The Life Satisfaction Advantage of Being Married and Gender Specialization

Abstract

This investigation examined whether the life satisfaction advantage of married over unmarried people decreased over the past 3 decades and whether the changes in contextual gender specialization explained this trend. Contextual gender specialization was defined as a country-year-specific share of married women who fully specialize in household work. The author used representative data from the World Values Survey-European Values Study integrated data set for 87 countries (N = 292,525) covering a period of 29 years (1981-2009). The results showed that the life satisfaction advantage of being married decreased over time among men but not among women. Furthermore, the decline of contextual gender specialization correlated with this trend in developed but not in developing countries. In developed countries the life satisfaction of unmarried people increased as the contextual gender specialization declined, whereas the life satisfaction of married people was not affected.

Headnote

This investigation examined whether the life satisfaction advantage of married over unmarried people decreased over the past 3 decades and whether the changes in contextual gender specialization explained this trend. Contextual gender specialization was defined as a country-year-specific share of married women who fully specialize in household work. The author used representative data from the World Values Survey-European Values Study integrated data set for 87 countries (N = 292,525) covering a period of 29 years (1981-2009). The results showed that the life satisfaction advantage of being married decreased over time among men but not among women. Furthermore, the decline of contextual gender specialization correlated with this trend in developed but not in developing countries. In developed countries the life satisfaction of unmarried people increased as the contextual gender specialization declined, whereas the life satisfaction of married people was not affected.

Key Words: families and individuals in societal contexts, family economics, gender roles, marriage, trends, well-being.

A large body of literature shows that married people are happier and more satisfied with their lives than unmarried persons (see, e.g., Gove, Style, & Hughes, 1990; Mastekaasa, 1994; Stack & Eshleman, 1998; Verbakel, 2012), yet growing divorce and cohabitation rates and falling marriage and fertility rates suggest a "retreat from marriage" (see, e.g., Adams, 2004; Cherlin, 2004; Huston & Melz, 2004; Popenoe, 1993). Research shows that the life satisfaction advantage of being married (defined as the difference between the population-based averages of the life satisfaction of married and unmarried persons) has decreased over time in the United States (Glenn & Weaver, 1988). This suggests that marriages in contemporary societies have become less advantageous than they were in the past.

Over recent decades, another change has occurred: Men and women now allocate their time in a more similar way than they did in the past (Bianchi, Milkie, Sayer, & Robinson, 2000). The employment of women, even those who are married and have children, is now accepted in most developed countries (Brewster & Rindfuss, 2000; Sayer & Bianchi, 2000). The general trend is of a decline in specialization, defined as gendered divisions of tasks within married couples between the labor market (typically assigned to men) and the household work (typically performed by women). Although theoretical approaches in both economics and sociology postulate that married couples benefit from gender specialization within marriages (e.g., Becker, 1981; Parsons, 1949; Stevenson & Wolfers, 2007), the relationship between specialization and the life satisfaction advantage of being married has so far received little attention.

The contribution of this article is twofold. First, I describe how the relationship between marriage and life satisfaction has changed over time and across countries. The second contribution is to verify whether macro-level characteristics-in particular, the contextual gender specialization-account for the changes to the life satisfaction advantage of being married.

CHANGES OVER TIME TO THE LIFE SATISFACTION ADVANTAGE OF BEING MARRIED

Previous research has investigated changes to the life satisfaction advantage of being married over recent decades, providing mixed results. Waite (2000) compared the population averages of the well-being of married persons versus never-married and previously married persons over the period 1972-1996 in the United States and found no significant shift in the life satisfaction advantage of being married. In contrast, Glenn and Weaver (1988) showed a decline in the cross-sectional relationship between marital status and declared happiness over the period 1972-1986, which mainly was due to the negative trend of life satisfaction among married women and a positive trend among never-married men. Both Waite (2000) and Glenn and Weaver focused on population averages; moreover, both examined the case of the United States using data from the General Social Survey.

If the life satisfaction advantage of being married decreases, it may be due to a lowered satisfaction of married persons or to increased life satisfaction of unmarried persons. The former has been examined by studies that have investigated changes in marital quality, marital interaction, and marital conflict. For instance, Waite (2000) and Glenn (1991) showed that the percentage of married men and women who declared that their marriage was very happy has been declining slightly in the United States since the 1970s. Similarly, Amato, Johnson, Booth, and Rogers (2003) showed that in the United States during the years 1980-2000 marital interactions declined, even though marital quality and divorce proneness changed little. In a similar vein, Rogers and Amato (2000) provided evidence that the cohort married between 1981 and 1997 reported less interaction and more marital conflict than the cohort married between 1964 and 1980. However, Corra, Carter, Carter, and Knox (2009) found no consistent trend of satisfaction with marriage during the years 1973-2006 across groups of White and Black husbands and wives. In contrast to these predominantly negative results, the life satisfaction of unmarried people, in particular men, increased during the 1970s and the 1980s (Glenn & Weaver, 1988; G. R. Lee, Seccombe, & Shehan, 1991). In sum, the majority of these results suggest that the life satisfaction advantages of marriage are declining over time.

GENDER SPECIALIZATION IN MARRIAGE

The decline of gender specialization is part of a broader transformation of marriage and family known as the Second Demographic Transition (Lesthaeghe, 2010). Theoretical approaches offer at least four arguments why declining gender specialization may be at the root of the declining life satisfaction advantage of being married.

First, declining gender specialization may directly lower the well-being of married couples. According to the economic household model (Becker, 1981), gender specialization increases the overall productivity of the household; its wealth; and, therefore, the life satisfaction of the couple (Stutzer & Frey, 2006). Similar to Becker's (1981) economic model, the sociological stream of functionalism theorized that the division of roles between men and women and their complementarity was functional for the institution of the family (Parsons, 1949). Both theories suggest that declining gender specialization may erode part of the benefits of marriage. Specialization in the couple intuitively should play a role; however, at the individual level specialization is endogenous to being in a good relationship (e.g., in bad relationships women seek employment to increase their independence; Oppenheimer, 1997). Thus, the relationship between individual specialization and life satisfaction advantage from marriage would reflect, at least in part, self-selection. The focus on contextual gender specialization of this article helps overcome this limitation.

Second, declining gender specialization might have improved the living conditions of unmarried people. Technological progress in household appliances, as well as the market availability of goods and services that replace those produced within the household (washing machines, ready-made meals, etc.) decrease the benefits of household-specific skills, thus lowering the cost of living as a single person (Stevenson & Wolfers, 2007). This change is likely to have increased the life satisfaction of unmarried people more than it increased the life satisfaction of married people, thus reducing the relative advantage of being married.

Third, contextual gender specialization may be related to the life satisfaction advantage of being married when the presence of other social institutions designed to benefit from specialization increase the benefits of specialization within the marriage. The institutions of interest may include educational systems, taxation, provision of welfare, and so on (Esping-Andersen, Boertien, Bonke, & Gracia, 2013). For example, the advantage of living as a married person may be stronger in societies that rely on private provision of care for the elderly than in societies where elder care is provided by the state.

Fourth and finally, the life satisfaction advantage of being married might have decreased with the decline of gender specialization because of the values change that accompanied the Second Demographic Transition (Lesthaeghe, 2010). In the context of lower gender specialization people may systematically expect that marriage satisfies their more complex needs (e.g., the need of self-actualization) compared to in gender-specialized societies, where the need for safety would be the dominating motive to marry (Finkel, Hui, Carswell, & Larson, 2014; Lundberg, 2012; Lundberg & Pollak, 2007, 2013; Stevenson & Wolfers, 2007). The satisfaction of higher needs requires a larger investment of psychological resources than the satisfaction of basic needs; therefore, it is more difficult. This, in turn, may lower marital satisfaction and the subjective well-being of married couples living in societies where gender specialization is rare.

All four of these arguments suggest that contextual gender specialization may correlate with the higher life satisfaction advantage of being married. The cross-country variations in the life satisfaction advantage of being married across social contexts has been so far neglected, even though researchers have examined how social context affects the life satisfaction gap between married and cohabiting couples (see, e.g., Diener, Gohm, Suh, & Oishi, 2000; K. S. Lee & Ono, 2012; Ryan, Hughes, & Hawdon, 1998; Soons & Kalmijn, 2009; Vanassche, Swicegood, & Matthijs, 2012; Verbakel, 2012).

GENDER DIFFERENCES

In gender-specialized societies women have a lower possibility to leave unsatisfactory marriages because gender specialization correlates with scarce employment opportunities for women (Sayer & Bianchi, 2000). Therefore, where the contextual gender specialization is high, marriages tend to be structured to men's advantage and to be more beneficial to men rather than to women. Hence, the gains from marriage in the context of high specialization may be higher for men than for women; moreover, the decline of specialization may benefit married women more than married men.

Taking this into account, in the present analysis I investigated the gender differences in the life satisfaction advantage of being married. To be specific, I allowed for a different trend of this advantage among men and women and for a different relationship between gender specialization and the advantage of being married.

THE PRESENT CONTRIBUTION

The goal of this analysis was twofold. The first goal was to provide evidence of the time trend of life satisfaction advantage of being married. Previous literature shows that this advantage in the United States has declined (Glenn & Weaver, 1988), but evidence in other countries and regions is missing. I verify the hypothesis that the declining life satisfaction advantage of being married is a general trend.

Second, I investigated how macro factors- in particular, contextual gender specialization- affect the life satisfaction advantage of being married. I should stress that I did not examine the life satisfaction difference between women who stay at home and women who work for pay. I was interested in the difference between societies where, because of normative, technological, and institutional reasons, it is easy for a woman to become a housewife and it is difficult to enter employment and societies where (for the same reasons) it is easy for a woman to work for pay and she is unlikely to become a housewife (McRae, 2003). In other words, I investigated the consequences of the social change that has occurred over the past few decades, which led to the normative acceptance of women's employment and the building of institutional support for it. Hence, I tested the hypothesis that marriage became less advantageous as men's and women's roles in marriages became less differentiated.

METHOD

Data

In this analysis I used the World Values Survey-European Values Study (WVS-EVS) integrated data set covering the period 1981-2009 (EVS, 2011; WVS, 2009). The main advantage of the WVS-EVS is its broad coverage: Its data represent nearly 90% of the world's population and range over a period of almost 30 years, allowing for a comparative analysis of countries' time series. In addition, the large number of countries and periods allows for a satisfactory variation in macro-level variables. The limitations of the WVS-EVS data are typical for secondary data (Hofferth, 2005). To be specific, they allow the use of only one measure of gender specialization (i.e., the share of homemakers) and do not allow the use of other measures proposed by the literature (for more details, see the section titled Measurement of Macro-Level Variables).

During the course of the WVS and EVS surveys individual country research agencies and institutions collected data on representative samples of adult populations (age 18 or older). The questionnaires were uniformly structured, and the translation of the English questionnaire into national languages was monitored. The modes of data collection included face-to-face and phone interviews in the case of WVS, face-to-face interviews (either a computer-assisted or paper-and-pencil personal interview) in the case of EVS, and an Internet panel (Finland in EVS). Currently, the WVS data contain five waves (1981-1984, 1989-1991, 1994-1999, 1999-2004, and 2004-2008), and the EVS data contain four waves (1981-1984, 1990-1993, 1999-2001, and 2008-2009). Detailed information on countries and years when the data were collected is shown in Table 1.

The integrated data set contains information for 102 countries and more than 420,000 respondents. Some questions were not included in all countries and waves; therefore, the sample used in the analysis consisted of 138,573 men and 153,952 women, for a total of 292,525 individuals. The percentage of missing cases in countries and waves included in the analysis was 7.7%, which guarantees that the risk of systematic bias of the estimates due to missing data is low.

Empirical Strategy

Focus on the life satisfaction advantage of being married over being unmarried. To answer the research questions I focused on the size of the life satisfaction advantage of being married over being unmarried. Technically, I investigated the coefficient of the "married" variable and its interactions with other variables of interest. This strategy allowed for a distinction between the general macro determinants of life satisfaction and the macro factors that correlate specifically with the life satisfaction of married people. Across social contexts, the average life satisfaction of married and of unmarried people is strongly correlated ( = .99 for the 211 country-waves), which suggests that the life satisfaction of married and unmarried people is-to some extent-determined by the same factors.

I examined the correlation of marital status with life satisfaction in various social contexts (e.g., varying in the level of contextual specialization); a similar strategy was used by Kalmijn (2010), as well as in studies that have examined the life satisfaction gap between married and cohabiting couples (see, e.g., K. S. Lee & Ono, 2012.)

Cross-country differences versus changes over time. A comparative analysis of time trends calls for a distinction between the effects of the cross-country differences and the effects of changes that take place over time. The changes over time and the cross-country differences in the macro factors may be interpreted analogously to within- and between-individual effects in regression models for panel data: The former shows what differences in life satisfaction are associated with within-country changes to the macro factors over time (e.g., a decline of contextual specialization), and the latter identifies what differences in life satisfaction are associated with the cross-country differences in the macro factors (e.g., differences between countries with low and high contextual specialization).

The distinction between changes over time and cross-country differences is relevant for translating the results into policy recommendations. The effects of changes over time control for the unobserved time-invariant differences between countries; therefore, they allow one to draw stronger conclusions. Interpretations of the effects of the cross-country differences in terms of the potential effects of policies are limited because the coefficients also capture the effects of unobserved time-invariant differences among countries, which may be large if countries are at different levels of development or have different cultural backgrounds.

Selection into marriage. Longitudinal studies have shown that the cross-sectional relationship between marriage and life satisfaction is partly causal (Evans & Kelley, 2004; Soons, Liefbroer, & Kalmijn, 2009) and partly shaped by selection: Happier unmarried people have a higher chance of marrying (Stutzer & Frey, 2006), and unhappy married people have a higher chance of divorcing (Zimmermann & Easterlin, 2006). Ignoring selection in the model may produce biased estimates of the relationship between marital status and life satisfaction. Hence, I controlled for selection to marriage. First, among individual control variables, I included the probability of being married, measured as the percentage of married people in the sociodemographic group of the respondent.

Both very high and very low probabilities of being married may indicate stronger selection; therefore, I included linear and quadratic terms to allow for nonlinear relationships.

Second, given that divorce is typically higher in societies where gender specialization is lower (see Oppenheimer, 1997, and Sayer & Bianchi, 2000, among others), among the control variables at the macro level I included divorce ratio. Given that divorces dissolve bad marriages, a high divorce ratio may serve as a measure of the selection out of marriage.

Statistical Method

The main analysis consists of multilevel regression of individual-level life satisfaction modeled as a function of both individual and country characteristics. I use multilevel-rather than ordinary least squares-regression because the hierarchical data (e.g., the multicountry WVS-EVS with individuals nested within country-waves nested within countries) do not satisfy the basic assumption of the independence of observations. This may bias downward the estimated standard errors, which in turn can result in wrongly rejecting or supporting theoretically important conclusions (Bryk & Raudenbush, 1992; Luke, 2004). Multilevel models properly account for the hierarchical structure of the data; they also simultaneously estimate the variation within and between countries and country-waves and attribute the variation unexplained by the model to the specific levels of data.

Random effect multilevel models, such as the one used in this analysis, assume that random effects are not correlated with the explanatory variables; if this assumption is not met, the results are inconsistent. Therefore, I validated the analysis by estimating models with fixed intercepts (dummy variables) for countries and country-waves (Snijders, 2005a). The fixed effects models provided the same results as the random effects models; therefore I present only the results of the random-effects model, which is considered more efficient.

I estimated a three-level model with individuals i nested within country-waves j nested within countries c. The number of waves observed per country varied between one and seven (in the case of Spain). Overall, I observed 211 country-waves, with an average of 2.4 country-waves per country. This small average cluster size at Level 3 is not an obstacle for estimating the effects at this level because the total sample size at this level is of prime importance (Snijders, 2005b).

Formally, the model is described by Equations 1 through 3:

... (1)

... (2)

... (3)

In this model, individual life satisfaction (LSijc) is regressed onto a set of individual, country-wave, and country-level predictors. In Equation 1 the coefficient 2 describes the trend of life satisfaction among unmarried men, and 5 informs how the trend of life satisfaction among unmarried women differs from the one among men. The coefficient 3 describes the overall trend of the life satisfaction advantage of being married among men, and 6 informs how much the trend among women differs from the trend among men. Coefficients 7, 9, 11, and 13 show how the life satisfaction advantage of being married changes (for men and women) with the level of specialization: 7 and 9 refer to within-country changes in specialization over time, whereas 11 and 13 refer to the cross-country differences in the level of specialization. The coefficients 8, 10, 12, and 14 describe how the life satisfaction advantage of being married among men and women correlates with the contextual specialization. The vector Xijc contains the individual-level control variables, Yjc is a vector of the country-wave level control variables, and Zc is a vector of the country-level control variables, whereas BK -BP are the vectors of their respective coefficients.

In the model (see Equations 2 and 3), the only coefficients allowed to vary randomly are the random intercepts jc and c. In other words, average life satisfaction was allowed to vary randomly across country-waves and across countries.

Measurement of Individual-Level Variables

The dependent variable was life satisfaction. Responses to the question "All things considered, how satisfied are you with your life as a whole these days? Please use this card to help with your answer," were given on a 10-point scale that ranged from 1 (dissatisfied)to10(satisfied). The variable has a distribution close to normal, but it is negatively skewed with the grand mean of 6.7. Country-year specific means vary between 3.72 (Moldova in 1996) and 8.5 (Puerto Rico in 2001). Table 2 shows how life satisfaction of married and unmarried people changed over time in the countries analyzed.

Life satisfaction has been shown to be a reliable indicator of subjective well-being, which correlates with physiological symptoms of stress and pleasure (see, e.g., Steptoe & Wardle, 2005; Urry et al., 2004), with third-person judgments (Schneider & Schimmack, 2009), and with satisfaction in particular domains of life (Schimmack, Krause, Wagner, & Schupp, 2010).

Marital status was measured with a set of dummy variables, including (a) married (59% of the final sample), (b) living together as married (~5%), (c) divorced (~4%), (d) separated (1.6%), (e) widowed (~7%), and (f) never married (23% of the final sample).

I controlled for a range of variables that may correlate with both life satisfaction and being married. These include self-declared unemployment (Kalmijn, 2007; Winkelmann & Winkelmann, 1998), self-declared status of housewife/househusband (Treas, van der Lippe, & Tai, 2011), education (Kalmijn, 2007; secondary and tertiary education levels were coded as dummy variables), age (linear and square components, centered at 40), family income class (Burgess, Propper, & Aassve, 2003; Frijters, Haisken-DeNew, & Shields, 2004; measured on a 10-point scale, centered on the country-wave-specific median, missing values replaced with median and flagged; for Wave 2008 of the EVS, a 12-point scale was recoded into a 10-point scale and used as in other waves), having children (Hansen, 2012; with a dummy taking the value of 1 for parents; for Wave 2008 of the EVS the information on children living in the household was used), and health problems (measured on a 5-point scale, centered on the overall mean). In some waves and countries education has been measured not as education level but as the age at which one finished one's education. To include these countries and waves in the analysis, I approximated the education level on the basis of the information about the age when respondents finished their education: I recoded the age 23-35 years into tertiary education and 18.5-23 years into secondary education.

To control for the selection to marriage at the individual level I included the percentage of married people in groups distinguished on the basis of country-wave, age, gender, and education (tertiary education vs. lower; Kalmijn, 2010). Selection into marriage was captured by the interaction of this variable (and its quadratic term, to allow for a nonlinear relationship) with being married. The coefficient shows how the life satisfaction advantage of being married correlates with the probability of being married.

Measurement of Macro-Level Variables

The economic specialization of spouses was approximated with the percentage of homemakers (i.e., women who declared that taking care of the home and children is their main activity) among married women age 18-60. The specialization variable had values between 0.5% (Sweden in 2009) and 87% (Egypt in 2008; overall = .35,SD= .23). (See Table 2 for an overview of levels of specialization in particular countries and their changes over time.)

This measurement directly refers to the work of Becker (1981), who exemplified full specialization as a situation wherein the husband is employed in the labor market and the wife focuses on domestic production. This definition also refers to seeing the growing employment of married women as a sign of declining specialization and interdependency within marriages (Becker, 1985; Oppenheimer, 1994). In constructing the measure I accounted for specialization among married people because specialization is supposed to take place within a marriage; I focused on women because employment is the norm for men and men's homemaking is likely to be driven by special circumstances. I limited the indicator to adults under 60, because after that age people often no longer choose between employment and housework as they retire. Note that this definition is similar to the one used by Stutzer and Frey (2006), who defined couple-level specialization as a situation in which one spouse is employed while the other is inactive or occasionally participates in the labor market. Other authors have measured gender specialization by the relative differences between spouses' wage rates (e.g., Stutzer & Frey, 2006) or between the number of working hours of spouses (e.g., Cohen, 2002); however, the information on working hours or earnings of spouses is not recorded in the WVS-EVS data.

Apart from accounting for contextual specialization, I also included the individual homemaker status in order to control for specialization in a specific couple. Note that estimating the effect of individual-level specialization on life satisfaction with cross-sectional data is problematic because of endogeneity (Oppenheimer, 1997).

To isolate the effect of specialization, I controlled for a range of macro factors that are likely to correlate with the level of specialization and that may affect the life satisfaction of married couples: the gross domestic product (GDP) and fertility rate (Alesina & Giuliano, 2014), the political and social rights of women (Orloff, 1993), and the divorce ratio (Kalmijn, 2007; Stevenson & Wolfers, 2006). I also controlled for the year when the survey was conducted (centered on the year 2000).

The real GDP per capita (retrieved from Heston, Summers, & Aten, 2012) is expressed in international dollars of the year 2000 transformed into a logarithm. The fertility rate (number of children who would be born to a woman if she were to live to the end of her childbearing years and bear children in accordance with current age-specific fertility rates) was taken from the United Nations (2013) database.

The variables referring to the political and social rights of women came from the Cingranelli and Richards (2008) database and ranged between 0 and 3 (0 indicates that women's rights were not guaranteed by law in a given year and country; 1 indicates that women's rights were guaranteed by law but were not enforced in practice; 2 indicates that the rights were guaranteed by law and enforced in some areas, but women were still discriminated against in practice; and 3 indicates that women's rights were guaranteed in both law and practice). Political rights included the right to vote, to run for political office, to hold elected and appointed government positions, to join political parties, and to petition government officials. Social rights covered the right to equal inheritance; to enter into a marriage on a basis of equality with men; to travel abroad; to obtain a passport; to confer citizenship on children or a husband; to initiate a divorce; to own, acquire, manage, and retain property brought into a marriage; to participate in social, cultural, and community activities; to an education; to the freedom to choose a residence/domicile; and to the freedom from female genital mutilation and forced sterilization.

The divorce ratio was measured as a (country-wave-specific) proportion of divorced women among all women age 18-60. I used this information rather than the divorce rate (United Nations, 2013) because the former was available only for 49 (out of 87) countries and 137 (out of 211) country-waves covered by this analysis. The correlation between the two measures on a country-wave level was = .69.

Country-level variables were included in the model as country-specific averages over the observation period (marked as Specializationc and the vector Zc in Equation 1); the values are centered on the grand mean for easier interpretation of the coefficients. These variables capture the effects of cross-country differences in macro factors. Country-wave-level variables (marked as ΔSpecializationjc and the vector Yjc in Equation 1) were defined as the country-wave-specific deviations from the country-specific mean ( ). These variables represent changes within countries over time. All variables were included as main effects as well as interactions with being married and with being a woman.

RESULTS

I started by estimating the null model, that is, an empty model containing only the (fixed and random) intercepts (not shown). In the null model 14.5% of the variation unexplained by the model was associated with the country level, and an additional 4% was associated with the country-wave level, which is a proportion large enough to justify the use of multilevel regression. The Akaike Information Criterion (AIC) ofthenullmodel(AIC=1,293,904,withdf =4) acted as a benchmark for assessing the fit of subsequent models.

To test the two research hypotheses I estimated two models. Model 1 (results shown in Table 3), estimating the unadjusted trend of the life satisfaction advantage of being married, tested the first hypothesis. This model contained the individual-level predictors and variables to capture general trends in life satisfaction (year), changes over time in the life satisfaction advantage associated with marriage (year × marriage), as well as variations in this trend by gender (year× marriage× gender).

Model 2 (shown in Table 3) included also the country- and country-wave-level variables as well as the interactions of contextual variables with being married. Accounting for the interaction of the life satisfaction advantage of being married with contextual specialization allowed verification of the second hypothesis. Note that to conclude that the results support this hypothesis, two conditions need to be met. First, the interaction of marital status with the changes of contextual gender specialization should be positive and significant. Second, the trend of the life satisfaction advantage of being married should decrease after the inclusion of contextual specialization in the model.

The comparison of AIC statistics between the null model and Models 1 and 2 shows that more complex models fit the data better than the simpler models (Model 1 vs. null model: ΔAIC=33,987, df =25, p<.000; Model 2 vs. Model1:ΔAIC=340,df =30,p<.000).Model 2 offered the best fit to the data.

Coefficients of Individual-Level Predictors

In Model 2, married men were 3.1% more satisfied with their lives than the never-married men. For women the life satisfaction advantage of being married over never married was 2.7%. These values are consistent with the literature (Diener et al., 2000).

Cohabitation was weaker but positively correlated with life satisfaction, and being widowed, divorced, or separated correlated negatively with life satisfaction. The correlation of separation with life satisfaction was strongly negative, which is consistent with the finding that the life satisfaction of divorcees is lowest during the periods right before and after divorce (Clark, Diener, Georgellis, & Lucas, 2008; Lucas, 2007). The coefficients for unemployment and health problems were negative, whereas the coefficients for education and family income were positive. The relationship between age and life satisfaction was U shaped.

Selection

The coefficients for selection variables (%married in age-sex-edu group and % married in age-sex-edu group2 , as well as their interactions with being married) were statistically significant for the linear, but not for the quadratic, component. Life satisfaction of unmarried people who had a higher probability of being married was, on average, lower than of those with a lower probability of being married. Life satisfaction of married people did not correlate with the probability of marrying.

Overall Trend of the Life Satisfaction Advantage of Being Married

The first hypothesis stated that the life satisfaction advantage of being married declined, regardless of the country. Model 1 estimated the average trend of the life satisfaction advantage of being married. The advantage of being married decreased for men at a rate of 0.09 per 10 years (see the coefficient of the Married × Year variable in Table 3). With the average advantage of being married of 0.36, such a rate of decline suggests that, at a constant rate, the advantage of being married would decrease to 0 within four decades. This trend resulted from increasing life satisfaction among the unmarried men (Year = 0.14, p = .004) and a nonsignificant trend among the married men (Year + Married × Year: = 0.048, p = .315).

Among women, the life satisfaction advantage of being married did not change over time (Married × Year + Woman × Married × Year: =-.009, p = .496; this trend was statistically significantly more positive than the trend for men). The life satisfaction trend was positive but statistically insignificant both among unmarried (Year + Woman × Year: = .087, p = .069) and among married women (Year + Woman × Year + Married × Year + Woman × Married×Year: =0.077,p=.103).

These results supported the first hypothesis for men, but not for women. Among men, the life satisfaction advantage of being married declined over time, across all countries. The predicted life satisfaction advantage of being married decreased from 0.54 in 1981 to 0.28 in 2009. Among women, the life satisfaction advantage of being married did not change over time; the predicted advantage declined slightly, from 0.38 in 1981 to 0.36 in 2009.

Decline in Specialization and the Life Satisfaction Advantage of Being Married

The results of Model 2 (see Table 3) did not support the second hypothesis; I found no evidence that the decline of contextual specialization explained the declining life satisfaction advantage of being married. The statistically insignificant interaction terms Married ×ΔSpecialization and Married × Specialization indicate that neither the within-country changes of specialization (ΔSpecialization) nor the cross-country differences of specialization ( Specialization) correlated with the life satisfaction advantage of being married.

Other Correlates of the Life Satisfaction Advantage of Being Married

Included primarily as controls, the changes over time of other macro factors were statistically significantly related to the life satisfaction advantage of being married. In particular, economic growth (ΔGDP) and the growing divorce ratio (Δdivorce ratio) correlated not only with higher life satisfaction overall but also with a greater life satisfaction advantage of being married. The expanding social rights of women significantly correlated with the declining life satisfaction advantage of being married.

The size of the life satisfaction advantage of being married also correlated with some cross-country differences. The advantage of married people over the unmarried was larger in countries with a higher GDP ( GDP) and in developing countries, and it was lower in countries with stronger protection of social rights of women, a higher divorce ratio, and a higher fertility rate. Moreover, the life satisfaction of unmarried people correlated positively with higher GDP, living in a developing country, stronger protection of the social rights of women, and with lower levels of gender specialization ( Specialization).

Robustness Checks

The increasing life satisfaction of unmarried people might have partly reflected the growing acceptance of divorce (Kalmijn, 2010). To exclude this factor, I reestimated Model 1 allowing for a different trend in life satisfaction among the divorced or separated people. The results showed that the life satisfaction trend of divorced or separated people did not differ significantly from the trend of other unmarried people. The estimates of trends of the life satisfaction advantage of being married, as well as the estimated trends of life satisfaction of married and unmarried people, did not differ from those obtained in Model 1.

Contextual specialization might have better reflected gender arrangements in young couples than gender arrangements of older couples; therefore, I reestimated Model 2 allowing for a different life satisfaction advantage of being married among people age 39 years or younger, a different trend of life satisfaction in this group, and a different effect of contextual specialization. The unmarried younger people were, on average, less satisfied with their lives than unmarried older people, and the life satisfaction advantage of being married was larger in the younger group ( = 0.35, p < .000) than in the older group ( = 0.24, p < .000). However, changes of contextual specialization did not statistically significantly correlate with a life satisfaction advantage of being married in any of the groups.

This analysis used a heterogeneous sample of countries; therefore, it is possible that the results were the outcome of a specific subsample. Therefore, I reestimated Models 1 and 2 allowing for different trends of life satisfaction in developing and developed countries. Reestimation of Model 1 yielded information about the trends of life satisfaction advantage from marriage in both groups of countries. In developing countries the life satisfaction advantage derived from marriage declined for men ( =-0.14, p < .000) and for women ( =-.06, p = .042), which was an outcome of nonsignificant trends of life satisfaction among both unmarried and married men and women. In developed countries the life satisfaction advantage declined only for men ( =-0.08, p < .000), whereas for women the change was not significantly different from 0 ( =-0.003, p = .849). This process was the result of increasing life satisfaction among single men ( = 0.16, p = .002), single women( =0.11,p=.039),andmarriedwomen ( = 0.11, p = .033), combined with a nonsignificant trend among married men.

Reestimation of Model 2 showed that changes in contextual specialization correlated with the life satisfaction advantage of being married in developed countries ( = .50, p = .017, among men and =0.39, p = .047, among women). In developing countries these relationships were statistically insignificant. The relationship in developed countries was driven by a negative correlation between the contextual gender specialization and life satisfaction of unmarried people ( =-1.04, p = .046, for men and =-0.91, p = .079, for women). It is important to note that the trend in the life satisfaction advantage of being married estimated in Model 2 was not considerably smaller than the trend estimated in Model 1 (men in developed countries: =-0.087, p = .003; women in developed countries: =-0.019, p = .510; men in developing countries: =-0.13, p < .000, women in developing countries: =-0.061, p = .070), which indicates that neither the contextual specialization nor the macro factors explained the trend in the life satisfaction advantage of being married.

DISCUSSION

In this analysis I tested the hypothesis that the decline of the life satisfaction advantage of being married is a general trend and not a feature exclusive to the United States. The results supported this hypothesis for men and showed that over the period 1981-2009 the life satisfaction advantage of married men over unmarried men declined, on average, from 0.54 to 0.28 (on a scale from 0 to 10). This decline resulted from the increasing life satisfaction of unmarried men and the constant life satisfaction of married men. In the same period, women did not experience a significant decline in the life satisfaction advantage of being married. The average life satisfaction of both married and unmarried women remained unchanged.

The analysis also documented that the life satisfaction advantage of being married decreased for men both in developed and developing countries. For women the trend was less negative than for men in both groups of countries: It was insignificant in developed countries and weakly negative in developing countries. Whereas the life satisfaction of married and unmarried men and women in developing countries changed little, in developed countries life satisfaction increased in all groups except for married men.

These results lead to conclusion that-even though marriage remains a preferred living arrangement for many people-life satisfaction of unmarried men and women in developed countries has increased, reducing the life satisfaction advantage of being married, especially for men. This is consistent with the sometimes-expressed opinion that contemporary societies create good conditions for a variety of life choices and living arrangements. Furthermore, the results suggest that the changes that took place over time in developed societies benefited married women more than they benefited married men, perhaps shaping marriage more to women's advantage. Such a change did not occur in developing countries.

Furthermore, in this analysis I tested the hypothesis that the change in the life satisfaction advantage of being married was shaped by the change in contextual gender specialization. The results showed that the life satisfaction advantage of being married correlated with the changes of contextual specialization only in developed countries; specifically, in developed countries the unmarried people benefited from declining contextual gender specialization, whereas the life satisfaction of married people was not affected by this change. Hence, declining contextual gender specialization did not erode the life satisfaction of married couples. Moreover, the trend in the life satisfaction advantage of being married did not change after including gender specialization in the estimation model, which leads to the conclusion that the changes in contextual gender specialization do not explain the decreasing life satisfaction advantage of being married for men.

These results are important because some policies attempt to strengthen the institution of marriage by creating incentives for a gender-specialized division of work. Such policies include, for example, tax systems discouraging the employment of a second earner in a couple or provision of long child care leaves for women. This study suggests that policies to strengthen gender-traditional family arrangements may not be an efficient tool for increasing the life satisfaction of married couples. My results support the claim that women's employment and egalitarian gender roles do not put at risk the well-being of marriages (previously formulated by, e.g., Oppenheimer, 1997, and Rogers & DeBoer, 2001). In other words, social environments that push women to assume homemakers roles seem to bring no inherent advantages to marriages. The current study is the first to demonstrate this result from a broad comparative perspective.

The analysis also identified other macro factors whose changes correlated with the life satisfaction advantage of being married. Growing GDP correlated positively with life satisfaction of unmarried people; it also correlated positively with the life satisfaction advantage of being married. In other words, married people benefit more from improving economic conditions than unmarried people. This is consistent with the observation that economic hardship affects the well-being of married people particularly negatively (Rogers & DeBoer, 2001). Moreover, GDP growth plausibly liberates people from the need to marry, and it eliminates the reasons to tolerate bad marriages. Similarly, as in the case of GDP, a growing divorce rate was also positively correlated with the life satisfaction of unmarried people and with the life satisfaction advantage of being married. These results are consistent with the observation that divorces dissolve unhappy and abusive marriages (see also Stevenson & Wolfers, 2006, on the consequences of the introduction of no-fault divorce). These two sets of results suggest that policies supporting economic development and liberal divorce regulations may positively contribute to the life satisfaction of both married and unmarried people.

There are limitations to this research. First, this study could not distinguish between first and subsequent marriages (Cherlin, 2004). However, Soons et al. (2009) showed that a remarriage allows for a return to the pre-separation levels of life satisfaction, which suggests that the inability to control for this factor does not necessarily create a considerable bias. The study also did not control for the duration of marriages. The process of adaptation gradually decreases the initially high life satisfaction advantages of marriage (Clark et al., 2008; Soons et al., 2009). Consistent with this, one of the robustness checks showed that the life satisfaction advantage of being married was larger among younger people. This factor could bias the results if the average duration of marriages differed systematically across countries and correlated with the explanatory factors. This may be the case for the divorce ratio, which may provide an additional explanation of the positive correlation between the divorce ratio and the life satisfaction advantage of being married. In countries where the divorce rate is higher, the average duration of marriages is shorter, and thus the average life satisfaction advantage of being married is larger. Finally, the control for selection in this study deserves a comment. I used a variable that measured the percentage of married people in an age-gender-education group in a given country and year. Although this is not a perfect measurement, it is probably the only feasible solution for cross-sectional data. Use of comparative panel data or data measuring the psychological traits of spouses (for more on this suggestion, see Huston & Melz, 2004) could solve this problem; they are, however, unavailable. This and other issues should still be addressed by future research.

Overall, this research leaves us with two take-home messages. First, the results suggest that developed countries have experienced a general decline in men's life satisfaction advantage of being married. In this group of countries married men were the only group whose life satisfaction remained stable over time; in contrast to that, the average life satisfaction of unmarried men and women as well as of married women increased over the past three decades. This change was not explained by declining gender specialization or other factors, such as increasing social or political rights of women.

Second, the analysis suggests a different source of life satisfaction advantage of being married than gender specialization. The specialization theory presents marriage as an arrangement that allows individuals to benefit from an exchange of productive skills, thus raising the total household productivity above the sum of the productivities of the spouses. In other words, marriage is seen as advantageous because it responds to economic necessities. In contrast to this approach, the results of this study suggest that the advantages of marriage are greater where the economic necessity is lower. The life satisfaction advantage of being married grew along with increases in GDP and the divorce ratio, which suggests that the advantages of marriage are greater not under the conditions of necessity but in the conditions of free choice. This conclusion is consistent with the literature addressing the transformation of contemporary marriages from an arrangement that satisfies lower level practical needs to an institution that helps achieve personal accomplishment and self-fulfillment.

NOTE

This research was supported by the Incoming Postdoctoral Fellowship cofunded by the Marie Curie Actions of the European Commission from the Université Catholique de Louvain and by the subsidy granted to the National University Higher School of Economics by the Government of the Russian Federation for the implementation of the Global Competitiveness Program. I thank Francesco Sarracino, Joshua Kjerulf Dubrow, Eve O'Callaghan, Jacques Hagenaars, Ronald Inglehart, Eduard Ponarin, Malina Voicu, Andrey Shcherbak, and colleagues from the Laboratory for Comparative Social Research for their helpful comments and discussions on earlier versions of this article.

References

REFERENCES

Adams, B. N. (2004). Families and family study in international perspective. Journal of Marriage and Family, 66, 1076-1088. doi:10.1111/j.0022-2445. 2004.00079.x

Alesina, A., & Giuliano, P. (2014). Family ties. In P. Aghion & S. N. Durlauf (Eds.), Handbook of economic growth (Vol. 2, pp. 177-215). New York: Elsevier. doi:10.1016/B978-0-444-535382.00004-6

Amato, P. R., Johnson, D. R., Booth, A., & Rogers, S. J. (2003). Continuity and change in marital quality between 1980 and 2000. Journal of Marriage and Family, 65, 1-22. doi:10.1111/j.1741-3737. 2003.00001.x

Becker, G. S. (1981). A treatise on the family.Cambridge, MA: Harvard University Press.

Becker, G. S. (1985). Human capital, effort, and the sexual division of labor. Journal of Labor Economics, 3(1, Pt. 2), S33-S58. doi:10.1086/298075 Bianchi,S.M.,Milkie,M.A.,Sayer,L.C.,&Robinson,

J. P. (2000). Is anyone doing the housework? Trends in the gender division of household labor. Social Forces, 79, 191-228. doi:10.2307/2675569

Brewster, K. L., & Rindfuss, R. R. (2000). Fertility and women's employment in industrialized nations. Annual Review of Sociology, 26, 271-296. doi:10.1146/annurev.soc.26.1.271

Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models: Applications and data analysis methods. Newbury Park, CA: Sage.

Burgess, S., Propper, C., & Aassve, A. (2003). The role of income in marriage and divorce transitions among young Americans. Journal of Population Economics, 16, 455-475. doi:10.1007/s00148003-0124-7

Cherlin, A. J. (2004). The deinstitutionalization of American marriage. Journal of Marriage and Family, 66, 848-861. doi:10.1111/j.0022-2445. 2004.00058.x

Cingranelli, D. L., & Richards, D. L. (2008). The Cingranelli-Richards Human Rights data set (Version 2008.03.12). Retrieved from http://www. humanrightsdata.org

Clark, A. E., Diener, E., Georgellis, Y., & Lucas, R. E. (2008). Lags and leads in life satisfaction: A test of the baseline hypothesis. The Economic Journal, 118, F222-F243. doi:10.1111/j.1468-0297. 2008.02150.x

Cohen, P. N. (2002). Cohabitation and the declining marriage premium for men. Work and Occupations, 29, 346-363. doi:10.1177/0730888402 029003004

Corra, M., Carter, S. K., Carter, J. S., & Knox, D. (2009). Trends in marital happiness by gender and race, 1973 to 2006. Journal of Family Issues, 30, 1379-1404. doi:10.1177/0192513X09336214

Diener, E., Gohm, C. L., Suh, E., & Oishi, S. (2000). Similarity of the relations between marital status and and subjective well-being across cultures. Journal of Cross-Cultural Psychology, 31, 419-436. doi:10.1177/0022022100031004001

Esping-Andersen, G., Boertien, D., Bonke, J., & Gracia, P. (2013). Couple specialization in multiple equilibria. European Sociological Review, 29, 1280-1294. doi:10.1093/esr/jct004

European Values Study. (2011). European Values Study 1981-2008, longitudinal data file (Version 2.0.0) [Data file]. GESIS Data Archive, Cologne, Germany. doi:10.4232/1.11005

Evans, M. D. R., & Kelley, J. (2004). Effect of family structure on life satisfaction: Australian evidence. Social Indicators Research, 69, 303-349. doi:10.1007/s11205-004-5578-9

Finkel, E. J., Hui, C. M., Carswell, K. L., & Larson, G. M. (2014). The suffocation of marriage: Climbing Mount Maslow without enough oxygen. Psychological Inquiry, 25, 1-41. doi:10.1080/1047840X. 2014.863723

Frijters, P., Haisken-DeNew, J. P., & Shields, M. A. (2004). Money does matter! Evidence from increasing real income and life satisfaction in East Germany following reunification. American Economic Review, 94, 730-740. doi:10.1257/ 0002828041464551

Glenn, N. D. (1991). The recent trend in marital success in the United States. Journal of Marriage and the Family, 53, 261-270. doi:10.2307/ 352897

Glenn, N. D., & Weaver, C. N. (1988). The changing relationship of marital status to reported happiness. Journal of Marriage and the Family, 50, 317-324. doi:10.2307/351999

Gove, W. R., Style, C. B., & Hughes, M. (1990). The effect of marriage on the well-being of adults: A theoretical analysis. Journal of Family Issues, 11, 4-35. doi:10.1177/019251390011001002

Hansen, T. (2012). Parenthood and happiness: A review of folk theories versus empirical evidence. Social Indicators Research, 108, 29-64. doi:10. 1007/s11205-011-9865-y

Heston, A., Summers, R., & Aten, B. (2012). Penn World Table (Version 7.1). Center for International Comparisons of Production, Income and Prices, University of Pennsylvania. Retrieved from https://pwt.sas.upenn.edu/

Hofferth, S. L. (2005). Secondary data analysis in family research. Journal of Marriage and Family, 67, 891-907. doi:10.1111/j.1741-3737.2005. 00182.x

Huston, T. L., & Melz, H. (2004). The case for (promoting) marriage: The devil is in the details. Journal of Marriage and Family, 66, 943-958. doi:10.1111/j.0022-2445.2004.00064.x

Kalmijn, M. (2007). Explaining cross-national differences in marriage, cohabitation, and divorce in Europe, 1990-2000. Population Studies, 61, 243-263. doi:10.1080/00324720701571806

Kalmijn, M. (2010). Country differences in the effects of divorce on well-being: The role of norms, support, and selectivity. European Sociological Review, 26, 475-490. doi:10.1093/esr/ jcp035

Lee, G. R., Seccombe, K., & Shehan, C. L. (1991). Marital status and personal happiness: An analysis of trend data. Journal of Marriage and the Family, 53, 839-844. doi:10.2307/352991

Lee, K. S., & Ono, H. (2012). Marriage, cohabitation, and happiness: A cross-national analysis of 27 countries. Journal of Marriage and Family, 74, 953-972. doi:10.1111/j.1741-3737.2012.01001.x

Lesthaeghe, R. (2010). The unfolding story of the Second Demographic Transition. Population and Development Review, 36, 211-251. doi:10.1111/j. 1728-4457.2010.00328.x

Lucas, R. E. (2007). Adaptation and the set-point model of subjective well-being: Does happiness change after major life events? Current Directions in Psychological Science, 16, 75-79. doi:10.1111/j.1467-8721.2007.00479.x

Luke, D. A. (2004). Multilevel modeling. Thousand Oaks, CA: Sage. doi:10.4135/9781412985147

Lundberg, S. (2012). Personality and marital surplus. IZA Journal of Labor Economics, 1, 1-21. doi:10.1186/2193-8997-1-3

Lundberg, S., & Pollak, R. A. (2007). The American family and family economics. Working Paper No. 12908, National Bureau of Economic Research, Cambridge, MA. doi:10.3386/w12908

Lundberg, S., & Pollak, R. A. (2013). Cohabitation and the uneven retreat from marriage in the US, 1950-2010. Working Paper No. 19413, National Bureau of Economic Research, Cambridge, MA. doi:10.3386/w19413

Mastekaasa, A. (1994). Marital status, distress and well-being: An international comparison. Journal of Comparative Family Studies, 25, 183-205.

McRae, S. (2003). Constraints and choices in mothers' employment careers: A consideration of Hakim's preference theory. The British Journal of Sociology, 54, 317-338. doi:10.1111/j. 1468-4446.2003.00317.x

Oppenheimer, V. K. (1994). Women's rising employment and the future of the family in industrial societies. Population and Development Review, 20, 293-342. doi:10.2307/2137521

Oppenheimer, V. K. (1997). Women's employment and the gain to marriage: The specialization and trading model. Annual Review of Sociology, 23, 431-453.

Orloff, A. S. (1993). Gender and the social rights of citizenship: The comparative analysis of gender relations and welfare states. American Sociological Review, 58, 303-328.

Parsons, T. (1949). The social structure of the family. In R. N. Anshen (Ed.), The family: Its function and destiny (pp. 173-201). New York: Harper.

Popenoe, D. (1993). American family decline, 1960-1990: A review and appraisal. Journal of Marriage and the Family, 55, 527-555. doi:10. 2307/353333

Rogers, S. J., & Amato, P. R. (2000). Have changes in gender relations affected marital quality? Social Forces, 79, 731-753. doi:10.1093/sf/79.2.731

Rogers, S. J., & DeBoer, D. D. (2001). Changes in wives' income: Effects on marital happiness, psychological well-being, and the risk of divorce. Journal of Marriage and Family, 63, 458-472. doi:10.1111/j.1741-3737.2001.00458.x

Ryan, J., Hughes, M., & Hawdon, J. (1998). Marital status, general-Life satisfaction and the welfare state. International Journal of Comparative Sociology, 39, 224-236. doi:10.1163/156854 298X00372

Sayer, L. C., & Bianchi, S. M. (2000). Women's economic independence and the probability of divorce: A review and reexamination. Journal of Family Issues, 21, 906-943. doi:10.1177/01925 1300021007005

Schimmack, U., Krause, P., Wagner, G., & Schupp, J. (2010). Stability and change of well-being: An experimentally enhanced latent state-trait error analysis. Social Indicators Research, 95, 19-31. doi:10.1007/s11205-009-9443-8

Schneider, L., & Schimmack, U. (2009). Self-informant agreement in well-being ratings: A meta-analysis. Social Indicators Research, 94, 363-376. doi:10.1007/s11205-009-9440-y

Snijders, T. A. (2005a). Fixed and random effects. In B. Everitt & D. Howell (Eds.), Encyclopedia of statistics in behavioral science (Vol. 2, pp. 664-665). Hoboken, NJ: Wiley. doi:10.1002/0470 013192.bsa234

Snijders, T. A. (2005b). Power and sample size in multilevel modeling. In B. Everitt & D. Howell (Eds.), Encyclopedia of statistics in behavioral science (Vol. 3, pp. 1570-1573). Hoboken, NJ: Wiley. doi:10.1002/0470013192.bsa492

Soons, J. P., & Kalmijn, M. (2009). Is marriage more than cohabitation? Well-being differences in 30 European countries. Journal of Marriage and Family, 71, 1141-1157. doi:10.1111/j.17413737.2009.00660.x

Soons, J. P., Liefbroer, A. C., & Kalmijn, M. (2009). The long-term consequences of relationship formation for subjective well-being. Journal of Marriage and Family, 71, 1254-1270. doi:10. 1111/j.1741-3737.2009.00667.x

Stack, S., & Eshleman, J. R. (1998). Marital status and happiness: A 17-nation study. Journal of Marriage and the Family, 60, 527-536. doi:10.2307/353867

Steptoe, A., & Wardle, J. (2005). Positive affect and biological function in everyday life. Neurobiology of Aging, 26, 108-112. doi:10.1016/j. neurobiolaging.2005.08.016

Stevenson, B., & Wolfers, J. (2006). Bargaining in the shadow of the law: Divorce laws and family distress. The Quarterly Journal of Economics, 121, 267-288. doi:10.1162/003355306776083572

Stevenson, B., & Wolfers, J. (2007). Marriage and divorce: Changes and their driving forces. Journal of Economic Perspectives, 21, 27-52. doi:10.1257/jep.21.2.27

Stutzer, A., & Frey, B. (2006). Does marriage make people happy, or do happy people get married? Journal of Socio-Economics, 35, 326-347. doi:10.1016/j.socec.2005.11.043

Treas, J., van der Lippe, T., & Tai, T.-O. C. (2011). The happy homemaker? Married women's well-being in cross-national perspective. Social Forces, 90, 111-132. doi:10.1093/sf/90.1.111

United Nations. (2013). World fertility data 2012 [Data Set No. POP/DB/Fert/Rev2012]. New York: United Nations Department of Economic and Social Affairs, Population Division.

Urry, H. L., Nitschke, J. B., Dolski, I., Jackson, D. C., Dalton, K. M., Mueller, C. J., ... Davidson, R. J. (2004). Making a life worth living: Neural correlates of well-being. Psychological Science,15,367-372.doi:10.1111/j.0956-7976.2004. 00686.x

Vanassche, S., Swicegood, G., & Matthijs, K. (2012). Marriage and children as a key to happiness? Cross-national differences in the effects of marital status and children on well-being. Journal of Happiness Studies, 14, 501-524. doi:10.1007/ s10902-012-9340-8

Verbakel, E. (2012). Subjective well-being by partnership status and its dependence on the normative climate. European Journal of Population, 28, 205-232. doi:10.1007/s10680-012-9257-2

Waite, L. (2000). Trends in men's and women's well-being in marriage. In L. J. Waite, C. Bachrach, M. Hindin, E. Thomson, & A. Thornton (Eds.), The ties that bind: Perspectives on marriage and cohabitation (pp. 368-392). New York: Aldine de Gruyter.

Winkelmann, L., & Winkelmann, R. (1998). Why are the unemployed so unhappy? Evidence from panel data. Economica, 65, 1-15. doi:10.1111/ 1468-0335.00111

World Values Survey. (2009). World Values Survey 1981-2008, official aggregate (Version 20090901) [Data file]. World Values Survey Association, Alcobendas, Spain.

Zimmermann, A., & Easterlin, R. (2006). Happily ever after? Cohabitation, marriage, divorce, and happiness in Germany. Population and Development Review, 32, 511-528. doi:10.1111/j.17284457.2006.00135.x