Does Fathers’ Involvement in Childcare and Housework Affect Couples’ Relationship Stability? ∗

Objective. Building on previous analysis conducted by Schober (2012), we explore how paternal involvement in different childcare and housework tasks affects the probability of relationship breakdown between parents. Methods. We use logistic regression on the U.K. Millennium Cohort Study to predict parental relationship breakdown from nine months to seven years post-childbirth. Paternal involvement in four childcare and three housework tasks during the ﬁrst year of parenthood, are used as explanatory variables. Results. The amount of time the father spends alone, caring for the baby during the ﬁrst year of parenthood, is associated with the stability of the parental relationship but the effect of involvement in other tasks is moderated by ethnicity and the mother’s employment status. Conclusion. These nonlinear relationships suggest further research is needed to explore the different associations between paternal involvement in childcare and housework and relationship breakdown, which are complex and variable according to different characteristics.

Understanding the conditions that lead to parental divorce or separation is important given the emotional, health, social, and economic costs for couples and their children (e.g., see Brewer and Nandi, 2014;Jones, 2010;Mooney, Oliver, and Smith, 2009). Previous research identifies a complex array of demographic and economic factors that increase the risk of relationship breakdown, beyond the obvious dissatisfaction and unhappiness with the relationship. These include financial hardship (e.g., Dew and Stewart, 2012), having parents who divorced (e.g., Goodman and Greaves, 2010), a low level of education (e.g., Kalmijn, 1999), and cohabiting rather than being married (e.g., Mooney, Oliver, and Smith, 2009). Divorce rates also vary between ethnic groups, influenced perhaps by variations in cultural norms about marriage and divorce. The risk of relationship breakdown is higher for interracial marriages (e.g., Zhang and van Hook, 2009;Aughinbaugh, Robles, and Sun, 2013). This scenario is likely to have become more prevalent in the United Kingdom given the increasing ethnic diversity in family structures (see Office for National Statistics, 2011;Phoenix and Husain, 2007;Kan and Laurie, 2018). The couple's division of unpaid domestic work also impacts on the quality of a relationship. If one partner takes on a disproportionate share of onerous and routine housework (e.g., cleaning, laundry, and cooking), it can lead to dissatisfaction, depression, and divorce (Ruppanner, Branden, and Turenen, 2017;Lively, Steelman, and Powell, 2010;Prince-Cooke, 2006;Baxter, 2000;Glass and Fujimoto, 1994;Hochschild and Machung, 2012). Conversely, involvement in childcare or other tasks that allow the parent to develop a bond with the child, or provide the opportunity to spend time together as a family (Schober, 2012), may have a positive effect on the couple's relationship. Yet the way in which involvement in the "type" of domestic activity affects the parental relationship has rarely been explored.
Empirical evidence shows that paternal involvement in childcare has a positive effect on the quality and stability of the parental relationship (e.g., McBride and Mills, 1993;Pleck and Masciadrelli, 2004;Hohmann-Marriott, 2009;Schober, 2012). However, paternal involvement tends to be treated as a singular phenomenon, and this presupposes in homogeneity in how the "type of involvement" may influence the parental relationship. The implications of engaging in different types of domestic activity are not considered even though this may affect the parental relationship in different ways. A factor in this is gender segregation in the types of domestic tasks men and women do (e.g., Wajcman, 2015;Kan, Sullivan, and Gershuny, 2011;Dermott, 2008;Spain and Bianchi, 1996;Oakley, 1985), which may suggest that paternal involvement in typically masculine or feminine activities has further (positive or negative) implications on the parental relationship. Another factor might be whether the mother-as well as the children-is a direct beneficiary of the involvement (solo childcare by the father and carrying out routine rather than "fun" childcare tasks might be two such areas).
To explore this further, we build on previous analysis conducted by Schober (2012), who found a significant relationship between a composite measure of paternal involvement in childcare and relationship stability. Yet it is not clear, from this analysis, whether (and which) particular childcare activities are important, and whether these associations are moderated by other sociodemographics. We build on Schober's analysis by modeling the probability of relationship breakdown using individual measures of paternal involvement in childcare and in housework using the same data. Our results show that the amount of time the father spends alone caring for the baby during the first year of parenthood has an association with the stability of the parental relationship, and the effect of paternal involvement in other tasks is moderated in complex ways by ethnicity and the mother's employment status.

The Gendered Division of Labor
Over the last 40 years, increasing numbers of women have entered the labor force, and at the same time, men have increased their contributions to household work, albeit not at an equivalent rate: women continue to do at least twice as much unpaid work compared to men across most industrialized countries (see Bianchi et al., 2000;Lyonette and Crompton, 2015;Craig and Mullen, 2011). Parenthood reinforces this traditional division of labor, with most mothers reducing their hours of employment to take on the "second shift" of running a home and looking after the children, while fathers maintain or increase their employment hours to manage the reduction in household earnings and additional expenses (Hochschild and Machung, 2012;Fox, 2009;Fagan and Norman, 2013;Baxter and Western, 1998;LaRossa and LaRossa, 1981;Kan, Sullivan, and Gershuny, 2011). The traditional division of labor is further perpetuated by social policies, gender inequalities in the labor market, and cultural ideologies about parenting roles, which posit preschool children suffer if their mothers work full-time, and "good fathers" are ones who provide for their children (Fox, 2009;Duncan, 2006;Norman, 2017).
An unbalanced division of household work can strain parental relationships and provide the organizational foundation of gender inequality within families (Hochschild and Machung, 2012;Fox, 2009). However, couples do not always fall into gender-specialized roles once they have children. It has become increasingly common for women to combine employment with raising children (e.g., see Fagan and Norman, 2012;Pfau-Effinger, 2012) and for men to be involved at home (e.g., Norman, Elliot, and Fagan, 2014;Fagan and Norman, 2016;Adler and Lenz, 2017), although these decisions are shaped by individual and structural constraints (e.g., see Norman, 2017). Most U.K. evidence on the gendered division of labor focuses on the white majority population even though rising immigration has increased the ethnic diversity of family structures (Kan and Laurie, 2018). Kan and Laurie's (2018) U.K. analysis provides evidence of ethnic variations in household work, highlighting clear intersections between gender, ethnicity, and the domestic division of labor. This may be at least partly shaped by cultural norms about what constitutes domestic work and how it is defined (Sayer and Fine, 2011).
Theory that offers insights into the possible effect on the stability of the parental relationship focuses overwhelmingly on the consequences of mothers engaging in paid work rather than fathers' involvement at home. Economic theory (e.g., Becker, 1991) suggests that couples' joint economic welfare is maximized by the higher earner, usually the man, specializing in employment and the other in domestic work. Thus, in this view, gender specialization (in heterosexual unions) increases the couple's mutual dependence and hence has a positive effect on relationship stability. This means that women's engagement in the labor market poses a threat to the benefits of specialization because it reduces their economic dependency on men and therefore increases the risk of divorce in unhappy relationships (Oppenheimer, 1997;Ono, 1998). Prince-Cooke (2006) speculates that under this model, men's greater involvement at home would threaten the mutual dependence created by specialization, thus increasing the risk of divorce. In contrast, social exchange theory regards the household division of labor to be the result of couple negotiations that reflect their relative power and resources. Divorce or separation represents a "threat" if a fair division of labor is not negotiated. If economically independent women are unable to negotiate what they consider to be a fair division of domestic tasks, they have the means to leave the marriage. Prince-Cooke (2006) deduces that fathers' greater domestic participation under this model would decrease the risk of separation because the "burden" of unpaid work is shared.
Beyond these economically framed arguments, paternal involvement in the primary care of children has been shown to be one of the most important sources of well-being and happiness for the father (e.g., Eggebeen and Knoester, 2001;Wilson and Prior, 2011;Brandth and Kvande, 2017)-traits that should have a positive effect on the stability of a parental relationship. Involvement in childcare has also been shown to enhance men's sense of obligation to the family (e.g., McClain, 2011), and lead to a greater respect for the carework that mothers have done (e.g., Brandth and Kvande, 2017), which should improve the stability of the parental relationship. Paternal involvement may also increase the mother's leisure time, potentially enhancing her personal well-being, sense of work-life balance, and satisfaction with the parental relationship. It may also enable her to engage in paid work outside the home and therefore compete on a more equal footing with her partner (Lyonette and Crompton, 2015), thus posing a challenge to "gender consciousness" and practices (Sullivan, 2004). In keeping with the balance of theoretical arguments, empirical research does consistently find that greater paternal involvement has a positive impact on both the stability and the quality of the parental relationship (e.g., McBride and Mills, 1993;Schober, 2012). However, the types of activities that fathers get involved in, and the potentially variable effects that this may have on the parental relationship, is usually overlooked.

Domestic Tasks and the Importance of Gender
Fathers and mothers often undertake different kinds of domestic activities, and previous studies have indicated that socioeconomic characteristics influence the quantity of paternal childcare time and the way it is used (e.g., Craig, 2006;Dermott, 2008;Fagan and Norman, 2016;Norman, Elliot, and Fagan, 2014). Mothers spend more time (and a greater proportion of their time) than fathers on the more onerous and less flexible tasks, such as routine housework (e.g., cleaning, doing laundry, and cooking) and on the more routine care of children (e.g., feeding, bathing, and changing nappies). Fathers tend to invest more time on the rewarding and enjoyable tasks that involve engaging with children, such as playing and talking, as well as on the more nonroutine household tasks (e.g., DIY and gardening) (Dermott, 2008;Craig, 2006;Sayer, Bianchi, and Robinson, 2004;Oakley, 1985;Spain and Bianchi, 1996;Hochschild and Machung, 2012;Kan, Sullivan, and Gershuny, 2011). Beyond physical household tasks, mothers are also more likely to assume a "parental consciousness" around the mental and emotional labor associated with caring for children (Waltzer, 1998:15), as well as most responsibility for the planning and management of household labor (Coltrane, 2000).
There are gendered connotations attached to the type of task that is performed, which may steer some fathers (and mothers) to take on certain tasks rather than others. The underlying domestic ideology that housework and primary care of children are traditional feminine tasks may push some women to take most responsibility because this is a way of "doing gender" (West and Zimmerman, 1987), providing a marker of feminine identity (Van der Lippe et al., 2017) and a means of expressing love for the family (Bianchi et al., 2000). Men may shy away from typically "feminine" work, opting for the nonroutine tasks because this affirms their masculine status. Empirical evidence suggests that the types of domestic activities that men and women do can have different consequences for the parental relationship. Baxter and Western's (1998) analysis showed that, for most wives, the amount of time spent by husbands was less important than what men actually did. When husbands participated in the more conventional "female" chores (e.g., preparing meals, cleaning the house, or doing the washing), women were more satisfied with their domestic labor arrangements compared to when husbands did not participate in such activities (also see Himsel and Goldberg, 2003). Baxter's (2000) analysis also found that both men's and women's perceptions of fairness in relation to the domestic division of labor were shaped by the distribution of childcare and housework tasks within the household rather than the amount of time spent doing them. She found that women were more likely to report that housework was shared fairly when men were involved in tasks that were typically considered to be "women's work"-such as cleaning, cooking, and laundry. Men's perceptions of fairness were also determined by how much they participated in these "typical" female tasks. This suggests that men's involvement in specific (core) household tasks has, for both parents, a more significant impact on levels of satisfaction and perceptions of fairness than an equal division of overall time on household work. Thus, it may follow that men's involvement in nontraditional types of household tasks creates the conditions for a more stable parental relationship.
To explore the way in which paternal involvement in different childcare and domestic activities affects the parental relationship, we build on earlier analysis by Schober (2012), who uses composite measures of paternal involvement to predict relationship stability. We add to this analysis by exploring whether these associations are moderated by ethnicity in light of the significant intersections that this may have with relationship stability and household work (e.g., see Kan and Laurie, 2018).

Building on Schober's Analysis
Schober (2012) uses four sweeps of the U.K. Millennium Cohort Study (MCS)-a nationally representative survey following a cohort of children born around the year 2000-to explore the association between fathers' childcare involvement and relationship stability. She derives two composite measures of paternal childcare involvement through factor analysis, measuring fathers' absolute childcare involvement and childcare contributions relative to the mother. These measures are used to model the risk of relationship breakdown across three increments of time-(i) sweep one and two (i.e., s1 and s2); (ii) sweep two and three (i.e., s2 and s3); and (iii) sweep three and four (i.e., s3 and s4)-using Cox proportional hazard models. 1 Schober's results show that, controlling for key sociodemographic, employment, attitudinal, and health-related characteristics, the risk of parental separation is reduced when fathers are involved in childcare between s1 and s2, and between s2 and s3. However, some of these associations become nonsignificant once perceptions of relationship quality are controlled for. Schober concludes that paternal involvement is associated with relationship quality, which lowers the risk of relationship breakdown until the child is at least school age.
A core issue with Schober's approach is that the model she employs embodies three levels of theory about paternal involvement. The level 1 theory is that the behaviors captured are all related to a latent factor (of paternal involvement) and this is what the factor analysis tests. The level 2 theory is that this latent factor is associated with an extraneous variable of interest (relationship stability) and is tested by the regression models that Schober runs and on which she bases her conclusions. The level 3 theory is that this latent factor accounts for all of the relationships between the observed variables (paternal involvement) and extraneous variables of interest (relationship stability); this theory is untested. To test the level 3 theory, and to explicitly investigate the ways in which paternal involvement in different types of childcare and housework may affect parental relationship stability, we use individual indicators of paternal involvement in childcare to predict relationship breakdown. We also use a larger sample that does not exclude households that are missing from any sweep of the MCS survey. This is important as subsequent data are available for many cases that suggest such parents are more likely to have experienced relationship breakdown with their partner. In Schober's paper, the proportion of such households is fairly substantial (32 percent between s1 and s2, 12 percent between s2 and s3, and a further 12 percent between s3 and s4), so by excluding these from her analyses, Schober was left with only a small sample of 5,753 households constituting less than half (46 percent) Household Trajectories from Sweep 1 (Nine Months) to Sweep 4 (Seven Years) in the MCS the original sweep one sample she derives. Schober attempts to account for attrition bias through sample design and nonresponse weights but this may not be sufficient given our exploratory analysis of MCS household trajectories suggests that couples have a higher probability of breaking up if they drop out of the survey between s1 and s2 (see Figure 1). This suggests that the characteristics of households that drop out of the survey differ from those that stay in the survey and so should not be ignored. Schober also only includes households in which both parents are interviewed, which filters out an additional 216 households in which the father is interviewed by proxy even though responses about childcare involvement can be imputed from information provided by the mother. In our analysis, we have included these households.

Data and Methods
We use data from the first four sweeps of the MCS. Mothers and their partners were interviewed when the cohort child was aged approximately nine months (2001-2002, sweep one), three years (2003-2004, sweep two), five years (2006, sweep 3), and seven years (2008, sweep 4). The sample was filtered to focus the analysis on married or cohabiting, heterosexual couples in which all fathers responded to the survey when the child was nine months old (n = 13,411-representing 72 percent of the original sweep one MCS sample). When the child reached aged three, there had been 679 parental separations (representing 5 percent of our derived sweep one sample). When the child was aged five, there had been a further 490 parental separations (representing an additional 4 percent of our derived sweep one sample), and by age seven, there had been a further 413 parental separations (representing an additional 3 percent of our derived sweep one sample). This resulted in a total of 1,819 parental separations (14 percent of our original sweep one sample) that occurred between nine months and seven years after the child's birth. During this period, 61 percent (n = 8,219) of the sweep one households from our derived sample definitely remained intact. The remaining 25 percent of households (n = 3,373) (labeled "excluded other" in Figure 1) were excluded from our final sample because they either dropped out of the sample at some stage during the period between sweeps 1 and 4 (22 percent), or it was not clear whether the relationship had ended (3 percent). 2 These relationship trajectories are shown in Figure 1.
We ran a logistic regression model to predict parental relationship breakdown from when the child was aged nine months to seven years old, using the father's involvement in four childcare and three housework tasks during the first year of parenthood as explanatory variables. We focus on this period given previous research finds it is pivotal for shaping parenting behavior when the child is older (e.g., see Fagan and Norman, 2016). The four childcare variables measure fathers' absolute involvement when the cohort child was nine months old, capturing the frequency they performed solo-childcare, changed nappies, fed the baby, and got up in the night. The three variables that measured fathers' contributions to domestic work relative to the mother captured whether fathers did the most, least, or shared the cooking, cleaning, and laundry.
We control for several factors that have been found to correlate with relationship stability-most of which are the same or similar controls used by Schober 3 (for a review of the literature explaining the potential correlations with relationship stability, see Schober, 2012). These were: the mother's employment status during pregnancy (as a proxy for the mother's labor market attachment); the mother's annual earnings; the unemployment of the father; household income; the parents' marital status; the mother's age; the age difference between the parents; the gender of the cohort child; the number of other children in the household; the mother's age when she left full-time education; the mother's gender role attitudes (derived as a sum of two attitude statements that asked if respondents agreed or disagreed that 1. the child suffers if the mother works before he or she starts school and 2. the woman and family are happier when she works); 4 the difference in the gender role attitudes between partners; the mother's attitudes toward divorce; a dummy variable that measured whether the father was present at the birth and binary variables for whether the child was born through caesarean section or had low birth weight, to indicate difficult recovery periods for the family following the birth. We also control for the ethnicity of both parents, which significantly alters the model; therefore, we run some further exploratory analyses to examine the association between ethnicity and relationship breakdown-which is of interest given previous research highlights a significant difference in how couples from different ethnic groups organize their domestic work (e.g., Kan and Laurie, 2018). Our analysis focuses on how the circumstances in the immediate post-birth period (sweep one) affect the accumulated risk of relationship breakdown up to when the child is aged seven (sweep four). Table 1 presents the odds ratios from two logistic regression models that use the paternal involvement variables to predict relationship breakdown between nine months and seven 2 Households were excluded if parents were lone parents because their partner died, their partner was "longterm absent," the reason for them becoming lone parents was not clear, or if the main or partner carer was a grandparent. 3 We did not control for mother's employment status or weekend work because this correlated with mothers' earnings. To explore the effect of excluding these variables, we ran the model three times: the first model controlled for mothers' earnings, the second model controlled for mothers' employment status, and the third controlled for mothers' weekend work. The odds ratios were broadly unchanged. 4 Note we also ran the model controlling for the father's gender role attitudes and the results were unchanged. However, we decided to follow Schober's analysis and present the model that controls for the mother's gender role attitudes. least * The model controls for the mother's annual earnings (also a proxy for employment status), employment status during pregnancy, the unemployment of the father, household income, marital status, the mother's age, the age difference between the parents, the gender of the cohort child, the number of other children in the household, the age when the mother left full-time education, the mother's gender role attitudes, the difference in the gender role attitudes between partners, the mother's attitudes towards divorce, the father's presence at the birth, type of delivery, and the child's birth weight. Model 1 also controls for the ethnicity of both parents. * * The mothers' gender role attitudes were derived from two variables: (i) whether the child suffers if the mother works before the child starts school and (ii) the woman and family are happier when the mother works. Both variables were coded (1) agree/agree strongly; (2) neither agree nor disagree; (3) disagree/disagree strongly. Variable (ii) was reverse coded and both variables were added together to give a "gender role attitude score" of 1 to 6 where 6 reflected an egalitarian gender role attitude. The derived variable was then recoded so that a score of 2 or 3 = traditional; a score of 4 = mixed or neutral; and a score of 5 or 6 = egalitarian. * * * The father's gender role attitudes were derived in the same way as the mother's gender role attitudes. The mother's gender role attitude score was deducted from the father's gender role attitude score and coded as 0 = same attitudes; between -2 and 0 and between 0 and 2 = "moderately different" attitudes; and between -4 to -3 and 3 and 4 = "radically different" attitudes. SOURCE: MCS sweeps one to four (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008). Sample weighted by survey weights for sweep four (DOVWT2). years post-birth. The models are the same except that in Model 1, the control measuring the parents' ethnicity is excluded in order to highlight the moderating effect that this has on paternal involvement and relationship stability. Model 1 shows that relationships are less likely to break down over the longer term if the father looks after the baby alone at least a few times a week during the first year of parenthood, regardless of gender role attitudes, employment status, and all the other variables controlled for in the model. However, the other paternal involvement variables have no significant effect on relationship stability. Model 2 shows the effect of excluding ethnicity as a control. Interestingly, the frequency fathers change the baby's nappy has a strong positive effect on relationship breakdown (and the frequency fathers get up in the night has a smaller, negative effect), which suggests an association between ethnicity and nappy changing. However, further analyses reveal that Indian and Pakistani parents have a lower rate of separation compared to the other ethnic groups and a much lower proportion of Indian and Pakistani fathers change nappies regularly than other groups, meaning that ethnicity is associated with both nappy changing and relationship breakdown (see Table 2). This suggests that the association between nappy changing and relationship breakdown in Model 2 (Table 1) is artifactual because it disappears once we control for ethnicity.

Results
To explore whether the association between paternal involvement and relationship breakdown was mediated by other characteristics, we reran the model with interaction effects. Specialization theory predicts paternal involvement to increase the likelihood of relationship breakdown. We included an interaction of each involvement variable with the mother's gender role attitudes as we speculated this would be most apparent for households in which the mother favored more traditional gender role ideologies. However, these interaction effects were not significant. We also included an interaction between the involvement variables and the mother's employment status given paternal involvement was predicted to decrease the risk of relationship breakdown for households in which mothers were employed. Interestingly, frequent nappy changing had a positive effect on relationship breakdown once we included an interaction between mothers' employment status and nappy changing, although the interaction variable itself was not significant (see Table 3).  Once a day or more × Full-time (35 or more hours per week) 0.4 * Ref: Never/rarely × Mother not employed * The model controls for the mother's employment status during pregnancy, the unemployment of the father, household income, marital status, the mother's age, the age difference between the parents, the gender of the cohort child, the number of other children in the household, the age when the mother left full-time education, the mother's gender role attitudes, the difference in the gender role attitudes between partners, the mother's attitudes towards divorce, the father's presence at the birth, type of delivery, and the child's birth weight. Model 3 also controls for the ethnicity of both parents. 2 The mothers' gender role attitudes were derived from two variables: i) whether the child suffers if the mother works before the child starts school and ii) the woman and family are happier when the mother works. Both variables were coded 1) agree/agree strongly; 2) neither agree nor disagree; 3) disagree/disagree strongly. Variable (ii) was reverse coded and both variables were added together to give a "gender role attitude score" of 1 to 6 where 6 reflected an egalitarian gender role attitude. The derived variable was then recoded so that a score of 2 or 3 = traditional; a score of 4 = mixed or neutral; and a score of 5 or 6 = egalitarian. 3 The father's gender role attitudes were derived in the same way as the mother's gender role attitudes (see 2 above). The mother's gender role attitude score was deducted from the father's gender role attitude score and coded as 0 = same attitudes; between -2 and 0 and between 0 and 2 = "moderately different" attitudes; and between -4 to -3 and 3 and 4 = "radically different" attitudes. SOURCE: MCS sweeps one to four (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008). Sample weighted by survey weights for sweep four (DOVWT2).
This preliminary, exploratory analysis illustrates the complexity of the relationship between paternal involvement and relationship breakdown, which is confounded by other characteristics, such as ethnicity and the mother's employment status. This suggests that the positive association between paternal involvement and the parental relationship shown by Model 1 and Schober's (2012) analysis is, in fact, not so clear cut.

Summary and Discussion
The aim of this article was to explore how paternal involvement in different childcare and housework tasks affects the probability of relationship breakdown. We built on Schober's (2012) work but argued that her approach implied that the composite measure (or latent factor) accounts for all of the relationships between the observed variables of paternal involvement and relationship breakdown. To examine this in detail, we used individual measures of childcare and housework in the first year of parenthood to explore whether paternal involvement in specific tasks had an association with the accumulated, long-term risk of relationship breakdown. We also explored these associations according to particular demographics that were found to be important by previous research, namely, the parent's ethnicity and the mother's employment status.
Schober found a negative association between paternal involvement and relationship breakdown, but our analysis shows that this association is not so definitive. It is paternal solo-childcare rather than overall "involvement" in childcare and housework that appears to have a positive association with relationship stability, and the association between paternal involvement in specific childcare tasks and relationship stability is moderated in complex ways by ethnicity and the mother's employment status. This supports other research that highlights the importance of accounting for ethnic and cultural variations, which are pivotal in shaping behavior yet are often ignored (e.g., Kan and Laurie, 2018). These nonlinear relationships suggest that further research is needed to explore the different associations between paternal involvement in childcare and housework and relationship breakdown, which appear to be complex and variable according to different characteristics.
Although our analysis suggests that the type of task that a father is involved in does not significantly alter the effect on relationship breakdown, it does suggest that the amount of time the father spends alone caring for the baby during the first year of parenthood is associated with the stability of the parental relationship. This complements other research that finds solo-paternal care as particularly important because of the positive effect it has on a father's happiness and well-being and the development of the father-child relationship (e.g., Brandth and Kvande, 2017;Wilson and Prior, 2010). This can also increase the mother's leisure time, which is likely to have a positive effect on her personal well-being, work-life balance, and thus relationship satisfaction. Men who engage more in care-work are also at least partly contributing to the "second shift" of household labor, which makes for a happier, more stable relationship (Hochschild and Machung, 2012), although our analysis shows that men's contributions to housework have no significant effect. However, this may suggest that the effect of fathers engaging in childcare goes beyond the economic gains hypothesized by gender-role specialization and exchange theory. For example, cultural factors, in relation to marriage and gender roles, appear to be important given the effect of paternal involvement on the parental relationship is moderated by the parent's ethnicity. While ethnicity (and the mother's employment status) moderates the effect that paternal involvement has on the relationship in different ways, it is likely that there are other factors interacting with these associations that have not been explored here. For example, do these associations hold for households in which the father is unemployed or works very long hours, and households that have different income levels? Moreover, while the analysis confirms a clear association between paternal solo-childcare and long-term relationship stability, even after controlling for either parents' prior attitudes toward gender roles, we cannot presume causality here. It is possible, for example, that both relationship stability and childcare are associated with some other factor (for example, the father having an agreeable personality type) and, specifically, we cannot infer that greater paternal involvement would necessarily be a remedy for relationship instability. Nevertheless, the finding is important and digging deeper into the causal mechanism is a key area for future research.