, Spain, and Switzerland), liberal (Australia, Canada, Iceland, Ireland,
Great Britain, and the United States), former socialist (Bulgaria, Croatia, the Czech Republic,
Hungary, Latvia, Lithuania, Poland, Russia, Slovakia, and Slovenia), social democratic
(Denmark, Finland, Norway, and Sweden), and developing countries (Argentina, Chile, Mexico,
Philippines, South Africa, and Venezuela). The reference category is familialist.
3.3. Analytical Strategy
There is no weight usable for international comparison on the ISSP across countries
(GESIS 2016). Even though the 2012 ISSP data sets contain a weight variable for some
countries, “the weights in this variable do not incorporate a common weighting scheme that can
be applied to all countries of the same ISSP module (GESIS 2016, p.XV).” As there is neither an
international weighting factor nor a national weighting factor which would be comparable for all
participating countries, all descriptive and multilevel analyses are based on the unweighted data
of the ISSP 2012 module throughout this dissertation.
3.3.1. Chapter 4.1
In Chapter 4.1, I focus on the cross-national variations in the division of household labor,
in terms of preferences on work-family arrangement. Given the multilevel nature of data sets and
hypotheses, I utilize a series of multilevel regression models using the mixed command in
STATA 13. The dependent variable is respondents’ share of household labor. The main
independent variable is preferences on work-family arrangement.
Based on previous theoretical approaches to the gender division of household labor, I
include several individual-level measures. In light of time availability perspective, respondents’
and their spouse/partner’s paid work hours are included. Consistent with relative resources
perspective, I control for respondent’s relative income in the analysis. Finally, based on the
doing-gender perspective, individual-level egalitarian gender ideology is included. I also take
account of several individual-level variables that affect household labor by controlling for age,
marital status, the presence of children, education, and spouse/partner’s housework hours.
With regard to country-level variables, I include aggregated egalitarian gender ideology,
female labor force participation, GDP, and welfare regimes. All individual- and country-level
variables are centered to their grand means, so that the intercept of each model can be interpreted
as the predicted probability of outcome measure for a respondent with average characteristics in
a country with average characteristics in the sample.
The analyses in Chapter 4.1 proceed as follows. I begin by summarizing descriptive
statistics of individual- and country-level variables. Then, I conduct multilevel regression models
predicting women’s and men’s share of household labor. I estimate the models separately for
women and men because the division of household labor is the highly gendered process. I first
estimate an unconditional random intercept model containing no independent- and country-level
variables in order to find the degree of variation in women’s share of housework across
countries. Next, I add individual- and country-level variables separately to the unconditional
model. Finally, I estimate the full model including variables at both levels. By comparing the
results for women and men, I examine whether and how the effects of individual- and
countrylevel variables on respondents’ share of housework differ by gender.
3.3.2. Chapter 4.2
In Chapter 4.2, I address the cross-national variations in work-family conflict, in terms of
the division of household labor and perceived fairness of that division. Building upon the prior
comparative studies of work-family conflict (Edlund 2007; Ruppanner 2010; Stier et al. 2012), I
further restrict the samples to men and women who currently participate in paid work and who
responded to the statements of work-family conflict. The dependent variable is work-family
conflict, which is created based on respondents’ answers to four statements measuring the
degrees of incompatibility between work and family. As discussed earlier, the preliminary results
show that all four statements retain one factor with a high level of reliability (alpha = .74).
Therefore, I use the single factor structure of work-family conflict. The main predictor variable is
respondents’ share of household labor.
A relative measure of time spent on paid work is included since this is the key
characteristic of work domain, which makes it difficult to fulfill the family demands, in terms of
work-family conflict. Moreover, I consider perceptions of fairness in the analyses, consistent
with prior housework research showing the important role of perceived fairness of the division of
housework in family life (Davis 2010; Greenstein 2009; Hu and Yucel 2018). Given the fact that
gender role ideology is known to play a significant role in explaining work-family conflict
(Minnotte et al. 2010; Somech and Drach-Zahavy 2007), I also include an indicator of
individual-level egalitarian gender ideology. Finally, I take account of several individual-level
measures by controlling for age, marital status, education, relative income, and the presence of
children. For country-level variables, the Gender Inequality Index (GII), GDP, and welfare
regimes are included. All individual- and country-level variables are grand mean-centered.
The analyses in Chapter 4.2 proceed as follows. After summarizing the main variables of
interest, I conduct a series of multilevel regression models to predict the respondent’s
workfamily conflict for men and women separately, using the mixed command in STATA 13. I
first estimate an unconditional model containing no independent- and country-level variables to
find the degree of variation in work-family conflict across countries. Then, I add level-1 and
level-2 variables in that order. By comparing the gender separate models, I explore whether and
how the direction or magnitude of the effects of level-1 and level-2 variables vary by gender.
3.3.3. Chapter 4.3
In Chapter 4.3, I investigate relationship between the division of housework, work-family
conflict, and women’s and men’s family life satisfaction, focusing on the possible moderating
effect of country-level gender equality. The dependent variable is women’s and men’s
satisfaction with family life. Based on my hypotheses, three key individual-level variables are
included: a respondent’s share of household labor, perceived fairness of the housework division,
and work-family conflict. I also control for a set of individual variables including age, marital
status, education, relative income, respondent’s share of paid work, presence of children, and
egalitarian gender ideology. With respect to country-level variables, the Gender Inequality Index
(GII), GDP, and welfare regimes are included. Similar to Chapter 4.2, I restrict the samples to
men and women who currently participate in paid work and report all of key variables. All
individual- and country-level variables are grand mean-centered.
I utilize a multilevel model using the mixed command in STATA 13. First, I describe the
patterns of women’s and men’s family life satisfaction and other key individual- and countrylevel
variables across countries. Next, I proceed the multilevel analyses for women and men in order
to examine how work-family conflict is associated with women’s and men’s satisfaction, while
considering the influences of the division of housework and perceptions of fairness on that
association. I expect that the possible negative association between work-family conflict and
family life satisfaction may be stronger in countries with a higher level of gender equality. To
demonstrate this hypothesis, I add the cross-level interaction between country-level gender
inequality (i.e., GII) with work-family conflict. Finally, by comparing the cross-level interaction
effects for women and men, I examine whether the moderating role of country-level gender
inequality in the association between work-family conflict and family life satisfaction differs by
gender.
Chapter 4. Results
Based upon previous literature and hypotheses, I conduct three sets of multilevel analyses
in the current chapter, each focusing on the division of household labor, work-family conflict,
and family life satisfaction. The first subchapter examines cross-national variations in the
division of household labor, with a particular emphasis on the role of individuals’ preferences on
work-family arrangement (Chapter 4.1). The second subchapter explores the extent to which and
how the unequal division of household labor and perceived unfairness of that division are
associated with individual-level work-family conflict from a comparative perspective (Chapter
4.2). The final subchapter investigates the relationship between the division of housework,
workfamily conflict, and family life satisfaction across countries (Chapter 4.3).
4.1. Exploring the Relationship between Preferences on Work-Family Arrangement and
the Division of Household Labor
4.1.1. Descriptive Analysis
Individual-Level Measures
I begin with a descriptive overview of the dependent and individual-level variables
(Table 1). Building on previous research on the division of household labor (Greenstein 2009;
Ruppanner 2010b), the sample is restricted to respondents who are 18 or older and report data on
key variables. The original sample size of respondents who are 18 or older are 52,301. I
eliminated 15,760 individuals who do not live with a spouse or partner and 5,750 individuals
who do not have valid values of the dependent variable, resulting in 30,791 cases. After listwise
deletion, the effective sample size is 22,309 (10,874 for women; 11,435 for men). I lost 8,482
cases (27.5%), and the key sources of missing data are relative income, preferences of
workfamily arrangement, and spouses’ paid work hours (7,161 cases total – 23.3%).
The dependent variable is a respondent’s share of household labor. Consistent with
previous literature on the gender division of household labor (Bianchi et al. 2000; Coltrane
2000), the results show that there is a substantial gender difference in the proportion of
housework in our analytic sample. On average, women report their share of housework as
67.45%, while men report their share of housework as 32.88%, indicating that women report
spending far more time than their male partner on household labor. Although it was reported that
men tend to overestimate their own contribution to housework (Kamo 2000), there seems to be
no substantial discrepancy between women’s and men’s self-reported contribution to household
labor for the entire sample (i.e., 67.45% for women’s self-reported housework proportion;
67.12% for female partner’s housework proportion reported by men). In terms of preferences on
work-family arrangement, women prefer male full-time/female part-time model to the greatest
extent (43.06%), followed by male breadwinner/female homemaker (32.03%), both part-time
workers (12.88%), and both full-time workers model (12.03%). For men, male
breadwinner/female homemaker model (40.01%) is most preferred as the way to organize their
family and work life, followed by male full-time/female part-time (37.12%), both full-time
workers (13.17%), and both part-time workers model (9.70%).
Table 1. Individual-Level Descriptive Statistics (N=22,309)
Variable
Mean
or %
Women
(N=10,874)
SD
Min
Max
Men (N=11,435)
Mean
or %
SD
Min
Max
Respondent’s share of household labor (%)
67.45
17.35
0
100
32.88
19.43
0
100
Preferences on work-family arrangement (%) Male
breadwinner/female homemaker
32.03
40.01
Male full-time/female part-time workers
43.06
37.12
Both full-time workers
12.03
13.17
Both part-time workers
12.88
9.70
Respondent’s paid work hours
22.45
20.57
0
96
31.40
23.29
0
96
Spouse’s paid work hours
31.09
23.18
0
96
19.75
20.75
0
96
Relative income (high = having higher income)
3.35
1.52
1
7
5.00
1.36
1
7
Egalitarian ideology (high = more egalitarian)
3.48
.78
1
5
3.24
.75
1
5
Age (in years)
46.49
14.25
18
92
51.31
15.02
18
95
Married (%)
87.17
88.92
Presence of children (%)
No child
51.31
57.52
One child
19.31
17.36
Two or more children
29.38
25.12
Education (in years, standardized by country)
.10
.97
-4.42
6.04
.05
.99
-4.42
5.44
Spouse’s housework hours
9.77
10.75
0
95
21.43
16.04
0
95
Source: ISSP (2016)
58
Gender differences are also found in the work-related measures—a respondent’s and
spouse’s paid work hours and relative income. On average, women report that they spend 22.45
hours per week on paid work, while their male partner spend 31.09 hours. Men also report that
they do more paid work compared to their female partner (31.40 hours for men, 19.75 hours for
women). The average score of relative income is 3.35 for women and 5.00 for men, indicating
that on average men report having a higher income than their female partner, while women
report having a lower income compared to their male partner (where 4=“My partner and I have
about the same income”). Moreover, women appear to have more egalitarian gender ideology
than men (3.48 for women, 3.24 for men), which is consistent with prior findings (Davis and
Greenstein 2009). To sum up, these results indicate that gendered practices still persist in our
analytic sample, in terms of the division of labor.
The average age of women and men is about 46 and 51 years, respectively. As for the
children variable, women tend to have more children than men. This may be attributable to the
fact that women are more likely to get custody of their children when it comes to separation or
divorce. Spouse’s housework hours are controlled given the prior literature showing the negative
association between respondent’s and spouse’s contribution to household labor (Coltrane 2000;
Greenstein 2009; Ruppanner 2010a). Women report that their spouse spend only 9.77 hours a
week on household labor, while men report their spouse’s weekly housework hours as much as
21.43 hours. As for education and marital status, no significant gender differences are found. The
majority of respondents are married or in civic relationship (87.17% for women, 88.92% for
men).
59
Country-Level Measures
Table 2 presents the mean values of women’s and men’s self-reported housework
proportions and country-level variables for each country. Among 38 countries, Japanese women
report performing the largest share of housework (81.13%), whereas Venezuelan women perform
the least (58.31%). As for men’s self-reported housework proportions, Swedish men report the
highest proportion of domestic labor (40.99%) and Japanese men the lowest (16.60%).
In terms of gender role ideology, Danish people hold the most egalitarian gender role ideology
(4.1), while people living in the Philippines have the most traditional gender ideology (2.8).
Iceland has the highest female labor force participation rate (70.6%), while South Africa has the
lowest (44.2%). The country with the highest GDP is Norway (101.6 in thousands of U.S.
dollars), and the Philippines has the lowest (2.6 in thousands of U.S. dollars). Overall, these
results show that there are substantial cross-national variations in the divisions of housework as
well as national characteristics, although the majority of household labor is still performed by
women in all countries.
60
Table 2. Country-Level Descriptive Statistics (N=38)
Mean
Mean Mean GDP
Difference Egalitarian
Housework Housework per capita Welfare
Country N Between Gender FLFP
Proportion for Proportion for (US$/ Regime
Women and Ideology
Women Men 1000)
Men
Argentina
371
66.16
31.10
35.05 ***
2.97
47.30
13.04
Developing
Australia
656
63.55
35.96
27.59 ***
3.33
58.80
67.65
Liberal
Austria
497
70.06
30.09
39.97 ***
3.22
54.60
48.33
Familialist
Belgium
986
67.70
35.21
32.49 ***
3.47
46.90
44.74
Familialist
Bulgaria
401
68.16
28.82
39.34 ***
3.08
47.80
7.38
Former socialist
Canada
413
66.44
38.88
27.56 ***
3.49
61.60
52.50
Liberal
Chile
463
72.43
25.15
47.29 ***
2.92
49.00
15.25
Developing
Croatia
534
76.44
30.32
46.12 ***
3.44
44.80
13.24
Former socialist
Czech Republic
893
68.83
32.62
36.20 ***
3.29
50.10
19.73
Former socialist
Denmark
784
62.39
40.54
21.85 ***
4.06
59.10
58.13
Social Democratic
Finland
597
66.75
39.60
27.15 ***
3.73
56.00
47.42
Social Democratic
France
850
69.59
35.43
34.16 ***
3.66
50.90
40.84
Familialist
Germany
864
70.35
31.32
39.02 ***
3.72
53.50
44.07
Familialist
Hungary
434
70.21
30.81
39.40 ***
2.98
44.70
12.83
Former socialist
Iceland
556
63.30
35.98
27.32 ***
3.76
70.60
44.26
Liberal
Ireland
514
70.66
38.76
31.90 ***
3.47
52.70
49.23
Liberal
Israel
500
67.98
30.56
37.42 ***
3.31
58.10
32.57
Familialist
Japan
440
81.13
16.60
64.53 ***
3.28
48.10
48.63
Familialist
South Korea
619
69.48
24.01
45.47 ***
2.86
49.90
24.45
Familialist
Latvia
449
64.69
37.46
27.22 ***
2.96
54.50
13.80
Former socialist
Lithuania
400
64.04
32.72
31.31 ***
3.09
55.80
14.34
Former socialist
Mexico
417
64.72
30.38
34.34 ***
2.84
45.00
9.72
Developing
Netherlands
694
69.16
37.02
32.14 ***
3.50
58.70
49.47
Familialist
Norway
721
67.79
40.06
27.72 ***
3.80
61.50
101.56
Social Democratic
Philippines
765
63.25
35.51
27.74 ***
2.81
51.00
2.60
Developing
(table cont’d.)
61
Mean
Mean Mean GDP
Difference Egalitarian
Housework Housework per capita Welfare
Country N Between Gender FLFP
Proportion for Proportion for (US$/ Regime
Women and Ideology
Women Men 1000)
Men
Poland 517 63.57 34.34
Portugal 290 72.04 26.45
Russia 483 63.09 33.57
Slovakia 612 65.72 35.92
Slovenia 484 68.83 24.34
South Africa 707 64.47 37.65
Spain 1129 66.38 26.70
Sweden 510 62.70 40.99
Switzerland 667 70.72 28.63
Taiwan 842 66.22 25.48
United Kingdom 395 72.13 38.84
United States 497 66.24 35.66
Venezuela 358 58.31 38.14
Source: ISSP (2016), UNDP (2014), World Bank (2018)
29.23 ***
45.59 ***
29.52 ***
29.80 ***
44.49 ***
26.81 ***
39.68 ***
21.71 ***
42.09 ***
40.74 ***
33.30 ***
30.58 ***
20.17 ***
62
3.14 3.35
2.93 3.18
3.51 3.18
3.43 3.89
3.23 3.12
3.40 3.29
3.03
48.90
55.40
57.00
51.00
52.30
44.20
52.60
60.20
61.20
50.20
55.70
56.80
50.90
13.15
20.58
15.04
17.27
22.49
7.57
28.65
57.13
83.16
21.31
41.54
51.43
12.77
Former socialist
Familialist
Former socialist
Former socialist
Former socialist
Developing
Familialist
Social
Democratic
Familialist
Familialist
Liberal
Liberal
Developing
66
Cross-National Variations in the Division of Household Labor
Table 2 also provides comparisons between women’s and men’s mean share of household
labor by country. The results show that there are considerable variations in the division of
household labor across countries, even though women perform the lion’s share of housework in
all countries. Overall, Japanese people are most traditional in their division of household labor
among the sampled countries. Japanese women report performing the largest share of housework
(81.13%), while Japanese men report performing the smallest share of housework (16.60%),
which results in the largest difference in women’s and men’s self-reported housework proportion
(Figure 1). On the other hand, people in social democratic countries (i.e., Denmark, Finland,
Norway, and Sweden) as well as in Venezuela seem to be relatively egalitarian in their division of
housework. In terms of men’s self-reported housework proportions, Swedish men report
performing the largest share of housework (40.99%), followed by Danish (40.54%), Norwegian
(40.06%), and Finnish men (39.60%). Women in those four countries also report performing the
relatively small share of housework compared to women in other countries. As a result, there are
relatively small gender gaps in a respondent’s share of housework in social democratic countries.
Meanwhile, Venezuelan women report performing the smallest share of domestic labor
(58.31%). The gender gap in a respondent’s share of housework in Venezuela also appears to be
the least (20.17%) among 38 countries. This indicates that Venezuelan are relatively more
egalitarian in their division of household labor, along with people in social democratic countries.
21.7
1
09
67
21.8
5
26.81
27.15
27.22
27.32
27.56
27.59
27.72
27.74
29.23
29.52
29.80
30.58
31.3
31.9
32.
32.
33.
34
34
1
0
14
49
30
.16
.34
5.05
6.20
37.42
39.02
39.34
39.40
39.68
39.97
40.74
42.
44
.49
5.47
5.59
6.12
47.29
68
3
3
4
4
69
Venezuela 20.17
Sweden
Denmark
South Africa
Finland
Latvia
Iceland
Canada
Australia
Norway
Philippines
Poland
Russia
Slovakia
United States
Lithuania
Ireland
Netherlands
Belgium
United Kingdom
France
Mexico
Argentina
Czech Republic
Israel
Germany
Bulgaria
Hungary
Spain
Austria
Taiwan
Switzerland
Slovenia
Korea
Portugal
Croatia
Chile
Japan 64.53
0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00
Figure 1. Differences between Women’s and Men’s Share of Household Labor by Country
4
70
4.1.2. Multilevel Analysis
Tables 3 and 4 provide the results of multilevel regression models predicting women’s and
men’s share of household labor, respectively. I estimate gender separate models because the
division of household labor is a highly gendered process (Ruppanner 2010a). Each set of models
introduces level-1 and level-2 variables. The full models for women and men are provided in
Model 4 of each Table.
Models 1-4 of Table 3 show the results for women’s share of household labor. First of all,
I conduct an unconditional model containing no independent- and country-level variables in order
to find the degree of variation in women’s share of housework across countries (Model 1 of Table
3). Note that all of the continuous variables at both levels are centered on their grand means.
Thus, the intercept in each model indicates the predicted share of housework for a woman (or a
man) with average characteristics in a country with average characteristics in the sample. The
intercept of Model 1 is 67.512 (p<.001), indicating that on average women perform about two
third of housework across countries. The between-country variance on the intercept is 15.9
(p<.001) and the within-country variance is 287.2 (p<.001). The intraclass correlation is .053,
indicating 5.3% of the overall variance in women’s share of housework is attributable to between-
country variance. These results show that women’s share of housework varies significantly both
between and within countries, although the more variation occurred at the individual-level.
Table 3. Multilevel Analysis of Women’s Share of Household Labor (N=10,874)
Model 1 Model 2 Model 3 Model 4
Intercept 67.512 *** 60.127 *** 69.11 *** 59.595 ***
Level 1: Individual-level variables Preferences
on work-family arrangement
(ref. = male breadwinner/ female homemaker)
-.663*
-.599
71
Male full-time/female part-time workers
Both full-time workers
Both part-time workers Time
availability
Respondent’s paid work hours
Spouse’s paid work hours
Relative resources
Relative income (high = higher income)
Doing-gender
Egalitarian ideology (high = egalitarian)
Individual controls
Age
Age squared
Married (ref. = not married)
Presence of children (ref. = no child)
One child
Two or more children
Education (standardized by country)
Spouse’s housework hours
Level 2: Country-level variables
Egalitarian ideology (high = egalitarian)
Female labor force participation
GDP per capita (US$/1000)
Welfare regimes (ref. = familialist)
Liberal
Former socialist
Social democratic
Developing
-2.097***
-2.587***
-.125***
.091***
-.719***
-1.787 ***
.418 ***
-.003 ***
.804
1.666 ***
2.173 ***
-.781***
-.919 ***
1.587
-2.054 ***
-2.465 ***
-.125 ***
.092 ***
-.723 ***
-1.761 ***
.413 ***
-.003 ***
.784
1.672 ***
2.198 ***
-.785 ***
-.920 ***
1.317
-.234*
.058
-.657
3.272 *
-3.644 *
.558
-.458 ***
.087 *
-1.286
-1.287
-5.408 *
-4.273
Variance components
Between-country 15.9*** 8.2*** 8.5 ***
4.7 ***
Within-country 287.2 *** 179 *** 287.2 *** 179.0***
ICC (%) 5.3% 4.4% 2.9% 2.5%
BIC 92541.3 87525.1 92584.5 87571.4
Wald chi-square (df) - 6582.2 (14) 29.4 (7) 6629.0 (21)
Log-likelihood -46256.7 -43683.6 -46245.8 -43674.2
Note: *p < .05, **p < .01, ***p < .001. Total number of countries for the analysis is 38.
Source: ISSP (2016), UNDP (2014), World Bank (2018)
72
Models 2 adds individual-level variables to Model 1. In Model 2, the between-country
(i.e., level-2) variance decreases by 48.3% (from 15.9 to 8.2) and the within-country (i.e., level1)
variation by 37.7% (from 287.2 to 179.0). In contrast, Model 3 include only country-level
variables. In Model 3, the level-2 variance reduces by 46.7% (from 15.9 to 8.5), but no change is
found in the level-1 variance. Finally, Model 4 is the full model, including all of the level-1 and
level-2 variables together. In Model 4, total 70.7% of reductions in the level-2 variance (from
15.9 to 4.7) and total 37.7% of reductions in the level-1 variance (from 287.2 to 179.0) are
achieved. The intralclass correlation decreases from .053 in the unconditional model to .025 in
the full model. Overall, the results indicate that there are substantial variations in women’s share
of domestic labor that can be explained by between- and within-country variations in our analytic
sample.
In the full model for women (Model 4), most level-1 and level-2 variables remain
significant and in the same direction as earlier models. Preferences on work-family arrangement
are significantly associated with women’s share of housework. Women who prefer either both
full-time workers (b=-2.054, p<.001) or both part-time workers model (b=-2.465, p<.001) tend to
perform a smaller share of housework, compared to the reference group of women who prefer a
traditional male breadwinner/female homemaker model. However, there is no statistically
significant difference in women’s share of housework between women who prefer a full-time
male/part-time female model and a reference group of women.
The results also show that individual-level measures reflecting time availability, relative
resources, and doing-gender perspectives are significantly related to women’s share of household
labor in the theoretically expected direction. Consistent with the time availability predictions,
respondent’s paid work hours are associated with a smaller share of housework (b=-.125, p<.001),
73
while spouse’s paid work hours are associated with a larger share of housework (b=.092, p<.001).
In light of relative resources perspective, earning higher income is expected to be associated with
doing less housework, which is consistent with our result. Women who have higher income
relative to their partner tend to report performing a smaller share of housework (b=-.723, p<.001).
As expected by the doing-gender perspective, women who hold more egalitarian gender ideology
tend to perform a smaller share of household labor (b=-1.761, p<.001).
With respect to other individual controls, all but marital status are significantly associated
with women’s share of household labor. The coefficient of age squared is negative and significant
(b=-.003, p<.001), indicating that the women’s share of housework reaches the highest point
when they are about 69 years old (=.413/(2*.003)=68.8), then decrease.
Considering the age range of women in our sample, the pattern of women’s share of housework
seems to be a monotonous increase. The child variable is also a significant predictor of women’s
share of household labor. Compared to women with no child, women with one child are more
likely to perform a larger share of housework (b=1.672, p<.001), and the degree of unequal
division of household labor appears to be higher for women who have two or more children
(b=2.198, p<.001). Finally, both respondent’s educational attainment (b=-.785, p<.001) and
spouse’s housework hours (b=-.920, p<.001) are negatively associated with a share of housework
for women.
At the country-level, the results show that women’s share of household labor could be
accounted for by country’s female labor force participation and welfare regimes. The coefficient
of female labor force participation is negative and statistically significant (b=-.234, p<.05),
indicating that women in countries with higher rates of female labor force participation report
performing a smaller share of housework. Further, women in former socialist countries report a
74
larger share of housework (b=3.272, p<.05), while women in social democratic countries report a
smaller share of housework (b=-3.644, p<.05), compared to a reference group of women in
familialist countries. Contrary to expectations, however, neither country-level gender ideology
nor GDP are statistically significant.
Now I turn to multilevel analyses of men’s share of household labor. Models 1-4 of Table
4 present the results for men’s share of household labor. Model 1 is an unconditional model
containing no level-1 and level-2 variables. The intercept of the unconditional model for men is
32.941 (p<.001), indicating that only about 33% of housework is done by men. The
betweencountry variance on the intercept is 29.1 (p<.001) and the within-country variance is
349.0
(p<.001). The intraclass correlation is .077, indicating 7.7% of the overall variance in men’s share
of housework is attributable to between-country variance. These results suggest that like women’s
share of housework, men’s share of housework varies significantly both between and within
countries, but the more variation occurred at the individual-level.
Table 4. Multilevel Analysis of Men’s Share of Household Labor (N=11,435)
Model 1
Model 2
Model 3
Model 4
Intercept 32.941 ***
40.009***
28.861 ***
35.865***
Level 1: Individual-level variables
Preferences on work-family arrangement
(ref. = male breadwinner/ female homemaker)
Male full-time/female part-time workers
.556
.570
Both full-time workers
1.921***
1.905***
Both part-time workers
2.463***
2.535***
Time availability
Respondent’s paid work hours
-.119***
-.121***
75
Spouse’s paid work hours
.088***
.088***
Relative resources
Relative income (high = higher income)
-1.384***
-1.373***
Doing-gender
Egalitarian ideology (high = egalitarian)
1.350***
1.375***
Individual controls
Age
-.102
-.099
Age squared
.001
.001
Married (ref. = not married)
-2.901***
-2.848***
Presence of children (ref. = no child)
One child
-.750
-.788
Two or more children
.521
.480
Education (standardized by country)
.337*
.332*
Spouse’s housework hours
-.447***
-.448***
Level 2: Country-level variables
Egalitarian ideology (high = egalitarian)
4.887
-1.737
Female labor force participation
.164
.190
GDP per capita (US$/1000)
-.005
-.006
Welfare regimes (ref. = familialist)
Liberal
6.845 **
6.012***
Former socialist
4.291 *
5.168**
Social democratic
7.892 *
4.804*
Developing
6.635 *
8.364***
Variance components
Between-country 29.1 *** 17.5*** 13.8***
7.3***
Within-country 349.0 *** 272.2*** 349.0***
272.2***
ICC (%) 7.7% 6.0% 3.8% 2.6%
BIC 99554.5 96832.2 99593.3 96867.1
Wald chi-square (df) - 3242.1 (14) 38.5 (7) 3318.7 (21)
Log-likelihood -49763.2 -48336.7 -49749.9 -48321.4
Note: *p < .05, **p < .01, ***p < .001. Total number of countries for the analysis is 38. Source:
ISSP (2016), UNDP (2014), World Bank (2018)
76
When individual-level variables are included (Model 2), the between-country (i.e., level2)
variance decreases by 39.7% (from 29.1 to 17.5) and the within-country (i.e., level-1) variation by
22.0% (from 349.0 to 272.2). When country-level variables are included (Model 3), the between-
country variance decreases by 52.6% (from 29.1 to 13.8), and almost no change is found in
within-country variance. Finally, in the full model for men (Model 4), total 74.8% of reductions
in the between-country variance (from 29.1 to 7.3) and total 22.0% of reductions in the within-
country variance (from 349.0 to 272.2) are achieved. The intraclass correlation changes from .077
in the unconditional model to .026 in the full model. Like women, these results suggest that there
are considerable variations in men’s share of housework, which can be accounted for by the
between- and within-country differences.
Consistent with expectations, preferences on work-family arrangement also appears to be
a significant factor of predicting men’ housework proportions. Men who prefer either both
fulltime workers model (b=1.905, p<.001) or both part-time workers model (b=2.535, p<.001) are
more likely to perform a larger share of housework, compared to a reference group of men who
prefer a traditional male breadwinner/female homemaker model. These results, taken together
with previous findings for women, suggest that individuals’ preferences for the equal contribution
for women’s and men’s housework play a significant role in their divisions of
household labor for both men and women.
Individual-level measures reflecting time availability, relative resources, and doinggender
perspectives are also significantly associated to men’s share of household labor.
Consistent with previous findings, the results show that men who spend less hours on paid work
(b=-.121, p<.001), who have a spouse/partner spending more hours on paid work (b=.088,
p<.001), and who have lower income relative to their spouse/partner (b=-1.373, p<.001) report
77
performing a larger share of household labor. On the other hand, men who hold more egalitarian
gender ideology (b=1.375, p<.001) tend to perform a larger share of housework, indicating that
egalitarian men are more likely to share household labor more equally with their partner, which is
consistent with predictions of doing-gender perspective.
With respect to individual controls, however, some gender differences are found. For
women, age and the presence of children are significant factors relating to the housework
distribution with their partner, but neither variable is statistically significant for men. In contrast,
marital status is not statistically significant for women, while being married is negatively
associated with men’s share of household labor. Such gender differences have important
implications for the mechanism of the gender division of household labor. First, as calculated
earlier, women’s share of housework continues to rise from the age of 18, reaches the peak at
around 70 years old, and then decreases while there is no age effect for men. Next, the result of
the positive association between the presence of children and women’s share of household labor
indicates that increased housework burden related to having children is likely to be put on
women, not on men. Finally, the negative association between being married and men’s share of
housework demonstrates that the benefits of marriage might be unequally distributed between
men and women in terms of domestic labor (Coontz 2006; Waite and Gallagher 2002),
particularly in light of the fact that marriage has historically been organized to men’s advantage
(Nock 1998).
As for country-level variables, cross-national differences in welfare regimes are only
significantly associated with men’s share of housework. All of four categories of welfare regimes
have a significant effect on men’s share of housework. Compared to a reference group of men
living in familialist countries, men who live in liberal (b=6.012, p<.001), former socialist
78
(b=5.168, p<.01), social democratic (b=4.804, p<.05), and developing (b=8.364, p<.001)
countries report performing a larger share of housework, when other individual- and countrylevel
variables are controlled. The results suggest that men living in familialist countries are most
traditional in their division of household labor.
4.1.3. Conclusion and Discussion
In Chapter 4.1, I examine the cross-national variations in the division of household labor
by focusing on the role of individuals’ preferences on work-family arrangement. In the context of
growing individualization of family and personal lives in modern societies, it is expected that
individuals’ preferences on how they want to organize their work and family life may play a role
in the actual division of household labor. However, the empirical investigation for this question
has not been provided. Using the ISSP 2012 data, I explore the relationship between individuals’
preferences on work-family arrangement and women’s and men’s share of household labor from a
comparative perspective. In light of preference theory (Hakim 2000, 2003) as well as the
expanding roles of personal choice to family life in modern societies (Cherlin 2004), I expect that
individuals’ preferences on work-family arrangement are associated with the actual division of
household labor for both women and men (Hypothesis 1). Consistent with previous comparative
research on the division of household labor, I also expect that individuals’ share of household
labor is conditional upon national contexts, which are measured by country-level differences in
egalitarian gender ideology, female labor force participation, GDP, and welfare state regimes
(Hypothesis 2). Finally, given the gendered nature of domestic work, I expect that there are
gender differences in the effects of individual- and country-level variables on the division of
household labor (Hypothesis 3).
79
The findings provide support for Hypothesis 1. Women who prefer either both full-time
workers or both part-time workers model tend to report performing a smaller share of housework,
while men who prefer those models tend to report performing a larger share of housework,
compared to their counterparts who prefer the traditional male breadwinner/female homemaker
model. In other words, individuals’ preferences for the equal contribution in terms of work-family
arrangement are significantly associated with relatively equal distribution of household labor
between partners. These results indicate that women’s and men’s preferences on work-family
arrangement play a significant role in explaining the actual division of household labor, even
though women still perform the lion’s share of housework when they prefer gender equal
arrangement of their work and family life.
With respect to the effects of country-level characteristics, the findings provide partial
support for Hypotheses 2 and 3. Female labor force participation appears to be a significant factor
influencing women’s share of household labor, but it does not show a significant result for men.
Specifically, women in countries with higher rates of female labor force participation report
performing a smaller share of household labor. These results suggest that national contexts that
allow women to actively participate in the labor market may have contributed to a decrease of
women’s share of household labor by reducing women’s housework hours, but it may not lead to
men’s greater involvement in domestic labor. Given this, the context of stalled revolution in terms
of domestic labor (Hochschild 1989) still seems to persist in our analytic sample. The findings
also show that country’s welfare regimes play a role in women’s and men’s share of housework
but have more significant effects for men. With a reference group of women who live in
familialist countries, women in social democratic countries report performing a smaller share of
housework, while women in former socialist countries report doing a larger share of housework.
80
For men, all of four categories of welfare regimes are positive and statistically significant,
indicating that on average men in familialist countries are most traditional in their division of
household labor.
Contrary to expectations, however, egalitarian gender ideology at the macro-level has no
significant effect on women’s or men’s share of household labor. There are some possible
explanations for these results. First, the division of housework in terms of personal preferences on
work-family arrangement may be more directly influenced by a structural condition—like female
labor force participation—because higher rates of women’s economic participation likely lead to
the necessity to reduce women’s housework load, and in turn, such a structural condition may
also affect individuals’ preference construction with respect to work-family arrangement. Given
this, country’s female labor force participation may have a stronger effect than aggregate
egalitarian gender ideology on the division of household labor, particularly for women who need
to directly reduce their housework burden in order to participate in paid labor. Our results
showing a significant effect of female labor force participation for women’s share of housework
but not for men’s share also support this explanation. Furthermore, the influence of personal
choice on their work and family life is likely to be stronger in modern, affluent societies where
the power of social norms and institutions that regulated personal life has weakened (Beck and
Beck-Gernsheim 1995, 2002; Giddens 1991, 1992). That is, individual-level gender ideology may
be a more influential factor explaining the division of household labor, as shown in our finding
that egalitarian gender ideology at the individual-level is strongly associated with the relatively
equal distribution of housework between partners for both women and men.
I acknowledge some limitations that need to be mentioned. As for national contexts, I
consider four country-level variables (i.e., aggregate egalitarian gender ideology, female labor
81
force participation, GDP, and welfare regimes), but I cannot exclude the possibility that there are
other national contexts, such as social policies, that affect the cross-national variations in the
division of household labor in terms of personal preferences on work-family arrangement.
Moreover, although I do not find any significant cross-level interaction effects in the current
analyses, it is possible that interactions between level-1 and level-2 variables can be found when
other national contexts are included. Future research could expand on the findings of this study
by elaborating the effects of various national contexts on the division of household labor, while
considering the role of personal preferences on work-family arrangement. Finally, it is important
to note that the current multilevel analyses cannot establish causality. Therefore, we need to be
careful when interpreting the results of this study. Nonetheless, our findings provide robust
evidence of the association between personal preferences on work-family arrangement and their
actual division of household labor across countries.
Overall, this study sheds new light on the role of personal preferences on work-family
arrangement in explaining the cross-national variations in the division of household labor. Despite
the increasing number of cross-national research on housework, empirical investigations of the
role of personal preferences on work-family arrangement have received insufficient attention,
partly because of the lack of available dataset. Using the most recent ISSP module on family and
changing gender roles, I seek to address this void. As a result, this study contributes to
comparative housework research by providing empirical evidence of the role of personal
preferences in terms of work-family arrangement as a significant factor influencing the division
of household labor in contemporary societies.
82
4.2. Exploring the Relationship between the Division of Household Labor and WorkFamily
Conflict
4.2.1. Descriptive Analysis
Individual-Level Measures
The descriptive statistics for the dependent and individual-level variables are presented in
Table 5. The sample includes respondents who are 18 or older, are living with a spouse or partner,
and report currently working in paid labor. The original sample size of respondents who are 18 or
older are 52,301. I eliminated 15,760 individuals who do not live with a spouse or partner and
14,288 individuals who do not currently employed in the labor market, resulting in 22,253 cases.
I have 21,461 cases with valid values of the dependent variable. After listwise deletion, the
effective sample size is 15,322 (6,904 for women, 8,418 for men). I lost 6,139 cases (28.6%), and
the key sources of missing data are relative income, perceived fairness of the division of
housework, and the respondent’s perceived share of household labor (4,601 cases total – 21.4%).
The dependent variable of this chapter is work-family conflict. The mean score of
workfamily conflict is 1.93 (SD=.71) for women and 1.83 (SD=.70) for men. Note that the higher
values of work-family conflict indicate that respondents experience work-family conflict more
often. The results show that on average women report more work-family conflict than men, which
is consistent with existing evidence of women’s disadvantage of work-family conflict.
Do roughly my fair share
Do less than my fair share
Age (in years)
Married (%)
Education (in years, standardized by country)
Relative income (high = having higher income)
Respondent’s share of paid work
Presence of children (%)
No child
One child
Two or more children
Egalitarian ideology (high = more egalitarian)
Source: ISSP (2016)
40.32
5.59
42.90
85.66
.33
3.73
54.46
46.47
21.70
31.84
3.62
10.82
.91
1.35
22.43
.74
18
-3.90
1
2.33
1
93
6.04
7
100
5
45.49
86.72
.21
5.18
69.68
46.14
21.39
32.47
3.35
47.33
41.91
11.65
.93
1.20
22.67
.75
18
-4.42
1
6.25
1
90
5.44
7
100
5
78
SD SD
Work-family conflict (high = more conflict) .71 1 4 .70 1 4
Table
5
. Individual
-
=15,322)
Level Descriptive Statistics (N
Female (N=6,904)
Male (N=8,418)
Variable
Mean
or %
Min
Max
Mean
or %
Min
Max
1.93
1.83
Respondent
’
s share of household labor
65.56
16.76
0
100
32.46
18.49
0
100
Perceived fairness (%)
Do more than my fair share
54.08
10.76
86
As expected, considerable gender differences are found in housework-related measures.
On average, women report that they perform 65.56% of household labor relative to their partner,
while men report only 32.46% of household labor are done by themselves. There are also
substantial differences in men’s and women’s reports on the perceived fairness of the division of
household labor. More than half of female respondents (54.08%) respond that they do more than
their fair share of the household labor, while only 10.76% of male respondents report that they do
more than their fair share. In contrast, only 5.59% of women report that they do less than their
fair share of the housework, whereas about 42% of men report they do less than their fair share.
As for the perception of doing one’s fair share of the housework, a gender gap is quite
attenuated, but more men perceive that they do roughly their fair share of the household labor
compared to women (40.32% for women; 47.33% for men). Gender differences are also found in
the work-related measures—a respondent’s share of paid work and relative income. On average,
men report their share of paid work as about 70%, while women report about 54% as their share
of paid work. The average score of relative income is 3.73 for women and 5.18 for men,
indicating that on average men report having a higher income than their female partner, while
women report having a lower income compared to their male partner (where 4=“My partner and
I have about the same income”).
With respect to age, marital status, educational attainment, and number of children, no
significant gender differences are found. The average ages of women and men are about 43 and
45 years, respectively. The majority of respondents are married or in civic relationship (85.66%
for women; 86.72% for men) and have no children (46.47% for women; 46.14% for men). After
listwise deletion of missing cases, the standardized measure of education is approximately
87
normally distributed for both women and men, with a mean of .33 for women and .21 for men, as
well as a standard deviation of .91 for women and .93 for men. Finally, women hold more
egalitarian ideology than men (3.62 for women; 3.35 for men).
Country Level Measures
Table 6 presents the mean values of women’s and men’s self-reported work-family
conflict and country-level variables for each country. Among 38 countries, Taiwanese women
(1.59) and men (1.42) report the lowest work-family conflict, while Venezuelan women (2.71)
and men (3.20) report the highest work-family conflict. In terms of the Gender Inequality Index
(GII), Denmark has the lowest score (.045) and Venezuela the highest (3.02). As for GDP and
welfare regimes for each country, please refer to Table 2. The descriptive statistics show a wide
range in the dependent variable and country-level measures.
To find if there are systematic gender differences in work-family conflict, I then conduct a
series of t-tests for each country. Overall, women report more work-family conflict than men in
the most countries (i.e., 32 out of 38 countries), and about half of the results (i.e., 15 countries)
are statistically significant. Among 32 countries where women report more work-family conflict
than men, Korean appears to have the highest gender disparities in work-family conflict
(difference=.38; p<.001). Among 6 countries in where men report more work-family conflict than
women, Venezuelan report the highest (difference=-.49; p<.001).
To sum up, the results indicate that women experience more work-family conflict in general than
men, although such a gender difference in work-family conflict is less systematic and less
significant than a gender difference in the division of household labor in the previous chapter. In
88
the next chapter, I conduct a series of multilevel regression models predicting women’s and men’s
work-family conflict, by focusing on the effects of the division of household labor,
perceptions of fairness of that division, and country-level gender equality.
Table 6. Country-Level Descriptive Statistics (N=38)
Mean Mean Mean
Gender
Work-Family Work-Family Difference
Country N Inequality
Conflict Conflict Between
Index
for Women for Men Women and Men
Argentina
243
1.91
1.61
.30 **
.37
Australia Austria
557
349
1.98
1.76
1.83
1.74
.15 **
.02
.13
.10
Belgium
Bulgaria
677
238
1.93
2.27
1.75
2.26
.19 ***
.01
.07
.22
Canada
291
1.94
1.72
.21 **
.11
Chile
243
2.47
2.12
.35 **
.35
Croatia
Czech Republic
Denmark
Finland
France
Germany
Hungary
Iceland
290
608
616
466
700
604
253
465
2.30
1.70
1.78
1.77
1.94
1.78
1.88
1.84
2.06
1.67
1.72
1.69
1.89
1.84
1.82
1.80
.24 *
.04
.05
.08
.05
-.07
.06
.04
.13
.13
.05
.07
.11
.09
.26
.08
Ireland
392
1.91
1.72
.19 ***
.14
Israel
Japan
380
309
2.27
1.70
1.93
1.77
.34 ***
-.07
.13
.13
South Korea
408
2.13
1.75
.38 ***
.08
Latvia
Lithuania
Mexico
Netherlands
329
269
304
435
2.08
1.86
2.12
1.66
1.86
1.90
2.03
1.59
.22 **
-.04
.09
.07
.19
.14
.37
.05
89
Norway
Philippines
Poland
Portugal
Russia
Slovakia
611
436
313
195
285
361
1.82
2.06
2.13
1.87
2.17
2.08
1.70
2.09
2.02
1.73
2.10
2.21
.12 **
-.03
.11
.14
.08
-.14
.06
.45
.15
.12
.29
.19
Slovenia
South Africa
279
315
1.87
1.82
1.68 1.82
.19 *
.00
.07
.41
Spain
698
2.19
1.91
.28 ***
.10
Sweden
Switzerland
404
469
1.83
1.61
1.65 1.58
.18 ***
.03
.05
.05
Taiwan
United Kingdom
United States
622
283
405
1.59
1.92
1.98
1.42
1.81
1.87
.17 **
.11
.11
.05
.16
.23
Venezuela
220
2.71
3.20
-.49 ***
.47
Source: ISSP (2016), UNDP (2014), Taiwan National Statistics (2018)
Note: *p < .05, **p < .01, ***p < .001.
4.2.2. Multilevel Analysis
Tables 7 and 8 provide results of multilevel regression models predicting women’s and
men’s work-family conflict, respectively. Each set of models introduces level-1 and level-2
variables. The full models for women and men are provided in Model 4 of each Table.
Models 1-4 of Table 7 show the results for women’s work-family conflict. First of all, I
conduct an unconditional model containing no individual- and country-level variables in order to
find the degree of variation in women’s work-family conflict across countries (Model 1 of Table
7). Again, all of continuous variables at both levels are centered on their grand means. Thus, the
intercept in each model indicates predicted work-family conflict for a woman (or a man) with
average characteristics in a country with average characteristics in the sample. The intercept of
Model 1 of Table 7 is 1.961 (p<.001), indicating women’s average self-reported work-family
conflict across countries. The between-country variance on the intercept is .050 (p<.001) and the
within-country variance is .457 (p<.001). The intraclass correlation is .099, indicating 9.9% of the
90
overall variance in women’s work-family conflict is attributable to between-country variance.
These results indicate that women’s work-family conflict varies significantly both between and
within countries, while the more variation occurred at the individual-level.
Table 7. Multilevel Analysis of Women’s Work-Family Conflict (N=6,904)
Model 1
Model 2
Model 3
Model 4
Intercept
1.961 ***
1.899 ***
1.993 ***
1.928 ***
Level 1: Individual-level variables
Respondent’s share of household labor
-.002 ***
-.002 ***
Respondent’s share of paid work
.001 **
.001 **
Perceived fairness
(ref. = do roughly my fair share) Do
more than my fair share
.151 ***
.150 ***
Do less than my fair share
.133 ***
.133 ***
Age
.017 **
.017 **
Age squared
.000 ***
.000 ***
Marital status (ref. = not married)
-.013
-.012
Education
.039 ***
.038 ***
Relative income
.021 **
.021 **
Presence of children (ref. = no child)
One child
.068 **
.067 **
Two or more children
.071 **
.070 **
Egalitarian ideology
-.127 ***
-.124 ***
Level 2: Country-level variables
Gender inequality index (GII)
1.340
.969
GDP per capita (US$/1000)
-.003
-.002
Welfare regimes (ref. = familialist)
Liberal
.013
.013
Former socialist
-.020
-.032
Social democratic
.041
.085
Developing
-.192
-.197
Variance components
Between-country
.05 ***
.037 ***
.031 ***
.031 ***
91
Within-country
.457 ***
.438 ***
.457 ***
.438 ***
ICC (%)
9.9%
7.8%
6.3%
6.5%
BIC
14327
14122.2
14363.1
14168.5
Wald chi-square (df)
-
319.2 (12)
21.3 (6)
328.9 (18)
Log-likelihood
-7150.2
-6994.8
-7141.8
-6991.4
Note: *p < .05, **p < .01, ***p < .001. Total number of countries for the analysis is 38.
Source: ISSP (2016), UNDP (2014), World Bank (2018), Taiwan National Statistics (2018) Model
2 of Table 7 adds individual-level variables to Model 1, while Model 3 of Table 7 includes
country-level variables only. Overall, the results indicate that individual-level variables are
significantly associated with women’s work-family conflict, whereas none of country-level
variable is statistically significant. Both measures of paid and unpaid labor are significantly
associated with work-family conflict. Women who perform a small share of housework (b=-.002,
p<.001) or who perform a large share of paid work (b=.001, p<.01) report more work-family
conflict. Note that the direction of women’s share of housework is somewhat opposite. The
coefficient is negative and significant, which indicates that women who do a larger share of
household labor relative to their partner report less work-family conflict than their female
counterparts who do a smaller share of housework, after controlling for other individual-level
variables. It might be because that women who do a higher share of housework are less sensitive
to work-family conflict caused from the unequal division of household labor. Fairness perception
of the division of household labor is also significantly related to women’s work-family conflict.
Women who perceive doing more or less than their fair share of housework report more
workfamily conflict than the reference group who perceives doing their fair share of household
labor
(b=.151, p<.001 for doing more than one’s fair share; b=.133, p<.001 for doing less than one’s
fair share). Being older (b=.017, p<.01), having a higher educational attainment (b=.039, p<.001),
92
earning higher income than one’s partner (b=.021, p<.01) and having more children (b=.068,
p<.01 for one child; b=.071, p<.01 for more than two children) are all associated with a higher
level of work-family conflict for women. Egalitarian gender ideology is negatively associated
with women’s work-family conflict, indicating that women who hold more egalitarian gender
ideology tend to report less work-family conflict (b=-.127, p<.001) than their traditional
counterparts.
Model 4 of Table 7 is the full model including both individual- and country-level
variables. The final model is consistent with previous models (Models 2 and 3). Including all of
variables at both levels reduces the between-country variance by 39.0% (from .050 to .031) and
the within-country variance by 4.3% (from .457 to .438). The intralclass correlation changes .099
in the unconditional model to .065 in the full model. Overall, the results indicate that both level-1
and level-2 variables included in the model play an important role in explaining women’s work-
family conflict in our analytic sample.
Models 1-4 of Table 8 present the results for men’s work-family conflict. Each set of
models introduces level-1 and level-2 variables. Compared to the results for women’s workfamily
conflict, several findings should be noted. First, the effect of respondents’ share of housework on
work-family conflict significantly varies by gender. While Table 7 shows that respondent’s share
of housework plays a significant role in explaining women’s work-family conflict, this variable is
not statistically significant for men (Table 8). This suggests that women are more influenced by
household labor in terms of work-family conflict than men. More notable gender differences are
found in the association between the GII and work-family conflict. Unlike women, the GII is
significantly related to men’s work-family conflict (b=2.412, p<.01 in Model 4 of Table 8),
93
indicating that men who live in high GII countries (i.e., countries with a lower level of gender
equality) tend to report more work-family conflict compared to their male counterparts who live
in low GII countries (i.e., countries with a high level of gender equality). These results suggest
that men are particularly influenced by country-level gender equality in
terms of work-family conflict, compared to women.
Table 8. Multilevel Analysis of Men’s Work-Family Conflict (N=8,418)
Model 1
Model 2
Model 3
Model 4
Intercept
1.864 ***
1.739 ***
1.949***
1.807 ***
Level 1: Individual-level variables
Respondent’s share of household labor
.000
.000
Respondent’s share of paid work
.001 *
.001 *
Perceived fairness
(ref. = do roughly my fair share) Do
more than my fair share
.140 ***
.140 ***
Do less than my fair share
.109 ***
.109 ***
Age
.006
.007
Age squared
.000 **
.000 **
Marital status (ref. = not married)
.022
.022
Education
.008
.008
Relative income
-.002
-.002
Presence of children (ref. = no child)
One child
.062 **
.062 **
Two or more children
.080 ***
.080 ***
Egalitarian ideology
-.062 ***
-.059 ***
Level 2: Country-level variables
Gender inequality index (GII)
2.585***
2.412 **
GDP per capita (US$/1000)
.000
.001
Welfare regimes (ref. = familialist)
Liberal
-.092
-.070
Former socialist
-.001
-.001
Social democratic
.026
.068
Developing
-.396
-.403
94
Variance components
Between-country
.077 ***
.067 ***
.041***
.041 ***
Within-country
.426 ***
.411 ***
.426***
.411 ***
ICC (%)
15.4%
14.1%
8.8%
9.1%
BIC
16874.6
16679.5
16905.7
16716.0
Wald chi-square (df)
309.3 (12)
31.8 (6)
335.1 (18)
Log-likelihood
-8423.7
-8271.9
-8412.2
-8263.1
Note: *p < .05, **p < .01, ***p < .001. Total number of countries for the analysis is 38.
Source: ISSP (2016), UNDP (2014), World Bank (2018), Taiwan National Statistics (2018)
Among other individual-level variables, the share of paid work, perceived fairness of the division
of housework, presence of children, and egalitarian gender ideology appear to play a significant
role in explaining men’s work-family conflict. Similar to women, performing a large share of
paid work, perceiving doing more or less than their fair share of housework, having more
children, or have more traditional gender ideology are significantly associated with higher levels
of work-family conflict for men. However, age, education, or relative income is not statistically
significant for men’s work-family conflict.
The variance components show that adding the level-1 and level-2 variables to an
unconditional model reduces the between-country variance by 46.8% (from .077 to .041) and the
within-country variance by 3.5% (from .426 to .411) for men. The intraclass correlation is .154 in
an unconditional model (Model 1 of Table 8) and .091 in the full model (Model 4 of Table 8).
Overall, the results show that the level-1 and level-2 measures included in full models explain the
substantial amount of total variation in men’s work-family conflict, like women.
4.2.3. Conclusion and Discussion
In Chapter 4.2, I examine the cross-national variation in work-family conflict in 38
countries. Comparative studies of the effect of country-level gender equality on work-family
conflict are scant compared to other family-related research, such as the division of household
95
labor, perceived fairness of that division, and family life satisfaction. Using the ISSP 2012 data, I
investigate women’s and men’s work-family conflict, with a particular emphasis on the role of
country-level gender equality. Building on the previous literature on work-family conflict and the
division of household labor, I posit that women experience more work-family conflict than men
(Hypothesis 4) and that the effect of household labor on work-family conflict is stronger for
women than men (Hypothesis 5). Consistent with the prior meta-analytic result examining mean
differences in work-family conflict across countries (Allen et al. 2015), I also expect that
individuals in countries with greater gender equality experience less work-family conflict
(Hypothesis 6). Finally, I explore whether and how the effect of country-level gender equality on
work-family conflict differs for men and women, by testing two competing hypotheses
(Hypotheses 7a and 7b).
The results provide support for Hypothesis 4. The results for the descriptive statistics
(Table 6) show that women report more work-family conflict than men in most countries (i.e., 32
out of 38 countries), and about half of these differences (i.e., 15 countries) are statistically
significant. The results for multilevel analysis also show the same tendency. The intercepts, which
indicate the predicted work-family conflict for a respondent with average characteristics in a
country with average characteristics in the sample, are greater for women than men throughout
the models (Tables 7 and 8). The gender difference in intercepts is about .097 (1.961 for women;
1.864 for men) in the unconditional models and about .121 (1.928 for women; 1.807 for men) in
the full models. Although such differences in intercepts do not tell us the significance level, these
results indicate that on average women report higher levels of work-family conflict than men. The
findings also support Hypothesis 5. I expect that work-family conflict is more susceptible to the
96
housework distribution among women than men, in light of the gendered nature of domestic
labor. The results provide empirical support for this, by showing that respondent’s share of
housework remains significant throughout the models for women, but this variable is not
statistically significant for men. It is important to note that perceived fairness of the division of
housework also plays a significant role in explaining work-family conflict for both women and
men throughout the analyses. Specifically, the results show that respondents who perceive doing
roughly their fair share of housework report significantly less work-family conflict than people
who perceive doing more or less than their fair share of housework. These results suggest that
along with the actual degree of equality in the division of household labor, the perceived fairness
of this division also needs to be considered for cross-national research on work-family conflict.
Furthermore, the results provide partial support for Hypothesis 6. Even though the
influence of country-level gender equality on work-family conflict in terms of the gender division
of housework has not been explicitly addressed cross-nationally, previous comparative studies
have shown that family-related behaviors and perceptions are significantly influenced by macro-
level contextual factors, and among those factors country-level gender equality plays a significant
role in family processes (Bataloa and Cohen 2002; Braun et al. 2008; Davis 2010;
Forste and Fox 2012; Fuwa 2004; Geist and Cohen 2011; Greenstein 2009; Hu and Yucel 2018;
Jansen et al. 2016; Kornrich and Eger 2016; Knudsen and Wærness 2008; Öun 2013; Ruppanner
2010a, 2010b; Treas and Tai 2016). Consistent with these findings, our results show that country-
level gender equality plays a significant role in explaining work-family conflict, but there is a
gender difference. For men, the effect of Gender Inequality Index (GII) is positive and
significant, indicating that men in countries with greater gender equality report significantly less
97
work-family conflict than others in more traditional countries. However, the GII is not
statistically significant in explaining women’s work-family conflict, which provides support for
Hypothesis 7a (i.e., The effect of country-level gender equality on work-family conflict varies by
gender). These findings suggest that work-family conflict might be more susceptible to country’s
gender equality climate among men than women.
Another important point this study found is the strong, negative association between
individuals’ egalitarian gender ideology and work-family conflict for both men and women.
Given the gendered nature of work-family conflict, individuals’ gender ideology is expected to
play a significant role in work-family conflict, and this is largely confirmed by the results of this
study. Egalitarian gender ideology at the individual-level is significantly associated with lower
levels of work-family conflict in all models, indicating that women and men with more
egalitarian gender ideology report significantly less work-family conflict than their traditional
counterparts. This is an interesting finding, particularly in light of the opposite impact of
individuals’ gender ideology on conflict over the division of household labor. In a prior
comparative study that examined housework conflict in national contexts, Ruppanner (2010a)
found that women and men with more egalitarian gender ideology report more conflict with their
partner over housework. These findings imply that although people with more egalitarian gender
ideology experience more conflict with their partner about how to divide the housework, they
may have less work-family conflict caused by the competing roles of the work and family
domains. Future research should investigate this nuanced mechanism for a deeper understanding
of gender equality and its consequences in work-family conflict.
98
I acknowledge some limitations that need to be mentioned. First, I cannot rule out the
possibility of other national contexts that may affect work-family conflict and its association with
country-level gender equality. Given the lack of prior cross-national studies that address the role
of country-level gender equality in women’s and men’s work-family conflict with respect to the
gender division of household labor, this study focuses on providing empirical evidence of that, by
controlling for two fundamental national contexts of GDP per capita and welfare state regimes.
Future research should continue to investigate the role of other national contexts in configuring
the relationship between country-level gender equality and work-family conflict for women and
men. Second, I measure work-family conflict without distinction between WIF (i.e., work
interference with family) and FIW (i.e. family interference with work), because it is expected that
the influences of the division of household labor and country-level gender equality may be
manifested regardless of the forms of work-family conflict. As mentioned in the methodology
section, the factor analysis results also indicate that all four items––which consist of the work-
family conflict measure––retain one factor with a high level of reliability. I also conduct
comparable multilevel models using two separate outcome variables of WIF and FIW as a
preliminary analyses and do not find any significant differences compared to the current
conclusions. Future research could expand on the findings of this study by elaborating whether a
different form of work-family conflict matters in explaining cross-national variations in
workfamily conflict. Finally, the current multilevel analyses cannot establish causality, and thus,
we need to be careful when interpreting the results of this study. Nonetheless, our findings still
provide evidence that work-family conflict is lower in countries with high levels of gender
equality.
99
Despite these limitations, this study provides robust evidence of the gendered process of
work-family conflict from a comparative perspective. On the one hand, the results show that men
and women who have more egalitarian gender ideology—and men who live in more egalitarian
countries—report significantly less work-family conflict than those living in more traditional
countries, suggesting that national and individual development toward more gender equality may
be conducive to reducing work-family conflict. On the other hand, the findings demonstrate that
the effects of country-level gender equality and the housework division on work-family conflict
vary by gender, suggesting that future research should consider such gender differences in
workfamily conflict. Work-family conflict has become an important issue for both researchers
and policy makers since the dominant family model largely shifted to the dual-earner family in
most industrial countries. Although important, cross-national studies of work-family conflict are
relatively scant compared to other family-related research. To my knowledge, there is no
comparable research examining the relationship between the division of household labor, gender
equality, and work-family conflict in a cross-national setting. Building on our findings, future
research should continue to explore the link between gender equality and men’s and women’s
work-family conflict, with a particular attention paid to factors to help reduce work-family
conflict in general and women’s disadvantage in work-family conflict.
4.3. Exploring the Relationship between the Divisions of Household Labor, Work-Family
Conflict, and Family Life Satisfaction
4.3.1. Descriptive Analysis
Individual-Level Measures
The descriptive statistics for the dependent and individual-level variables are presented in
Table 9. The sample includes respondents who are 18 or older, living with a spouse or partner,
100
and currently employed in the labor market. The original sample size of respondents who are 18
or older are 52,301. I eliminated 15,760 individuals who do not live with a spouse or partner and
14,288 individuals who do not currently employed in the labor market, resulting in 22,253 cases.
I have 22,006 cases with valid values of the dependent variable. After listwise deletion, the
effective sample size is 15,274 (6,888 for women, 8,386 for men). I lost 6,732 cases (30.6%), and
the key sources of missing date are relative income, perceived fairness of the division of
housework, and respondent’s perceived share of household labor (4,777 cases total – 21.7% ).
Table 9. Individual-Level Descriptive Statistics (N=15,274)
Satisfaction with family life (high = satisfied)
Work-family conflict (high = more conflict)
Respondent’s share of household labor
Perceived fairness (%)
Do more than my fair share
Do roughly my fair share
Do less than my fair share
Age (in years)
Married (%)
Education (in years, standardized by country)
Relative income (high = having higher income)
5.72
1.93
65.56
54.07
40.33
5.60
42.90
85.70
.33
3.73
.94
.71
16.77
10.82
.91
1.35
1
1 0
18
-3.90
1 2.33
7
4
100
93
6.04
7
5.82
1.83
32.47
10.77
47.32
41.92
45.48
86.74
.21
.89
.70
18.50
11.65
86.74
.93
1.20
1
1
0
18
-4.42
1
7
4
100
90
5.44
Mean Mean
SD SD
or % or %
Respondent’s share of paid work
Presence of children (%)
No child
One child
Two or more children
Egalitarian ideology (high = egalitarian)
Source: ISSP (2016)
54.45
46.43
21.70
31.87
3.62
22.42
.74
93
1
100
5
5.18
69.68
46.09
21.38
32.53
3.35
22.67
.74
6.25
1
7
100
5
Female (N=6,888)
Male (N=8,386)
Variable
Min
Max
Min
Max
103
The dependent variable of this chapter is satisfaction with family life. The mean score of
family life satisfaction is 5.72 (SD=.94) for women and 5.82 (SD=.89) for men, indicating on
average women are less satisfied with family life than men. The t-test result for an overall
comparison by gender shows that there is a statistically significant difference between women’s
and men’s family life satisfaction at the .001 level (difference=-.10, t= -6.67). Women also
appear to have more work-family conflict than men (1.93 for women, 1.83 for men), consistent
with the findings in Chapter 4.2. More substantial gender differences are found in
houseworkrelated measures. As expected, women do the majority of household labor in our
analytic sample. Women report that they perform 65.56% of housework relative to their partner,
while men report only 32.47% of housework done by themselves. As for perceived fairness of
the housework division, both men and women perceive unfairness, but in different directions.
More than half of female respondents (54.07%) perceive that they do more than their fair share
of housework, while only 10.77% of male respondents perceive that they do more than their fair
share. In contrast, only 5.60% of women report doing less than their fair share, whereas 41.92%
of men report doing less than their fair share.
Gender differences are also found in the work-related measures—a respondent’s share
of paid work and relative income. Women report their relative share of paid work as 54.45%,
while men as 69.68%. Note that the mean value of respondent’s share of paid work reported by
employed women is over 50%. Even though we should consider response bias and retired men,
this result is somewhat greater than anticipated. This would be because that our sample is
restricted to respondents who are currently employed in the labor market, regardless of their
spouses’ employment status. Another possibility would be that this result itself indicates a slow
104
change in the division of household labor, but a relatively fast change in the division of paid
work, particularly in light of the conventional result for women’s and men’s share of housework.
The average score of relative income is 5.18 for men and 3.73 for women, indicating that on
average men report having a higher income than their partner, while women report having a
lower income than their partner (where 4=“My partner and I have about the same income”).
Finally, women appear to hold more egalitarian gender ideology than men (3.62 for women, 3.35
for men). These results indicate that gendered practices still persist in our analytic sample, in
terms of work and family life.
For age, marital status, educational attainment, and number of children, no significant
gender differences are found. The average age of women and men is about 43 and 45 years,
respectively. The majority of respondents are married or in civic relationship (85.70% for
women, 86.74% for men), and about half of them have no children (46.43% for women, and
46.09% for men).
Country Level Measures
Table 10 presents the mean values of women’s and men’s family life satisfaction and
country-level variables for each country. Among 38 countries, Korean women (4.89) and
Lithuanian men (5.23) report the lowest family life satisfaction, while Argentine women (6.20)
and Venezuelan men (6.20) report the highest. The level of gender equality in each country is
measured by the Gender Inequality Index (GII) by the UNDP. The GII ranges from 0 to 1, with
higher values representing a lower level of gender equality in a given country. The county with
the highest level of gender equality is Denmark with the lowest GII score (.045), while
Venezuela appears to be a country with the lowest level of gender equality with the highest GII
105
score (3.02). As for GDP and welfare regimes for each country, please refer to Table 2. The
descriptive statistics show a wide range of the dependent variable and country-level measures.
Table 10. Country-Level Descriptive Statistics (N=38)
Country
N
Mean
Family Life
Satisfaction for
Women
Mean
Family Life
Satisfaction for
Men
Mean Difference
Between
Women
and Men
Gender
Inequality
Index
Argentina
243
6.20
6.17
.03
.37
Australia
552
5.77
5.84
-.07
.13
Austria
349
5.76
5.83
-.06
.10
Belgium
674
5.83
5.77
.06
.07
Bulgaria
238
5.64
5.94
-.30 *
.22
Canada
290
5.61
5.79
-.18
.11
Chile
243
5.87
6.07
-.20
.35
Croatia
290
5.96
6.11
-.14
.13
Czech Republic
601
5.39
5.45
-.06
.13
Denmark
615
5.82
5.84
-.02
.05
Finland
464
5.76
5.69
.08
.07
France
697
5.68
5.70
-.02
.11
Germany
600
5.65
5.77
-.12
.09
Hungary
253
5.76
5.79
-.03
.26
Iceland
464
6.02
5.95
.07
.08
Ireland
390
5.71
5.85
-.14
.14
Israel
379
5.94
6.04
-.10
.13
Japan
307
5.25
5.48
-.23
.13
South Korea
408
4.89
5.34
-.45 ***
.08
Latvia
329
5.34
5.59
-.25 *
.19
Lithuania
267
5.04
5.23
-.19
.14
Mexico
303
5.93
6.09
-.16
.37
Netherlands
435
5.80
5.94
-.14
.05
Norway
606
5.67
5.78
-.11
.06
Philippines
436
5.94
6.01
-.06
.45
Poland
313
6.00
6.13
-.13
.15
Portugal
195
5.91
6.02
-.10
.12
Russia
284
5.39
5.66
-.27 *
.29
Slovakia
361
5.66
5.67
-.01
.19
Slovenia
279
5.78
5.78
.00
.07
South Africa
315
5.85
5.97
-.12
.41
Spain
695
5.79
5.91
-.12
.10
106
Sweden
403
5.69
5.74
-.05
.05
Switzerland
468
5.97
5.95
.02
.05
Taiwan
622
5.53
5.64
-.11
.05
United Kingdom
281
5.83
5.89
-.05
.16
United States
405
5.97
6.03
-.06
.23
Venezuela
220
6.19
6.20
-.01
.47
Source: ISSP (2016), UNDP (2014), World Bank (2018), Taiwan National Statistics (2018)
Note: *p < .05, **p < .01, ***p < .001.
To find if there are any systematic gender differences in family life satisfaction for each
country, I conduct a series of t-tests for each country (Table 10). As discussed earlier, there is a
statistically significant difference between women’s and men’s family life satisfaction for an
overall comparison. However, the t-test results for each country do not show many significant
results. Only four countries (i.e., Bulgaria, South Korea, Latvia, and Russia) show statistically
significant differences in family life satisfaction between women and men. Otherwise, a
difference is not significant, even though women tend to report less family life satisfaction than
men in most of the sampled countries.
To sum up, the results indicate that on the average women are less satisfied with family
life than men, but almost no systematic gender difference in family life satisfaction is found in
each country. Yet, there are still significant gender disparities in work-family conflict and
household characteristics, such as the housework division and perceived fairness of that division,
along with cross-national variations in other country-level measures. Recent comparative
research has found that cross-national variation in men’s and women’s family life satisfaction are
determined by different processes at the individual-level (Kornrich and Eger 2016). Given this,
even though there are less systematic gender gaps in family life satisfaction across countries, the
effects of individual- and country-level variables on family life satisfaction might vary by
gender.
107
Furthermore, previous cross-national research shows that different national contexts play
a significant role in shaping individuals’ satisfaction with family life, by creating different
environments within countries (Greenstein 2009, Hu and Yucel 2018). This suggests that national
contexts may moderate the relationship between work-family conflict and family life
satisfaction, and the moderating effect of national contexts may also vary by gender. In the next
chapter, I attempt to answer these questions by conducting a multilevel model of women’s and
men’s family life satisfaction.
4.3.2. Multilevel Analysis
Multilevel Models of Women’s Satisfaction with Family Life
Table 11 presents the results of multilevel regression models predicting women’s family
life satisfaction. Model 1 includes two housework-related measures (i.e., the division of
household labor and perceived fairness of that division), net of individual controls. Consistent
with prior research, the results indicate that the division of housework and perceptions of fairness
are significantly associated with women’s family life satisfaction. Specifically, women who do a
smaller share of household labor and who perceive their share of housework to be fair tend to
report greater satisfaction with family life. Among other individual variables, being younger,
being married, doing a smaller share of paid work, and holding more egalitarian gender role
ideology are positively associated with women’s satisfaction with family life.
Model 2 includes the work-family conflict measure, net of individual controls. Building
on previous work-family conflict research, it is expected that work-family conflict is negatively
associated with family life satisfaction. Our results provide support for this hypothesis. The
coefficient of work-family conflict is significant and negative (b=-.263, p<.001), indicating that
108
women who experience more work-family conflict are significantly less satisfied with their
family life. In Model 2, most of individual-level controls remain significant and in the same
direction, except for the standardized measure of education. Education becomes significant in
Model 2, but it is insignificant in other models.
Table 11. Multilevel Analysis of Women’s Satisfaction with Family Life (N=6,888)
Model 1
Model 2
Model 3
Model 4
Model 5
Model 6
Model 7
Intercept 5.695***
5.555 ***
5.700 ***
5.637 ***
5.635 ***
5.645 ***
5.645 ***
Level 1: Individual-level variables
Age -.034***
-.030 ***
-.030 ***
-.029 ***
-.030 ***
-.029 ***
-.031 ***
Age squared
.000***
.000 **
.000 **
.000 **
.000 **
.000 **
.000 **
Married1
.200***
.204 ***
.198 ***
.201 ***
.201 ***
.202 ***
.204 ***
Education
.010
.026 *
.020
.020
.020
.020
.023
Relative income2
-.014
-.004
-.009
-.009
-.009
-.010
-.008
R’s share of paid work
-.003***
-.002 ***
-.003 ***
-.003 ***
-.003 ***
-.003 ***
-.003 ***
Presence of children3
One child
-.035
-.033
-.019
-.021
-.019
-.021
-.015
Two or more children
-.017
-.023
.000
-.004
-.002
-.004
-.003
Egalitarian ideology4
.076***
.057 ***
.044 **
.048 **
.046 **
.048 **
.044 **
R’s share of household labor
-.003***
-.003 ***
-.003 ***
-.003 ***
-.003 ***
-.003 ***
Perceived fairness5
Do more than a fair share
-.215***
-.177 ***
-.181 ***
-.176 ***
-.183 ***
-.178 ***
Do less than a fair share
-.149**
-.114 *
-.119 *
-.115 *
-.112 *
-.113 *
Work-family conflict6
-.263 ***
-.252 ***
-.252 ***
-.253 ***
-.252 ***
-.250 ***
Level 2: Country-level variables
GII7
.684
.461
.605
.614
GDP
.002
.002
.002
.002
Welfare regime8
Liberal
.156
.156
.158
.172
Former socialist
-.018
-.028
-.055
-.017
Social democratic
.018
.017
.008
.023
Developing
.270
.270
.301
.261
Cross-level interaction
R’s share of household labor X GII
.016 **
Do more than my fair share X GII
-.061
Do less than my fair share X GII
-.254
Work-family conflict X GII
.425 *
(table cont’d.)
99
Model 1
Model 2
Model 3
Model 4
Model 5
Model 6
Model 7
Variance components
Between-country
.066***
.076 ***
.075 ***
.050 ***
.052 ***
.051 ***
.048 ***
Within-country
.790***
.773 ***
.762 ***
.762 ***
.761 ***
.760 ***
.757 ***
R’s share of household labor slope
.000
Perceived fairness slope
.006
Work-family conflict slope
.007 *
ICC (%)
7.7%
8.9%
9.0%
6.1%
6.3%
6.3%
6.0%
BIC
18160.5
17998.0
17922.6
17961.2
17980.8
17993.3
17969.2
Wald chi-square (df)
284.9 (12)
437.7 (10)
546.9 (13)
563.2 (19)
570.5 (20)
539.1 (21)
438.9 (20)
Log-likelihood
-9014.0
-8941.6
-8890.6
-8883.4
-8879.9
-8881.8
-8874.1
Note: *p < .05, **p < .01, ***p < .001. Total number of countries for the analysis is 38.
Source: ISSP (2016), UNDP (2014), World Bank (2018), Taiwan National Statistics (2018)
1. Reference is not married. 2. A high score indicates that a respondent has a high income than their partner, where 4 represents a
respondent and his/her partner have about the same income. 3. Reference is no child. 4. A high score indicates more egalitarian
gender role ideology. 5. Reference is doing roughly a fair share. 6. A high score indicates more work-family conflict. 7. GII is
Gender Inequality Index, where 0 indicates the most egalitarian country. 8. Reference is familialist.
100
112
Model 3 is the final level-1 model that includes both housework-related and work-family
conflict measures, net of individual controls. The net coefficient of work-family conflict is
almost unaffected by the inclusion of the housework division and perceptions of fairness
(decreased from -.263 to -.252) indicating that such housework-related variables hardly mediate
the association between work-family conflict and women’s family life satisfaction. The net
coefficients of perceived fairness of the housework division are somewhat mediated by the
inclusion of work-family conflict, decreasing from -.215 to -.177 for the perception of doing
more than a fair share, and decreases from .146 to .114 for the perception of doing less than a fair
share. However, they remain significant in the same direction. This suggests that work-family
conflict, the division of housework, and perceptions of fairness are all significantly associated
with women’s family life satisfaction, and their net influences on women’s satisfaction are
largely independent of one another.
Model 4 adds the country-level variables to Model 3: Gender Inequality Index (GII),
country’s gross domestic product (GDP) per capita, and welfare regimes. No significant changes
are found with respect to individual-level variables, but the model fit significantly improves.
Compared with Model 3, adding country-level variables reduces the between-country variance of
country-level intercept by 33.5% (from .075 to .050). In other words, three country-level
variables explain 33.5% of the total country-variance in women’s satisfaction with family life.
Inclusion of these country-level variables also reduces the intraclass correlations from .090 (in
Model 3) to .061 (in Model 4). Nonetheless, none of the main effects of country-level variables
are statistically significant. This suggests that the level-1 variations may have a stronger and
113
more straightforward influence on women’s family life satisfaction than the level-2 variations in
the model.
I then examine the cross-level interaction with GII to test whether the effects of
workfamily conflict, along with the division of housework and perceived fairness of that
division, on women’s family life satisfaction vary by country-level gender equality. To avoid
multicollinearity due to a relatively small number of level-2 variables, a piecewise block of
interaction terms is added to Models 5, 6, and 7, respectively. The results show that there are
significant interaction effects between GII and work-family conflict (Model 7), as well as
between GII and the division of household labor (Model 5).
To facilitate interpretation, Figure 2 depicts predicted values of women’s family life
satisfaction by work-family conflict in low and high GII countries, holding all other variables at
their means. Here, I operationalize low GII (i.e., a higher level of gender equality) as 2 standard
deviations below the mean and high as 2 standard deviations above the mean. Likewise, low
work-family conflict is operationalized as 2 standard deviations below the mean and high as 2
standard deviations above the mean. In Figure 2, it is clear that despite the overall negative effect
of work-family conflict on women’s satisfaction with family life, such a negative effect is
stronger in countries with a higher level of gender equality (i.e., low GII countries). As a result,
when work-family conflict is low, there is almost no difference in women’s satisfaction with
family life by country-level gender equality, but the difference is widening as work-family
conflict increases. These results indicate that women who live in countries with a higher level of
gender equality are more susceptible to the adverse effect of work-family conflict on family life
satisfaction.
114
The effect of the division of household labor also significantly varies by country-level
gender equality. Figure 3 presents predicted values of women’s family life satisfaction by their
share of housework in low and high GII countries, holding all other covariates at their means. As
before, low GII and a small share of housework are operationalized as 2 standard deviations
below the means and high GII and a large share of housework as 2 standard deviations above the
means. In countries with a lower level of gender equality (i.e., high GII countries), the slope is
nearly flat, indicating that women’s satisfaction with family life is almost unaffected by changing
their share of housework, when work-family conflict and other covariates are controlled.
However, the negative association between doing a larger share of housework and women’s
family life satisfaction is substantially stronger in countries with a higher level of gender equality
(i.e., low GII countries), indicating that women who perform a large share of housework in
countries with great gender equality have substantially lower family life satisfaction than their
counterparts in countries with a lower level of gender equality.
Overall, the findings suggest the moderating role of country-level gender equality in the
associations of women’s family life satisfaction with work-family conflict, as well as the division
of household labor.
115
Figure 2. Women’s Satisfaction with Family Life, by Work-Family Conflict and Gender
Inequality Index (GII)
Figure 3. Women’s Satisfaction with Family Life, by a Respondent’s Share of Household Labor
and Gender Inequality Index (GII)
Low
Moderate
High
Work-family conflict
Low GII
High GII
Small
Large
Moderate
Respondent's share of housework
Low GII
High GII
116
Multilevel Models of Men’s Satisfaction with Family Life
I now turn to results of men’s satisfaction with family life. Table 12 provides the results
of multilevel regression models predicting men’s family life satisfaction. Model 1 includes two
housework-related measures (i.e., the division of housework and perceived fairness of that
division), net of individual controls. The results of Model 1 are mostly consistent with those for
women, even though the strengths of housework-related measures somewhat decrease. Like
women, men who do a smaller share of household labor and who perceive their share of
housework to be fair tend to report greater satisfaction with family life. The distinctive gender
difference is found in the measure of egalitarian gender role ideology. For women, egalitarian
gender ideology is significantly associated with greater satisfaction with family life. However, it
does not seem to be a significant factor in explaining men’s family life satisfaction. Men’s
egalitarian gender ideology is only significant in Model 1 and loses significance in other models.
Model 2 includes work-family conflict, net of individual controls. The coefficient of
work-family conflict for men is significant and negative (b=-.197, p<.001), indicating that men
who have more work-family conflict are significantly less satisfied with their family life. This
result shows that work-family conflict plays an important role in shaping family life satisfaction
for men, like for women.
Table 12. Multilevel Analysis of Men’s Satisfaction with Family Life (N=8,386)
Intercept
Model 1
5.662***
Model 2
5.627 ***
Model 3
Model 4
Model 5
Model 6
Model 7
5.635***
5.598 ***
5.594 ***
5.595***
5.581 ***
Level 1: Individual-level variables
Age
-.037***
-.035 ***
-.036***
-.035 ***
-.035 ***
-.035***
-.035 ***
Age squared
.000***
.000 ***
.000***
.000 ***
.000 ***
.000***
.000 ***
Married1
.209***
.221 ***
.213***
.217 ***
.217 ***
.217***
.216 ***
Education
.015
.015
.017
.017
.017
.017
.016
Relative income2
.006
.011
.006
.005
.005
.005
.006
R’s share of paid work
-.001
.000
.000
.000
.000
.000
.000
Presence of children3
One child
-.031
-.015
-.020
-.021
-.022
-.020
-.021
Two or more children
.033
.051 *
.047
.044
.043
.044
.045
Egalitarian ideology4
.036*
.019
.023
.027
.028
.027
.026
R’s share of household labor
-.002***
-.002***
-.002 ***
-.002 ***
-.002***
-.002 ***
Perceived fairness5
Do more than a fair share
-.116***
-.089**
-.094 **
-.095 **
-.114**
-.092 **
Do less than a fair share
-.065**
-.044*
-.045 *
-.047 *
-.046
-.043 *
Work-family conflict6
-.197 ***
-.194***
-.194 ***
-.195 ***
-.194***
-.198 ***
Level 2: Country-level variables
GII7
.871
.972
.659
.912
GDP
.002
.002
.002
.002
Welfare regime8
Liberal
.100
.135
.110
.126
Former socialist
-.038
-.041
-.033
-.010
Social democratic
-.002
.019
-.004
.017
Developing
.196
.190
.193
.188
Cross-level interaction
R’s share of household labor X GII
.006
Do more than my fair share X GII
.920***
Do less than my fair share X GII
.042
Work-family conflict X GII
.121
(table cont’d.)
106
Variance components
Between-country
Model 1
.047***
Model 2
.056 ***
Model 3
Model 4
Model 5
Model 6
Model 7
.058***
.034 ***
.037 ***
.055***
.033
***
Within-country
.738***
.724 ***
.722***
.722 ***
.722 ***
.719***
.719
***
R’s share of household
labor slope
.000
Perceived fairness slope
.004
Work-family conflict
slope
.006 *
ICC (%)
6.0%
7.2%
7.5%
4.4%
4.9%
7.1%
4.4%
BIC
21490.0
21320.6
21321.9
21357.1
21381.6
21372.6
21373.0
Wald chi-square (df)
154.0 (12)
310.6 (10)
337.7 (13)
356.9 (19)
358.5 (20)
371.1 (21)
284.3
(20)
Log-likelihood
-10677.2
-10601.6
-10588.7
-10579.2
-10577.9
-10568.9
-10573.5
Note: *p < .05, **p < .01, ***p < .001. Total number of countries for the analysis is 38.
Source: ISSP (2016), UNDP (2014), World Bank (2018), Taiwan National Statistics (2018)
1. Reference is not married. 2. A high score indicates that a respondent has a high income than their partner, where 4 represents a
respondent and his/her partner have about the same income. 3. Reference is no child. 4. A high score indicates more egalitarian
gender role ideology. 5. Reference is doing roughly a fair share. 6. A high score indicates more work-family conflict. 7. GII is
Gender Inequality Index, where 0 indicates the most egalitarian country. 8. Reference is familialist.
107
120
Model 3 is the final level-1 model that includes both the housework-related measures and
work-family conflict, net of individual controls. The net coefficient of work-family conflict
remains almost the same, after controlling for the division of housework and perceptions of
fairness. This indicates that work-family conflict plays a significant and independent role in
explaining men’s family life satisfaction, net of the division of housework and perceptions of
fairness. On the other hand, the influences of perceived fairness on men’s satisfaction are
somewhat mediated by the inclusion of work-family conflict (from -.116 to -.089 for perception
of doing more than a fair share; from -.065 to -.044 for perception of doing less than a fair share),
but they still substantially explain men’s family life satisfaction in the model. This indicates that
two housework-related measures have significant net influences on men’s family life satisfaction,
largely independent of work-family conflict. These results are generally consistent with those for
women.
Model 4 adds the country-level variables to Model 3: Gender Inequality Index (GII),
country’s gross domestic product (GDP) per capita, and welfare regimes. Similar to women,
including the level-2 variables does not yield distinguishable differences in the association
between the level-1 variables and men’s family life satisfaction, but the model fit significantly
improves. Compared with Model 3, adding country-level variables reduces the between-country
variance of country-level intercept by 42.3% (from .058 to .034). In other words, three
countrylevel variables explain 42.3% of the total country-variance in men’s satisfaction with
family life. Inclusion of country-level variables also reduces the intraclass correlations by 40.4%
(from .075 in Model 3 to .044 in Model 4), indicating that the proportion of the between-country
variance in men’s family life satisfaction decreases by 40.4% after introducing country’s GII,
121
GDP, and welfare regimes. However, none of the main effects of country-level variables provides
significance.
Models 5, 6, and 7 include a piecewise block of cross-level interaction terms, to test
whether the effects of work-family conflict and housework-related measures on men’s family life
satisfaction differ by country-level gender equality. As for women, significant interaction effects
are found between GII and work-family conflict (Model 7 of Table 11), as well as between GII
and housework division (Model 5 of Table 11). In contrast, the significant cross-level interaction
effect for men is only found between GII and perceived fairness of the housework division
(Model 6 of Table 12), indicating that the association between perceived fairness and men’s
family life satisfaction is moderated by country-level gender equality. Specifically, in countries
with a high level of gender equality, men who have the perception of doing more than their fair
share are substantially less satisfied with their family life, compared to a reference group who
perceive their share of housework to be fair. On the contrary, in countries with a low level of
gender equality, men’s satisfaction with family life varies little by how they perceive their share
of housework.
4.3.3. Conclusion and Discussion
In Chapter 4.3, I examine the cross-national variations in women’s and men’s satisfaction
with family life among 38 countries. Recent cross-national research has documented that the
unequal division of household labor and perceived housework unfairness have negative
association with satisfactory family life, but the influence of work-family conflict on that
association has received insufficient attention. Using the ISSP 2012 data, I investigate the
relationship between work-family conflict and women’s and men’s family life satisfaction,
122
focusing on the possible moderating effect of country-level gender equality. Building on the
previous literature on work-family conflict, I expect that work-family conflict is negatively
associated with family life satisfaction for both women and men, net of the division of
housework, perceived fairness of this division, and other individual controls (Hypothesis 8).
Building on the relative deprivation framework and the recent cross-national literature, I also
hypothesize that the effect of work-family conflict on family life satisfaction may vary by
country-level gender equality (Hypothesis 9), with a possible differential effect of country-level
gender equality on women’s and men’s satisfaction (Hypothesis 10).
The results provide strong support for Hypothesis 8. A higher level of work-family
conflict is significantly associated with a lower level of family life satisfaction in all models, for
both women and men. The negative association between work-family conflict and family life
satisfaction is almost unchanged after controlling for the division of housework and perceptions
of fairness. In other word, the negative effect of work-family conflict is not mediated by the
inclusion of such housework-related variables. The net influences of the division of housework
and perceptions of fairness also remain significant in the same direction when work-family
conflict is taken into account, even though the association between perceptions of fairness and
family life satisfaction is somewhat mediated. This suggests that the adverse effect of
workfamily conflict on family life satisfaction is largely independent of the well-documented
negative influences of the unequal division of household labor and perceived unfairness, and vice
versa.
The results also find that the association between work-family conflict and family life
satisfaction varies by countries, and the moderating effect of country-level gender equality on
123
that association differs by gender. For women, country-level gender equality moderates the
association of women’s satisfaction with work-family conflict and the division of household
labor. Specifically, the negative effects of work-family conflict and unequal divisions of
housework on women’s satisfaction are significantly stronger in countries with a higher level of
gender equality. This suggests that country-level gender equality is a significant moderator in
explaining women’s family life satisfaction, in terms of work-family conflict and the division of
household labor. For men, however, country-level gender equality only moderates the association
between perceptions of fairness and men’s satisfaction, indicating that the moderating role of
country-level gender equality differs by gender. Taken together, the findings provide support for
Hypotheses 9 and 10. Yet, it is important to note that the main effect of country-level gender
equality is not statistically significant either for women or men. Given this, further examination
is needed to elaborate the moderating role of national contexts in the association between work-
family conflict and women’s and men’s satisfaction.
This study has some limitations. First, I do not distinguish WIF (i.e., work interference
with family) from FIW (i.e. family interference with work), because it is expected that
workfamily conflict has the negative influence on family life satisfaction, regardless of the type
of work-family conflict. As mentioned earlier in the methodology section, the factor analysis
results also indicate that all four items––which consist of the work-family conflict measure––
retain one factor with a high level of reliability. Thus, given the lack of cross-national research
on the role of work-family conflict in the association between the gender division of housework
and family life satisfaction, this study focuses more on providing empirical evidence of the
negative contribution of work-family conflict to women’s and men’s satisfaction in general,
124
without distinction between WIF and FIW. Future research could expand on the findings of this
study by exploring whether and how the different types of work-family conflict matter. Second,
there are likely other national contexts not captured in this study that may affect the association
between work-family conflict and family life satisfaction. For instance, social policies play a
significant role not only in shaping women’s and men’s work-family conflict (Stier et al. 2012)
but also in distributing housework between partners. (Fuwa and Cohen 2007). This suggests that
policies in ameliorating work-family incompatibilities at the national-level may play a role in the
mechanism over family life satisfaction in relation to work-family conflict and housework
divisions. Future research should continue to investigate the influence of other national contexts–
–such as family friendly policies––in configuring the relationship between work-family conflict
and family life satisfaction. Finally, the current multilevel analyses cannot establish casual order.
For instance, although the findings demonstrate that the negative association between
workfamily conflict and women’s satisfaction is stronger in countries with greater gender
equality, it can also be interpreted that a level of gender equality is greater in countries with the
stronger association between work-family conflict and women’s satisfaction. Thus, we should be
careful of interpreting the results.
Despite these limitations, this study provides robust evidence of the negative association
between work-family conflict and satisfactory family life for both men and women. The findings
also reveal that the moderating role of country-level gender equality in that association varies by
gender. These results have a straightforward policy implication: Social policies that reduce
individuals’ work-family conflict have the potential to increase family life satisfaction for both
women and men, and the effects of such policies may be stronger for women than men, in
125
countries with a higher level of gender equality. Even though a growing number of comparative
research has examined the relationship between the division of household labor, perceptions of
housework fairness, and family life satisfaction, to my knowledge, there is no comparable
research examining the role of work-family conflict on the association between the gender
division of housework and family life satisfaction in a cross-national setting. Building on the
findings of this study, future comparative research should continue to examine the important and
independent role of work-family conflict as the key mechanism underlying women’s and men’s
family life satisfaction, in terms of the gender division of household labor.
126
Chapter 5. Conclusion
The purpose of this dissertation is to provide contextual understanding of the persistence
of gender inequality at home from a comparative perspective. Contemporary societies have
experienced substantial developments leading to more gender equality in the past several
decades. Although equality between men and women has been considered desirable in the
domain of paid labor in most industrial societies, unpaid labor in the domestic sphere is one of
the most obstinate realms of gender inequality. In particular, women still perform the majority of
household labor in most societies regardless of the individual- and country-level characteristics,
which may result in women’s disadvantage in family life. Given this, I seek to address how
gendered relations are (re)produced and legitimated in the family, and whether and how national
contexts play a role in explaining the cross-national variations in the division of household labor
and related-family life. To achieve this goal, I analyze cross-national variations in the division of
household labor, work-family conflict, and family life satisfaction by conducting a series of
multi-level analyses using data from the 2012 International Social Survey Programme.
In order to examine the cross-national variations in the division of household labor, I
focus on the role of personal preferences on work-family arrangement, in light of preference
theory (Hakim 2000, 2003) as well as the expanding roles of personal choice to family life in
modern societies (Cherlin 2004). The results demonstrate that individuals’ preferences on
workfamily arrangement play a significant role in explaining the division of housework across
countries. Specifically, personal preferences for the equal contribution in terms of work-family
arrangement are significantly associated with the relatively equal distribution of household labor
between partners. At the macro-level, female labor force participation and welfare state regimes
127
are significantly related to the division of housework. This findings sheds new light on the role of
personal preferences on work-family arrangement in explaining the cross-national variations in
the division of household labor.
Next, I investigate the cross-national variations in work-family conflict in terms of the
gender division of household labor. Despite growing interest in work–family conflict across the
globe, examining work-family conflict with respect to unequal gender relations from a
comparative perspective has received insufficient attention. The results show that men and
women who have more egalitarian gender ideology—and men who live in more egalitarian
countries—have significantly less work-family conflict than their traditional counterparts. This
suggests that national and individual development toward more gender equality may be
conducive to reducing work-family conflict. On the other hand, the findings demonstrate that the
effects of country-level gender equality and the housework division on work-family conflict vary
by gender, suggesting that future research should consider such gender differences in workfamily
conflict.
Finally, I explore the relationship between the division of household labor, work-family
conflict, and family life satisfaction from a comparative perspective. Recent cross-national
research has documented that the unequal division of housework and perceived unfairness of that
division have the adverse effects on family life satisfaction, but studies examining the influence
of work-family conflict on that association have been scarce. Building on relative deprivation
theory (Crosby 1976) and previous literature (Allen et al. 2015; Greenstein 2009; Kornrich and
Eger 2016), I investigate the relationship between work-family conflict and women’s and men’s
family life satisfaction, focusing on the possible moderating effect of country-level gender
128
equality. The results show that work-family conflict is negatively related to satisfactory family
life for both men and women. Moreover, I find that the moderating role of country-level gender
equality in that association varies by gender. The negative effects of work-family conflict and the
unequal divisions of housework on family life satisfaction are significantly stronger in countries
with a higher level of gender equality for women, but not for men. This suggests that social
policies that reduce individuals’ work-family conflict may have the potential to increase family
life satisfaction for both women and men, but such policies may be effective for women than
men, in countries with a higher level of gender equality.
In sum, this dissertation provides robust evidence that gender inequality at home is an
important factor in configuring individuals’ work-family conflict and family life satisfaction, as
well as that national contexts play a significant role in explaining the cross-national variations in
housework, work-family conflict, and family life satisfaction. Given the findings, I argue that the
shift toward gender equality at both individual- and country-levels is the key in reducing
workfamily conflict and achieving satisfactory family life. By shedding light on the cross-
national variations in the relationship between housework, family life, and gender inequality, this
dissertation contributes to a better understanding of the mechanism underlying the gender
division of household labor and its consequences in family life.
