Parents often put their own relationship on the back burner to concentrate on their children, but a new study shows that when spouses love each other, children stay in school longer and marry later in life.
Research about how the affection between parents shapes their children’s long-term life outcomes is rare because the data demands are high.
This study uses unique data from families in Nepal to provide new evidence.
The study, co-authored by researchers at the University of Michigan and McGill University in Quebec, was published in the journal Demography.
“In this study, we saw that parents’ emotional connection to each other affects child rearing so much that it shapes their children’s future,” said co-author and U-M Institute for Social
Research researcher William Axinn. “The fact that we found these kinds of things in Nepal moves us step closer to evidence that these things are universal.”
The study uses data from the Chitwan Valley Family Study in Nepal. The survey launched in 1995, and collected information from 151 neighborhoods in the Western Chitwan Valley.
Married couples were interviewed simultaneously but separately, and were asked to assess the level of affection they had for their partner.
The spouses answered “How much do you love your (husband/wife)? Very much, some, a little, or not at all?”
The researchers then followed the children of these parents for 12 years to document their education and marital behaviors.
The researchers found that the children of parents who reported they loved each other either “some” or “very much” stayed in school longer and married later.
“Family isn’t just another institution. It’s not like a school or employer. It is this place where we also have emotions and feelings,” said lead author Sarah Brauner-Otto, director of the Centre on Population Dynamics at McGill University.
“Demonstrating and providing evidence that love, this emotional component of family, also has this long impact on children’s lives is really important for understanding the depth of family influence on children.”
Nepal provides an important backdrop to study how familial relationships shape children’s lives, according to Axinn. Historically, in Nepal, parents arranged their children’s marriage, and divorce was rare.
Since the 1970s, that has been changing, with more couples marrying for love, and divorce still rare, but becoming more common.
Education has also become more widespread since the 1970s. In Nepal, children begin attending school at age 5, and complete secondary school after grade 10, when they can take an exam to earn their “School-Leaving Certificate.”
Fewer than 3% of ever-married women aged 15-49 had earned an SLC in 1996, while nearly a quarter of women earned an SLC in 2016. Thirty-one percent of men earned SLCs in 2011. By 2016, 36.8% of men had.
The researchers found that the children of parents who reported they loved each other either “some” or “very much” stayed in school longer and married later.
The researchers say that their next important question will be to identify why parental love impacts children in this way.
The researchers speculate that when parents love each other, they tend to invest more in their children, leading to children remaining in education longer.
The children’s home environments may also be happier when parents report loving each other, so the children may be less likely to escape into their own marriages.
Children may also view their parents as role models, and take longer to seek similar marriages.
These findings still stood after researchers considered other factors that shape a married couple’s relationship and their children’s transition to adulthood.
These include caste-ethnicity; access to schools; whether the parents had an arranged marriage; the childbearing of the parents; and whether the parents had experience living outside their own families, possibly being influenced by Western ideas of education and courtship.
“The result that these measures of love have independent consequences is also important,” Axinn said. “Love is not irrelevant; variations in parental love do have a consequence.”
A large number of studies have focused on subjective well-being1(see Diener 2000; Helliwell 2003; Kahneman et al. 1999), and many have found that married people have higher subjective well-being (Mikucka 2016; Stutzer and Frey 2006; Waite and Gallagher 2000).
Yet the increase in cohabitation – not just as a prelude to marriage but also as an alternative partnership type and an accepted setting for parenthood (Perelli-Harris et al. 2012) – raises questions as to whether only marriage has beneficial effects.
Given that cohabitation shares many of the same characteristics as marriage – for example, intimacy, emotional and social support, and joint residence – cohabitors may have similar well-being to those who are married (Soons et al. 2009; Zimmerman and Easterlin 2006).
One of the key issues when analyzing the relationship between partnership status and subjective well-being (SWB) is selection. Cross-sectional studies have often recognized the inability to disentangle selection and causality (e.g., Lee and Ono 2012; Soons and Kalmijn 2009).
Longitudinal studies tend to use fixed-effects (FE) models, which focus on within-individual transitions in partnership status, a process usually occurring in younger adulthood and covering a relatively short period of the life course (Kalmijn 2017; Musick and Bumpass 2012; Soons et al. 2009).
These studies did not examine the long-term effects of partnership in midlife, after the majority of individuals have made decisions about marriage and childbearing. Midlife2 – after women’s prime reproductive period, and when being married may influence one’s identity and well-being – is an understudied part of the life course (Lachman 2015).
At this point in life, the initial boost in happiness may have declined (Soons et al. 2009; Zimmermann and Easterlin 2006), and raising children may confound SWB (Balbo and Arpino 2016; Margolis and Myrskylä 2015).
FE models also do not directly consider individuals who do not experience a change in partnership status: that is, those who cohabit and never marry.
Thus, we cannot tell whether certain groups – for example, those living in a coresidential partnership who have a low propensity to marry – would benefit if they married rather than cohabited; or alternatively, whether the benefits to marriage may be more pronounced for those who have a higher propensity to marry.
Given the interest in marriage promotion policies, targeting low-income individuals in countries such as the United Kingdom, it is important to examine whether those unlikely to marry would be happier if they did marry.
To address these selection processes, we use propensity score–weighted regression analysis to pinpoint how early life and/or current conditions influence conditions in midlife.
This method allows us to address baseline bias and differential treatment bias (Morgan and Winship 2014), which occur when the link between marriage and well-being varies across subgroups.
Cross-sectional research has found a “happiness gap” between cohabitation and marriage in most countries, but the size of the gap appears to vary and may be linked to the acceptance and prevalence of cohabitation in a society (Soons and Kalmijn 2009) or gender context and religious norms (Lee and Ono 2012).
However, it is unclear whether the long-term effects of selection in different countries operate similarly, especially because union duration, childbearing experience, and meanings of cohabitation differ across countries (Hiekel et al. 2014; Perelli-Harris et al. 2014).
In addition, the heterogeneity of treatment effects may vary, indicating that marriage has different benefits for different groups, depending on the country. Here we compare the association between partnership type and SWB in the United Kingdom, Australia, Germany, and Norway, which have experienced substantial increases in cohabitation over the past few decades but have different family policies (Perelli-Harris and Sánchez Gassen 2012) and cultural orientations toward marriage (Perelli-Harris et al. 2014).
Each of these contexts leads us to predict a certain relationship between partnership status and SWB.
This study addresses a number of gaps in the literature. First, we provide new insights into selection due to childhood background and current characteristics for the association between union type and SWB.
Second, we examine whether marriage may be especially advantageous for partnered individuals who have a lower or higher propensity to marry.
Third, we analyze the extent to which the relationship between partnership type and SWB varies by country and gender.
More broadly, analyzing how cohabitation differs from marriage for individuals’ well-being will contribute to our understanding of the meaning and consequences of cohabitation as well as the extent to which these meanings differ across contexts.
A large body of research has investigated the beneficial aspects of marriage for well-being (for reviews, see Nelson-Coffey 2018; Waite and Gallagher 2000). These studies posited that married partners benefit from sexual and emotional intimacy, companionship, and daily interaction (Kamp Dush and Amato 2005; Umberson et al. 2010).
Spouses help each other cope with stress by providing social and emotional support. Recognition from a spouse may provide symbolic meaning in life (Umberson et al. 2010). Additionally, sharing a household can lead to economies of scale, and married spouses could profit from a larger friendship and kin network (Ross and Mirowsky 2013; Umberson and Montez 2010). All these mechanisms could enhance SWB.
Nonetheless, given dramatic social change over the past decades, the benefits to marriage may be declining (Liu and Umberson 2008).
A recent study comparing 87 countries found that the life satisfaction advantage of married men compared with unmarried men has waned over the last three decades, suggesting that marriage has become less advantageous (Mikucka 2016).
This decline may be partially due to the increase in cohabitation, especially in high-income countries. Cohabitation may be taking on much of the form and function of marriage (Cherlin 2004), especially as cohabiting unions become longer and involve children. Similar to married couples, cohabiting couples share a household and may benefit from similar intimacy, support, care, and family networks. Normative expectations to marry have become weaker, and the tolerance for nonmarital arrangements has increased in most countries (Treas et al. 2014).
A large body of research, however, has found that cohabitors often differ from married couples. Across countries, cohabitors have lower second birth rates (Perelli-Harris 2014), are less likely to pool incomes (Gray and Evans 2008; Lyngstad et al. 2010), have lower relationship quality (Wiik et al. 2012), and are more likely to dissolve their relationships (Galezewska et al. 2017), even if they have children (Musick and Michelmore 2018). Qualitative research from Europe and Australia has suggested that many still think of cohabitation as a less-committed type of union than marriage and instead oriented toward freedom and independence (Perelli-Harris et al. 2014). Marriage may thus still be desired by most people but more as a cultural ideal or status symbol.
Recently, most scholars have used FE models to examine whether cohabitation is similar to marriage in increasing SWB. This approach allows the testing of set-point theory, which posits that individuals have a baseline level of happiness that cannot be permanently modified by life events, such as union formation. This theory has been tested in a range of settings, and the findings support a positive effect of marriage and cohabitation on SWB (Musick and Bumpass 2012; Soons et al. 2009; Zimmermann and Easterlin 2006), with cohabitation having a weaker effect (Kalmijin 2017). Some studies, however, have questioned set-point theory and found that different model specifications can result in long-term improvements for marriage (Anusic et al. 2014). Overall, however, most studies indicated that, on average, marriage provides a boost to well-being, with cohabitation providing a weaker boost, and individuals return to original happiness levels in the long term.
Prior studies examining the effects of SWB on partnership status have found that individuals receive a minor boost in happiness after moving in with a partner and a larger boost after marriage, although these effects generally wear off as married partners return to their set-point happiness (Kalmijn 2017; Musick and Bumpass 2012; Soons et al. 2009). What these studies cannot show is whether marriage is beneficial to those who are unlikely to marry, and the extent to which any marriage benefits are due to characteristics that select people into marriage. Prior studies have also not specifically focused on the effects of marriage relative to cohabitation in midlife, after most people have married, and the initial honeymoon period of marriage is over. Our study produced some surprising findings that indicate not only differences by country and gender but also differences by the propensity to marry.
First, contrary to prior studies (e.g., Ono and Lee 2012; Soons and Kalmijn 2009), our results indicate that relative to cohabitation, marriage does not automatically provide a boost to SWB in all countries. On average, cohabiting men in the United Kingdom and Norway and women in Germany have levels of SWB that are similar to those of married men in midlife, even without controls. These findings suggest that in some countries, cohabitation may provide benefits similar to those of marriage, such as shared intimacy, pooled resources, and emotional support (Musick and Bumpass 2012; Perelli-Harris and Styrc 2018).
Second, our results show that, on average, marriage does differ from cohabitation for Australian men and Norwegian women. Without any controls for selection, married individuals in these countries have higher levels of SWB than those in cohabiting partnerships. In Australia, these differences disappear after we include controls, indicating that cohabitation is selective of disadvantage, in accordance with prior studies (Evans 2015; Heard 2011). For Norwegian women, however, our entire battery of controls cannot eliminate average differences between cohabitation and marriage. This finding is quite surprising because some have argued that cohabitation and marriage in Norway, and Scandinavian countries in general, are indistinguishable (Heuveline and Timberlake 2004). Yet here, we see persistent differences between the two partnership types. Our models, however, take into account important sociodemographic status and childhood background variables but may be missing key psychological characteristics or other attributes that are associated with marriage. We cannot rule out the possibility that people with certain personality traits or preferences are more likely to marry. Thus, we are reluctant to interpret our results as having a causal effect.
Nonetheless, the findings suggest that marriage is more important in Norway than often assumed. Focus group research found that marriage is associated with romance and love, even if it occurs sometime after childbearing (Lappegård and Noack 2015), or as a capstone later in life (Holland 2017). Although cohabitation may seem to be identical to marriage superficially, marriage may be indicative of a closer partnership on a deeper level. Marriage for women may be symbolic of a more committed loving relationship, and if marriage does not happen by midlife, the lack of marriage might have detrimental effects on SWB. The elimination of marriage effects when we include relationship satisfaction suggests this may be the case. On the one hand, the results may indicate that relationship quality is more important than type of partnership, and cohabiting and married women have similar SWB. On the other hand, relationship satisfaction may instead be a proxy for marriage given that happier couples are generally more likely to marry (Wiik et al. 2009). Then again, the marital contract may in fact improve relationship quality for women. Thus, although we urge caution in interpreting this result one way or the other, it seems to be plausible that marriage, on average, has positive effects for women in Norway.
Third, our results demonstrate the heterogeneity of treatment effects for German men and British and Australian women. Partnered individuals who have a lower propensity to marry based on childhood selection mechanisms (ATC) have lower SWB if they cohabit rather than marry. Those who have a high propensity to marry (ATT), on the other hand, would not receive any benefits from marriage. For German men and Australian women, controlling for selection mechanisms and partnership experiences eliminates differences between cohabitation and marriage, again demonstrating that selection is more consequential for SWB than partnership status.
For British women, however, we find the persistent effect of marriage on SWB for those who are less likely to marry, despite the large number of controls. These results suggest that marriage may provide some benefits for disadvantaged women, as was also found in a study on mental well-being (Perelli-Harris and Styrc 2018). Our findings here, however, indicate that women from disadvantaged backgrounds would be better off if they did marry, but not because they or their partners have low education, poor employment conditions, or low income, which would make them unhappy. Instead, marriage seems to be associated with happiness for other reasons. The findings could be due to unobserved selection mechanisms related to personality, appearance, or other psychological factors that make them less-attractive marriage partners. On the other hand, they may be unhappy because they do not want to marry partners who do not live up to their expectations, or they may disagree about whether to get married, which could have a greater effect on women than men. The mediating effect of relationship quality suggests this may be the case; those who have higher-quality relationships marry and have higher levels of SWB. Qualitative research provides a deeper explanation for this finding: focus group participants from all educational levels in Britain generally agreed that marriage signaled a more committed relationship than cohabitation, but low-educated women stated that although they would like to marry, cohabitation was more common among their peers. For these women, marriage was a low priority compared with other responsibilities such as housing and children, but they nonetheless aspired to have a wedding (Berrington et al. 2015), and perhaps they were unhappier because they could not achieve their goals.
Our study is not without limitations. First, as mentioned, life satisfaction may be endogenous to partnership decisions: happier people may be more likely to marry than cohabit (Luhmann et al. 2013). Although we focus on controlling for selection mechanisms in childhood, before individuals enter into partnerships, our data do not include a direct measure of happiness in childhood or around the time of entrance into partnership. Propensity score–weighted regression is also unable to control for unobserved factors not available in our surveys. Marriage may be selective of other individual characteristics such as personality, emotional control, or attractiveness. This is particularly important for Norwegian and British women, where differences between cohabitation and marriage persist until relationship satisfaction is included in the models. Second, despite our concerted effort to harmonize the variables across our surveys, differences in survey design and variable construction may limit comparability across surveys. Finally, our definition of midlife (ages 38–50) is relatively narrow. But because cohabitation has increased only within the past few decades, the sample size for cohabiting individuals in the older cohorts is too small. Therefore, future research must continue to evaluate these relationships as cohabitation increases throughout midlife.
Despite these limitations, this study provides evidence that the relationship between partnership and SWB is not straightforward and is context-specific. The context, especially the policy context, does not always operate in predictable ways. For example, given that the German government privileges the marital breadwinner model, we would have predicted that married women in Germany would be happier than cohabiting women; however, we find no differences between marriage and cohabitation. This finding suggests that despite policies to encourage marriage, cohabitation in Germany may not be stigmatized; indeed, Treas et al. (2014) found that Germans have a more positive view of living in cohabitation without marriage intentions than those in Great Britain and Australia. Thus, the policy climate may not shape the association between partnership status and SWB as much as changing values and the specific meaning of cohabitation. This also holds true for Norwegian women, who (as discussed earlier) appear happier if they are married, despite a gender-friendly policy regime that stipulates few legal distinctions between cohabitation and marriage (Lappegård and Noack 2015). Regardless of an increasing number of children born in cohabitation and a lack of stigmatization toward cohabitors, Norwegian women still seem to value marriage and the symbolic implication of the wedding. Research on individual countries needs to recognize that context may be shaping the relationship between factors.
Finally, this study not only demonstrates the role of selection in accounting for the association between marriage and SWB but also illustrates how selection is heterogeneous and differs according to the propensity to marry. Such findings can have important policy implications given that those with a low propensity to marry—namely, the disadvantaged—are likely to be targeted by pro-marriage policymakers. At first glance, our findings seem to suggest that in some countries, those least likely to marry would benefit from marriage-promotion policies: if they were to marry, they would be happier. However, for German men and Australian women, the effect of marriage disappears after the inclusion of more controls. For disadvantaged women in the United Kingdom, marriage may matter, especially if it provides legal protection and a sense of security (Barlow 2014; Berrington et al. 2015). The effect in the United Kingdom disappears after a control for relationship satisfaction is included, implying that policymakers could focus on improving relationship quality, possibly through relationship support organizations that provide counseling, although it is also important to recognize that these women may be unhappy because they are unable to find a suitable partner. On the whole, however, our study indicates that especially after selection and relationship satisfaction are taken into account, differences between marriage and cohabitation disappear in all countries. Marriage does not cause higher SWB; instead, cohabitation is a symptom of economic and emotional strain. Thus, our findings imply that in order to increase SWB, policymakers should aim to reduce disadvantages—both in childhood and adulthood—instead of creating incentives to marry.
University of Michigan