Fathers’ Time Off Work After the Birth of a Child and Relationship Dissolution among Socioeconomically Disadvantaged U.S. Families

Abstract

Relationship dissolution is common among socioeconomically disadvantaged parents. This study utilizes longitudinal data from the Fragile Families and Child Wellbeing Study (FFCWS) to assess whether fathers’ time off work after the birth of a child reduces the likelihood of parents dissolving their relationship. We also consider whether the association between fathers’ time off work and relationship dissolution is mediated by fathers’ support of mothers and moderated by union type. Results indicate that the risk of relationship dissolution is lower when fathers take time off work after the birth of a child. Results also suggest that longer periods of time off work (i.e., two or more weeks) are associated with a lower risk of relationship dissolution among married couples, although overall evidence for variations by union type are mixed. Additionally, there is evidence that the association between fathers’ time off work and relationship dissolution is at least partially explained by higher levels of relationship support among fathers who took time off work after the birth of a child. Overall, findings suggest that providing fathers with opportunities to take time off for the birth of a child may help to promote relationship stability among socioeconomically disadvantaged couples in the U.S.

Relationship dissolution is considerably more likely for socioeconomically disadvantaged couples. For example, the likelihood that first marriages among college educated U.S. women will last 20 years is 78%, compared to only 40% for women with a high school education or less (Copen et al. 2012). Rates of marriage are also lower among less educated couples (Allred 2018), and nonmarital unions are much more likely to dissolve than marriages (Manning and Smock 2002; Manning, Smock, and Majumdar 2004).

At a time when maintaining a stable, satisfying relationship depends more than ever on sharing paid and unpaid labor (Carlson, Hanson, and Fitzroy 2016; Carlson, Miller, and Sassler 2018; Frisco and Williams 2003; Schwartz and Gonalons-Pons 2016), socioeconomically disadvantaged couples face substantial barriers to achieving egalitarian relationships. These barriers include fewer job opportunities, a greater likelihood of having a low-paying and inflexible job, and the challenges of arranging and paying for childcare – all of which can lead to family stress, particularly for mothers (Craig and Churchill 2018; Nomaguchi and Johnson 2016). Such stress may lead working women especially to end unequal parenting relationships rather than be primarily responsible for domestic work (McLanahan and Percheski 2008; Pedulla and Thebáud 2015).

To address the problem of relationship dissolution in socioeconomically disadvantaged communities in the U.S., programs have been initiated to promote healthy marriages and responsible fatherhood (e.g., Supporting Healthy Marriage) that provide relationship workshops and family support services to low-income married parents (Hsueh et al. 2012; Lichter, Graefe, and Brown 2003; Wood et al. 2014). Yet, these programs are often unsuccessful because they do not address key external strains (e.g., financial insecurity, resources to balance work and family) that contribute to relationship dissolution (Trail and Karney 2012; Wood et al. 2014). Policies that better address external strains by enabling couples to balance work and family and encouraging equality in relationships, such as parental leave programs, may help to strengthen parental relationships (Huerta et al., 2014; Pragg and Knoester 2017; Rehel 2014). For example, paternity leave allows fathers to maintain their work commitments while also providing time after a birth for parents to learn how to manage family demands, presumably through encouraging greater fathering commitments (Petts and Knoester 2018; Rehel 2014). In doing so, paternity leave may help facilitate a more egalitarian division of labor, fathers’ support of mothers, and greater relationship satisfaction (Almqvist and Duvander 2014; Bünning 2015; Kotsadam and Finseraas 2011; Petts and Knoester 2019; 2020; Pragg and Knoester 2017), which may reduce the likelihood of relationship dissolution (Carlson et al. 2016; Carlson et al. 2018; Schwartz and Gonalons-Pons 2016). Evidence from Nordic countries – and one study using U.S. data – also suggests that paternity leave-taking reduces the likelihood of relationship dissolution (Lappegård et al., 2019; Olah, 2001; Author Citation; Viklund, 2018). However, previous work has not focused on socioeconomically disadvantaged families or considered potential variations by union type, which is important given that paternity leave may be particularly likely to contribute to marital stability given cultural norms in the U.S. that link fatherhood, marriage, and employment (Townsend 2002).

In this study, we examine associations between fathers’ time off work after a birth and the likelihood of relationship dissolution over a 15-year period using prospective data from the Fragile Families and Child Wellbeing Study (FFCWS), which follows families connected to a national U.S. birth cohort from the late 20^th century. These families are disproportionately socioeconomically disadvantaged (defined as families who are disproportionately racial/ethnic minorities and lower SES). We also assess the extent to which fathers’ relationship support may explain these associations. Finally, we consider whether the associations between fathers’ time off work and the likelihood of relationship dissolution may vary by union type (i.e., married, cohabiting, and romantically involved).

This study is novel in many ways. It considers unique questions within the U.S. context and uses data from one of the few national U.S. studies that contains information about fathers’ time off work and family outcomes. We examine the potential of both mediating and moderating effects in the associations between fathers’ time off work and relationship dissolution. Also, the FFCWS’s focus on socioeconomically disadvantaged parents offers insights into whether policies that increase access to time off work after a birth may help to strengthen relationships among disadvantaged couples. Although data limitations prevent us from analyzing paternity leave policies specifically, our findings about the implications of fathers’ time off work after a birth can inform discussions about parental leave policies within the U.S.

BACKGROUND

Socioeconomically Disadvantaged Couples and Relationship Dissolution

Since the last quarter of the 20^th century, dramatic increases in economic inequality in the U.S. have coincided with vast differences in family formation and stability. Although rates of marriage among high-SES individuals remain high, socioeconomically disadvantaged individuals are less likely to marry, more likely to cohabit as an alternative to marriage, and more likely to have children outside of marriage – and with multiple partners – compared to those with high SES (Carlson 2015; Manning and Smock 2002; Upchurch, Lillard, and Panis 2002). Relationship dissolution is also more likely to occur among socioeconomically disadvantaged couples (Bulanda and Brown 2007; Copen et al. 2012). Nonmarital unions are even more likely to dissolve than marriages (Lichter, Qian, and Mellott 2006; Manning et al. 2004). These trends are important because although family members may benefit from the dissolution of violent or high-conflict relationships (Amato 2000), relationship dissolution is generally associated with a range of negative economic, social, and health outcomes (Amato 2000; Lee and McLanahan 2015; Meadows, McLanahan, and Brooks-Gunn 2008; Peterson 1996).

Cherlin (2016) describes the decline of marriage among America’s working class as a function of men’s declining ability to be sole financial providers and women’s increasing ability to support themselves. Indeed, financial pressures increasingly require multiple incomes to make ends meet, especially for disadvantaged parents. These pressures add to the challenges and stresses of meeting family care responsibilities (Damaske 2011). Consequently, socioeconomically disadvantaged parents, and mothers especially, face great difficulties in balancing work and family demands (Craig and Churchill 2018). Childcare is a particular challenge for socioeconomically disadvantaged families given its high costs (Whitehurst 2008). These parents often have few options (e.g., flexible work hours) to manage childcare (Bond and Galinsky 2011), and often rely on an inconsistent patchwork of friends, relatives, and split work shifts—the combination of which is associated with lower relationship quality and a greater likelihood of relationship dissolution (Kalil, Ziol-Guest, and Epstein 2010). Overall, difficulty balancing work and family responsibilities is a primary reason for high rates of relationship instability (Boling 2015).

Taking Time Off Work for the Birth of a Child

The United States is the only high-income country that does not have a national policy that provides paid leave to mothers and/or fathers for the birth of a child. All high-income countries provide paid leave to new mothers, and most also have policies that allow fathers to take paid leave after childbirth (Koslowski et al. 2019; World Bank Group 2018). The only national policy in the U.S. is the Family and Medical Leave Act (FMLA), which provides up to 12 weeks of unpaid leave after childbirth to approximately 60% of the employed workforce who meet eligibility requirements (Koslowski et al., 2019; Klerman, Daley, and Pozniak 2012). Additionally, eight states have passed or implemented paid family leave policies at the state level (National Partnership for Women and Families 2019).¹

Due to the lack of public policies, most workers rely on employers for access to leave. However, only 15% of employers provide at least partially-paid leave to men after having a child, and this number has not changed much since the mid-2000s (Matos, Galinsky, and Bond 2017). Access to leave is especially rare for socioeconomically disadvantaged fathers, who are also less likely to be employed in jobs that allow for flexible schedules, and are more likely to be penalized for requesting leave than advantaged fathers (Petts, Knoester, and Li 2020; Rudman and Mescher 2013; Williams, Blair-Loy, and Berdahl 2013). Yet, most disadvantaged fathers still take a short period of time off (i.e., 2 weeks or less) when they have a child, often using some form of personal or sick time off (Petts et al. 2020; Pragg and Knoester 2017). Even so, these fathers face more barriers to leave-taking than more advantaged fathers in the U.S.

CONCEPTUAL FRAMEWORK

Division of Labor and Parental Relationship Quality

Parental relationship quality frequently declines after having a child, often because mothers come to perceive the division of labor as unfair as they take on a greater share of the domestic work (Cowan et al. 1985; Keizer and Schenk 2012; Twenge et al. 2003). However, couples may experience fewer strains when responsibilities for breadwinning, housework, and childcare are shared more equally. Indeed, more equitable arrangements are associated with greater sexual intimacy, greater relationship satisfaction, and a lower risk of relationship dissolution (Carlson, Fitzroy, and Hanson 2016; Carlson, Miller, Sassler, and Hanson 2016; Frisco and Williams 2003; Schwartz and Gonalons-Pons 2016).

Although most Americans desire egalitarian relationships, they are hard to achieve given the cultural frames, norms, and structures that reinforce traditionally gendered behaviors (Gerson 2010; Ridgeway 2009; Risman 1999; Shu and Meagher 2017). Disadvantaged parents also face barriers due to a lack of workplace flexibility (Ciabattari 2007; Williams 2010). Moreover, if an egalitarian balance cannot be achieved, socioeconomically disadvantaged women are more likely to want to dissolve a relationship and be self-reliant rather than manage all unpaid family work while partnered (Pedulla and Thebáud 2015). Thus, policies that provide socioeconomically disadvantaged parents with flexibility to meet the demands of both work and family life, while also encouraging egalitarian practices, may reduce the risk of relationship dissolution.

Fathers’ Time Off Work After the Birth of a Child and Relationship Dissolution

Contemporary fathering role expectations suggest that men should be engaged, nurturing parents, and most fathers – including nonresident and socioeconomically disadvantaged fathers – state that it is important to be emotionally available and engaged in their child’s life (McGill 2014). Yet, social and economic barriers, such as challenges in fulfilling breadwinning roles, make it difficult for disadvantaged fathers to meet these expectations (Edin and Nelson 2013).

Taking time off work after a child’s birth may help fathers to bond with their child while also preserving their provider status. In doing so, taking time off work after a birth – and especially longer periods of time off – may help fathers better balance work and family responsibilities by providing them time to become comfortable engaging in childcare and housework (Petts and Knoester 2018; Rehel 2014; Tanaka and Waldfogel 2007). Taking time off work may also demonstrate fathers’ commitment to more egalitarian parenting and help parents establish patterns of sharing in domestic labor that fosters feelings of equity (Nomaguchi, Brown, and Leyman 2017; Petts and Knoester 2018; Rehel 2014).

Indeed, fathers who take time off work after a child’s birth, and especially fathers who take longer periods of time off, engage more with children than fathers who do not take time off work (Huerta et al. 2014; Petts and Knoester 2018; Pragg and Knoester 2017). Time off work is also associated with fathers’ increased participation in housework, and mothers’ reports of more satisfying parental relationships (Almqvist and Duvander 2014; Bünning 2015; Kotsadam and Finseraas 2011; Petts and Knoester 2019; 2020). Fathers’ involvement in childcare, and more equitable sharing of domestic labor more generally, is positively associated with relationship satisfaction (Carlson et al. 2016; Hohmann-Marriott 2009; Kalmijn 1999; Keizer and Shenk 2012; McClain and Brown 2017).

Overall, fathers’ time off work following a child’s birth may promote a more equitable division of labor, which may help to strengthen parental relationships. Indeed, evidence from Nordic countries, and one U.S. study, suggests that paternity leave-taking is associated with a lower risk of relationship dissolution (Lappegård et al. 2019; Olah 2001; Author Citation; Viklund 2018). Although fathers take far less time off work in the U.S. compared to fathers in Nordic countries (Petts et al. 2020; Pragg and Knoester 2017), evidence suggests that even short leaves can indicate fathers’ commitments to be engaged in family life and provide benefits to families (Pailhé, Solaz, & Tô, 2018; Author Citation). Thus, we expect:

Hypothesis 1: Fathers’ time off work following a child’s birth and length of time taken off work will be negatively associated with relationship dissolution.

The above arguments suggest that fathers’ time off work after a birth will reduce the likelihood of relationship dissolution at least in part because it may enable fathers to provide additional support to mothers. Indeed, mothers are more likely to view relationships with fathers favorably if fathers take time off work after a birth (Petts and Knoester 2020; Kotsadam and Finseraas 2011), and this is a key reason why paternity leave is believed to reduce the risk of relationship dissolution (Lappegård et al. 2019; Viklund 2018). Demonstrating support to mothers may be particularly beneficial for socioeconomically disadvantaged couples, as they are especially likely to care for their children themselves due to high childcare costs (Ciabattari 2007; Williams 2010). Thus, we expect:

Hypothesis 2: The associations between fathers’ time off work after a birth and relationship dissolution will be at least partially mediated by fathers’ relationship support toward mothers.

Although fathers’ time off work may reduce the likelihood of relationship dissolution, this association may differ by union type, as fathers’ time off work may be more likely to reduce the risk of relationship dissolution for married parents compared to cohabiting or dating parents. Fatherhood norms in the U.S imply that fatherhood, marriage, and employment are part of a “package deal”, and there are benefits associated with achieving the package deal (Townsend 2002). For example, married residential fathers receive an employer wage premium that unmarried and nonresident fathers do not (Killewald 2013). Marital relationships are also more institutionalized and less likely to be dissolved than other unions (Manning et al. 2004; Sassler and Miller 2017). Married fathers in the U.S. are also more likely to take time off work for childbirth (Huerta et al. 2014; Petts et al. 2020). Thus, married fathers who take time off work following a birth may be especially likely to be viewed by mothers as committed to their family, decreasing the likelihood of relationship dissolution among married couples (Schober 2012).

In contrast, the positive impact of fathers’ time off work following a birth may be more limited in cohabiting and dating relationships. Nonmarital unions are less stable than married unions, and nonmarital partners are less likely to pool resources (Addo 2017; Manning et al. 2004). Cohabiting and dating relationships are also less institutionalized, and unwed fathers may be viewed as less committed partners (Sassler and Miller 2017; Townsend 2002). Moreover, due to cultural norms linking marriage and fatherhood in the U.S., unmarried fathers face greater workplace penalties when requesting leave, which may reduce their economic stability (Petts et al. 2020; Williams et al. 2013). Taking time off work following a birth may ultimately increase the risk of dissolution if there are economic penalties associated with taking time off work, and may only minimally improve feelings of equity in dating relationships since economic resources are less likely to be shared. Overall, although taking time off work after a child’s birth may signal unmarried men’s commitments to being engaged partners and co-parents, the lack of institutionalized norms and support for unmarried parents may limit the implications of fathers’ time off work for unmarried parents’ risk of relationship dissolution. Thus, we expect:

Hypothesis 3: Negative associations between fathers’ time off work following a birth and relationship dissolution will be more pronounced for married couples than for cohabiting couples and couples who are romantically involved but not living together.

Selection Effects and Confounding Factors

Although there is reason to believe that fathers’ time off work after a birth may reduce the likelihood of relationship dissolution among socioeconomically disadvantaged parents, there are other plausible explanations if such an empirical association is found. First, reverse causality is possible; fathers in more stable relationships may be more likely to take time off work when a child is born. Indeed, married fathers are more likely to take leave, and longer periods of leave, than unmarried fathers (Huerta et al. 2014; Petts et al. 2020). Second, any association between fathers’ time off work following a birth and relationship dissolution may be due to selection; fathers who are relatively more socioeconomically advantaged (e.g., higher income, more educated, professional occupation) are more likely to have access to leave and are less likely to experience relationship dissolution than fathers who are relatively more socioeconomically disadvantaged (e.g., lower income, lower education, non-professional occupation) (Conger, Conger, and Martin 2010; Huerta et al. 2014; Petts et al. 2020). Additionally, fathers who take time off work may have jobs that generally provide better work-family balance (e.g., paid leave, flexible schedules), which may enable fathers to provide more support to mothers and reduce the likelihood of relationship dissolution. Third, selection into employment may also influence the association between fathers’ time off work and relationship dissolution. By definition, fathers can only take time off work if they are employed. Thus, a focus on employed fathers (i.e., those who are eligible to take time off work) may create selection concerns by excluding a group of fathers who may be especially likely to experience relationship dissolution (Jensen and Smith 1990; Jalovaara 2003).

Although we are unable to fully address concerns of selection, we attempt to minimize the likelihood of these alternative explanations through the use of inverse probability of treatment weighting (IPTW), which allows us to account for observed selection factors in our analysis (Austin 2016). We account for key factors such as occupation type, income, education, and union type at the time the child was born in an attempt to minimize selection on these factors. We also account for parental relationship quality at the time of the child’s birth to minimize the likelihood of reverse causality. Unfortunately, given the nature of our research question, we are unable to fully account for selection effects due to employment. However, the vast majority of fathers in these data are employed, and our focus on how time off work for the birth of a child is associated with relationship dissolution is not relevant to unemployed fathers.²

In addition to selection, there may also be factors that confound the association between fathers’ time off work following a birth and relationship dissolution. Following previous work, we account for the possibility that the association between fathers’ time off work and relationship dissolution may be affected by age, education, income, work characteristics, and gender attitudes (Lappegård et al. 2019; Olah 2001; Viklund 2018). We also control for race/ethnicity given the racial/ethnic differences in rates of leave-taking and union dissolution in the U.S. (Osborne, Manning, and Smock 2007; Petts et al. 2020). We consider relationship duration, marital history, and religious participation as well, as these factors are associated with family commitments as well as relationship dissolution (Berger et al. 2008; Lyngstad & Jalovaara 2010; Petts 2007).

DATA AND METHODS

Data

Data is taken from the Fragile Families and Child Wellbeing Study (FFCWS).³ The FFCWS is a longitudinal birth cohort study that follows the families of 4,898 children who were born in the U. S. between 1998 and 2000. These data consist of urban parents who are disproportionately low-income, unmarried (although married parents are also included), and racial/ethnic minorities. Although these data present some challenges particularly regarding selection, it is one of the few national datasets to contain information on fathers’ leave-taking and the only one to focus on disadvantaged parents specifically. These data have also been used in studies that have considered other consequences of fathers’ leave-taking (Knoester, Petts, and Pragg 2019; Petts and Knoester 2020; Pragg and Knoester 2017). We use all available data to assess the long-term association between fathers’ time off work and relationship dissolution in this study: parents were interviewed shortly after birth (W1) and when children were approximately one (W2), three (W3), five (W4), nine (W5), and fifteen years old (W6).

The sample is restricted to families in which couples were romantically involved at W1 (N = 4,224) and fathers participated in both the W1 and W2 surveys to obtain information about fathers’ time off work following a birth (N = 2,937). The sample is further restricted to families in which fathers were employed at the time of their child’s birth (N = 2,498) and reported working after the birth (N = 2,419). We also omit couples in which the mother does not participate in a survey after W1 and couples in which fathers did not answer the questions about time off work after a birth, resulting in a sample of 2,036 couples (8,968 couple-years).

Fathers’ Time Off Work After a Birth

In W2 of the FFCWS, fathers reported on whether they took any time off work due to the birth of the focal child and how many weeks of leave, paid or unpaid, they took.⁴ Although these questions do not allow us to know whether fathers took paternity leave specifically (as opposed to using sick time, vacation time, etc.), they do indicate whether fathers took time off work specifically for the birth of their child, which is how paternity leave is typically defined in the U.S. and elsewhere (Koslowski et al. 2019). These questions were used to construct two indicators. Time off work indicates whether fathers took time off work for the birth of their child (1 = yes). Length of time off work is a categorical variable indicating whether fathers took no time off (reference group), one week, or 2 or more weeks off. These categories are informed by research on paternity leave in the U.S., which suggests that two weeks off work is often key for promoting positive outcomes for American families (Nepomnyaschy and Waldfogel 2007; Petts and Knoester 2018).⁵

Relationship Dissolution

The primary dependent variable in this study is relationship dissolution. Relationships are considered dissolved if mothers report no longer being romantically involved with fathers. This includes both formal (i.e., divorce) and informal types of dissolution (e.g., separation, moving out, and reporting no romantic involvement). Couples are at risk of dissolution starting at W1 until either the relationship ended or the couple was censored. Relationship dissolution is treated as a discrete event based on mothers’ reported relationship status at each survey wave (couples are treated as dissolved in the wave that the mother reports no longer being romantically involved with the father), and couples are right-censored if they dropped out of the survey or mothers reported still being romantically involved with fathers at W6.

Union Type

Union type is taken from mothers’ reports at W1 of being: (a) married (used as reference group), (b) cohabiting, or (c) in a romantic relationship, but not residing with, the child’s father.⁶

Relationship Support

Relationship support is a time-varying⁷ indicator taken from mothers’ mean responses to questions about how often fathers (1 = never to 3 = often): (a) are fair and willing to compromise when you have a disagreement, (b) express affection or love, (c) insult or criticize you or your ideas (reverse coded), and (d) encourage or help you do things that are important (α > .64).⁸

Control Variables

A number of variables are included as controls. Time-varying indicators of mothers’ and fathers’ income are measured in logged dollars. Race/ethnicity is coded as (a) both parents are white (used as reference group), (b) both parents are black, (c) both parents are Latino, (d) both parents report other race/ethnicity, and (e) each parent reports a different race/ethnicity. Time-varying indicators of parents’ work hours are categorized as (a) does not work (this is included for fathers to allow for this possibility in later waves), (b) part-time (less than 35 hours a week), (c) full-time (35–44 hours a week, used as reference category), and (d) more than full-time (45 hours a week or more). For mothers, the categories of works full-time and more than full-time are combined. Time-varying indicators of fathers’ occupation type are coded as (a) professional (used as reference category), (b) labor, (c) service, (d) sales, or (e) other occupation. We also include time-varying indicators of each parent’s age, number of other children, and whether each parent attended religious services at least a few times a month (1 = yes), and time-invariant indicators of whether each parent attended at least some college (1 = yes), whether either parent was previously married, how long the couple had been together prior to W1 (in years), whether the focal child is the father’s first child, and how much time mothers took off work for the birth of the focal child (in months, taken from W2).

We also include W1 indicators of parenting role attitudes. Positive father attitudes are indicated by fathers’ mean level of agreement (1 = strongly disagree to 4 = strongly agree) on whether (a) being a father and raising children is one of the most fulfilling experiences for a man, (b) I want people to know that I have a new child, and (c) not being a part of my child’s life would be one of the worst things that could happen to me (α = .73). Both parents also indicated the fathering role they believed was most important (options: provide regular financial support, teach child about life, provide direct care, show love and affection to the child, provide protection for the child, or serve as an authority figure and discipline the child), and separate variables indicate whether each parent identified either providing direct care or showing love and affection as most important (1 = yes). We also include separate measures of traditional gender attitudes for each parent, which indicate whether parents agree that it is much better for everyone if the man earns the main living and the woman takes care of the home and family (1 = yes).

Analytic Strategy

Discrete-time logistic event history models are used to assess whether fathers’ time off work and length of time off are associated with union dissolution, accounting for potential confounders. We also consider whether the association between time off work and relationship dissolution is mediated by relationship support. Formal tests of mediation are conducted using the KHB method, which decomposes the overall effect into direct and indirect effects using the same scale, and is appropriate for nonlinear models (Breen, Karlson, & Holm, 2013; Kohler, Karlson, & Holm, 2011). To assess variations by union type, we include interaction terms between fathers’ time off work and union type.⁹ Because the direction and significance of interaction terms in nonlinear regression models are not always accurate (Long and Freese 2014; Mize 2019), we follow the guidelines put forth by Mize (2019) to correctly interpret nonlinear interaction effects. Indicators of each survey wave were included to estimate the baseline hazard (with W6 used as the reference year). For all analyses, continuous control variables are mean centered, and multiple imputation from ten imputed models is used to account for missing data.¹⁰

Selection

Given variations in access to leave and high rates of relationship dissolution among socioeconomically disadvantaged populations, selection is an important concern. Although we cannot account for selection effects due to unobservable factors (e.g., motivation to be a good father and/or husband), we attempt to minimize selection concerns in our multivariate models by using inverse probability of treatment weighting (IPTW). IPTW uses propensity score matching to generate weights (weights are equal to the inverse of the probability of receiving the treatment) that are included in regression models to account for variations between the treatment (i.e., took time off work) and control (i.e., did not take time off) groups, such that these groups differ in whether they received the treatment but are similar on all other baseline characteristics (Austin 2016). To calculate the weights, we used a logistic regression model to predict ever experiencing dissolution based on W1 control variables and W1 relationship support, estimated propensity scores, and matched respondents in the treatment and control groups with the closest propensity scores. Diagnostic analyses were conducted to assess the propensity score model, which included ensuring balance was achieved and that the assumption for common support (i.e., propensity scores overlap between the treatment and control groups) was met. To minimize the possibility that large propensity scores may inflate the value of the treatment effect, weights were trimmed such that all weights above the 90^th percentile were trimmed to equal the 90^th percentile (Lee, Lessler, and Stuart 2011). Thus, to the extent we are able, this approach enables us to incorporate the potential for selection effects due to occupation type, relationship support at birth, union type at birth, and other key sociodemographic characteristics known to be related to both leave-taking and relationship dissolution.

RESULTS

Mean values for all variables are presented in Table 1. Overall means in the first column of Table 1 reflect the socioeconomically disadvantaged nature of the sample; 59% of fathers did not attend college, mean income for fathers is $30,000 (and only 14% of fathers earned above the median household income in the year 2000), and approximately half of fathers are in labor occupations. Despite this, results in Table 1 show that 80% of fathers took time off work after the birth of their child. Of the fathers who took time off work, most took one week (73%). Results in Table 1 also suggest that fathers who took time off work are more socioeconomically advantaged (e.g., higher income, higher education) than fathers who did not take time off work.

Table 1.

Summary Statistics (W1)

	Full sample		Did not take time off		Took time off
Variable	M	SD	M	SD	M	SD
Relationship Dissolution	0.52	-	0.66	-	0.48***	-
Time Off Work
Time Off Work	0.80	-	0.00	-	1.00***	-
Length of Time Off Work
No Time Offa	0.20	-	1.00	-	0.00***
One Week	0.59	-	0.00	-	0.73***
Two or More Weeks	0.21	-	0.00	-	0.27***
Union Type
Marrieda	0.36	-	0.20	-	0.40***	-
Cohabiting	0.42	-	0.44	-	0.42	-
Romantic	0.22	-	0.36	-	0.18***	-
Relationship Support	2.78	0.27	2.72	0.29	2.79***	0.25
Control Variables
Father’s Income (logged)	9.96	1.40	9.68	1.56	10.03***	1.35
Mother’s Income (logged)	6.23	4.68	6.15	4.62	6.24	4.70
Father Age	28.31	6.93	27.39	6.80	28.52**	6.94
Mother Age	25.88	6.13	24.53	5.68	26.21***	6.19
Both Whitea	0.24	-	0.10	-	0.28****	-
Both Black	0.37	-	0.53	-	0.33***	-
Both Latino	0.22	-	0.19	-	0.22	-
Both Other Race	0.02	-	0.01	-	0.02	-
Mixed Race	0.15	-	0.17	-	0.15	-
Attended College (Father)	0.41	-	0.29	-	0.45***	-
Professional Occupationa	0.17	-	0.11	-	0.18***	-
Labor Occupation	0.49	-	0.54	-	0.48*	-
Sales Occupation	0.08	-	0.08	-	0.09	-
Service Occupation	0.24	-	0.26	-	0.23	-
Other Occupation	0.02	-	0.01	-	0.02	-
Works Part-Time (Father)	0.11	-	0.18	-	0.10***	-
Works Full-Time (Father)a	0.47	-	0.47	-	0.47	-
Works Overtime (Father)	0.41	-	0.35	-	0.43**	-
Attended College (Mother)	0.45	-	0.35	-	0.47***	-
Works Part-Time (Mother)	0.21	-	0.19	-	0.21	-
Works Full-Time (Mother)a	0.44	-	0.45	-	0.44	-
Not Employed (Mother)	0.35	-	0.36	-	0.35	-
Monthly Religious Participation (Father)	0.32	-	0.29	-	0.33	-
Monthly Religious Participation (Mother)	0.40	-	0.39	-	0.40	-
First Child	0.41	-	0.36	-	0.43***	-
Number of Other Children	1.02	1.18	1.26	1.35	0.96***	1.12
Relationship Duration	2.44	2.93	1.95	2.67	2.57***	2.99
Previously Married	0.21	-	0.21	-	0.21	-
Length of Maternity Leave	2.46	3.13	2.42	3.14	2.47	3.13
Positive Father Attitudes	3.77	0.39	3.74	0.39	3.78	0.39
Traditional Gender Attitudes (Father)	0.39	-	0.39	-	0.39	-
Traditional Gender Attitudes (Mother)	0.30	-	0.31	-	0.30	-
Engaged Father Attitudes (Father)	0.66	-	0.62	-	0.67	-
Engaged Father Attitudes (Mother)	0.74	-	0.72	-	0.75	-
N	2046		401		1635

^aUsed as reference category. Significant differences determined by two-tailed t-tests

^*p < .05.

^**p < .01.

^***p < .001.

Results from discrete-time logistic regression models are presented in Tables 2 and 3. Looking first at the association between fathers’ time off work and relationship dissolution in Table 2, results from zero-order models support hypothesis 1 by showing that couples in which fathers take time off work have 34% lower odds of dissolving their relationship compared to couples in which fathers do not take time off (Model 1). Supplementary analyses suggest that the association between taking time off work and relationship dissolution does not vary over time (i.e., although the risk of dissolution differs across waves, interactions between time off work and survey wave were not statistically significant). The negative association between fathers’ time off work and relationship dissolution persists when controls are included in Model 2, although the effect size is reduced. The association is further reduced (i.e., odds ratio changes from .81 to .83) when relationship support is included in Model 3. Formal tests of mediation suggest that the indirect effect of relationship support is statistically significant (b = −0.03, p < .001) and explains approximately 17% of the relationship between fathers’ time off work and relationship dissolution.¹¹

Table 2.

Results from Discrete-Time Logistic Event History Models Predicting the Association Between Fathers’ Time Off Work and Relationship Dissolution

	1		2		3		4
Variable	OR	SE	OR	SE	OR	SE	OR	SE
Time Off Work	0.66***	0.05	0.81*	0.07	0.83*	0.08	0.59*	0.12
Union Type
Cohabiting			1.84***	0.24	1.78***	0.24	1.48	0.33
Romantic			2.86***	0.42	2.72***	0.40	2.23**	0.51
Relationship Support					0.47***	0.06
Control Variables
Father’s Income (logged)			0.96	0.04	0.95	0.04	0.96	0.04
Mother’s Income (logged)			0.97	0.09	0.98	0.09	0.99	0.09
Father Age			0.99	0.01	0.99	0.01	0.99	0.01
Mother Age			0.95***	0.01	0.95***	0.01	0.95***	0.01
Both Black			1.30*	0.17	1.25	0.16	1.28	0.16
Both Latino			0.79	0.11	0.77	0.11	0.78	0.11
Both Other Race			0.47	0.20	0.45	0.19	0.48	0.19
Mixed Race			1.34*	0.19	1.31	0.19	1.33*	0.19
Attended College (Father)			0.96	0.10	0.96	0.10	0.96	0.10
Labor Occupation			1.20	0.17	1.18	0.17	1.18	0.17
Sales Occupation			1.18	0.24	1.18	0.24	1.15	0.24
Service Occupation			1.32	0.21	1.32	0.21	1.30	0.21
Other Occupation			0.58	0.30	0.62	0.31	0.60	0.32
Works Part-Time (Father)			0.95	0.14	0.91	0.13	0.98	0.13
Works Overtime (Father)			1.00	0.10	1.01	0.10	1.00	0.10
Not Employed (Father)			0.94	0.39	0.90	0.37	0.93	0.38
Attended College (Mother)			0.96	0.10	0.95	0.10	0.96	0.10
Works Part-Time (Mother)			0.97	0.12	0.96	0.12	0.98	0.12
Not Employed (Mother)			0.73	0.66	0.80	0.71	0.87	0.79
Monthly Religious Participation (Father)			1.01	0.10	1.04	0.10	1.02	0.10
Monthly Religious Participation (Mother)			0.82*	0.08	0.85	0.08	0.83	0.08
First Child			1.02	0.11	1.03	0.10	1.02	0.12
Number of Other Children			0.99	0.03	0.98	0.03	0.99	0.03
Relationship Duration			1.02	0.02	1.01	0.02	1.02	0.02
Previously Married			1.28*	0.15	1.30*	0.15	1.28*	0.15
Length of Maternity Leave			1.03	0.01	1.03*	0.01	1.02	0.01
Positive Father Attitudes			1.05	0.11	1.06	0.11	1.04	0.11
Traditional Gender Attitudes (Father)			0.94	0.08	0.93	0.09	0.92	0.08
Traditional Gender Attitudes (Mother)			1.04	0.10	1.00	0.10	1.04	0.10
Engaged Father Attitudes (Father)			0.87	0.08	0.87	0.08	0.87	0.08
Engaged Father Attitudes (Mother)			0.74**	0.07	0.75**	0.07	0.74**	0.07
Survey Wave
Wave 2	0.87	0.12	0.33***	0.06	0.36***	0.07	0.34***	0.07
Wave 3	1.07	0.15	0.49***	0.09	0.48***	0.09	0.50***	0.09
Wave 4 (5 years old)	0.34***	0.06	0.18***	0.04	0.17***	0.03	0.18***	0.03
Wave 5 (9 years old)	2.17***	0.31	1.55**	0.25	1.54**	0.25	1.55**	0.25
Interactions
Time Off Work x Cohabiting							1.44	0.34
Time Off Work x Romantic							1.53	0.38

N = 2,036 (8,150 person-years); Models are weighted using IPTW with trimmed weights;

^*p < .05.

^**p < .01.

^***p < .001.

Table 3.

Results from Discrete-Time Logistic Event History Models Predicting the Association Between Fathers’ Length of Time Off Work and Relationship Dissolution

	1		2		3		4
Variable	OR	SE	OR	SE	OR	SE	OR	SE
Length of Time Off Work
One Week	0.71***	0.06	0.83*	0.08	0.84	0.08	0.70	0.15
Two or More Weeks	0.53***	0.06	0.75*	0.09	0.77*	0.09	0.37***	0.10
Union Type
Cohabiting			1.84***	0.24	1.78***	0.23	1.48	0.33
Romantic			2.85***	0.42	2.72***	0.40	2.24**	0.51
Relationship Support					0.46***	0.06
Control Variables
Father’s Income (logged)			0.95	0.04	0.95	0.04	0.96	0.04
Mother’s Income (logged)			0.99	0.09	0.99	0.09	0.99	0.09
Father Age			0.99	0.01	0.99	0.01	0.99	0.01
Mother Age			0.95***	0.01	0.95***	0.01	0.96***	0.01
Both Black			1.29*	0.17	1.24	0.16	1.28	0.16
Both Latino			0.78	0.11	0.76	0.11	0.78	0.11
Both Other Race			0.47	0.20	0.45	0.19	0.49	0.20
Mixed Race			1.34*	0.19	1.30	0.19	1.32	0.19
Attended College (Father)			0.97	0.10	0.96	0.10	0.96	0.10
Labor Occupation			1.22	0.17	1.19	0.17	1.19	0.17
Sales Occupation			1.20	0.24	1.20	0.24	1.13	0.24
Service Occupation			1.35	0.21	1.34	0.21	1.31	0.21
Other Occupation			0.57	0.30	0.60	0.32	0.64	0.34
Works Part-Time (Father)			0.96	0.14	0.93	0.13	0.99	0.13
Works Overtime (Father)			1.00	0.10	1.00	0.10	1.00	0.10
Not Employed (Father)			0.92	0.39	0.88	0.37	0.92	0.38
Attended College (Mother)			0.96	0.10	0.95	0.10	0.97	0.10
Works Part-Time (Mother)			0.96	0.12	0.96	0.12	0.99	0.13
Not Employed (Mother)			0.87	0.67	0.93	0.72	0.90	0.81
Monthly Religious Participation (Father)			1.02	0.10	1.04	0.10	1.01	0.10
Monthly Religious Participation (Mother)			0.82*	0.08	0.85	0.08	0.83*	0.08
First Child			1.02	0.11	1.06	0.12	1.02	0.11
Number of Other Children			0.99	0.03	0.98	0.03	0.99	0.03
Relationship Duration			1.02	0.02	1.01	0.02	1.01	0.02
Previously Married			1.30*	0.15	1.30*	0.15	1.29*	0.15
Length of Maternity Leave			1.03	0.01	1.03*	0.01	1.02	0.01
Positive Father Attitudes			1.05	0.11	1.07	0.11	1.05	0.11
Traditional Gender Attitudes (Father)			0.94	0.08	0.93	0.09	0.92	0.08
Traditional Gender Attitudes (Mother)			1.04	0.10	1.00	0.10	1.03	0.10
Engaged Father Attitudes (Father)			0.87	0.08	0.88	0.08	0.87	0.08
Engaged Father Attitudes (Mother)			0.74**	0.07	0.75**	0.07	0.74**	0.07
Survey Wave
Wave 2	0.86	0.12	0.33***	0.06	0.36***	0.07	0.34***	0.06
Wave 3	1.06	0.15	0.49***	0.09	0.47***	0.09	0.49***	0.09
Wave 4 (5 years old)	0.34***	0.06	0.18***	0.04	0.17***	0.03	0.18***	0.03
Wave 5 (9 years old)	2.16***	0.31	1.54**	0.25	1.54**	0.25	1.55**	0.25
Interactions
One Week x Cohabiting							1.21	0.29
Two or More Weeks x Cohabiting							2.29**	0.70
One Week x Romantic							1.21	0.31
Two or More Weeks x Romantic							3.09**	1.04

N = 2,036 (8,968 person-years); Models are weighted using IPTW with trimmed weights;

^*p < .05.

^**p < .01.

^***p < .001.

Although results in Model 4 of Table 2 suggest that the relationship between paternity leave-taking and relationship dissolution does not vary by union type, interaction coefficients are not always accurate in nonlinear regression models (Long and Freese 2014; Mize 2019). Thus, we estimate predicted probabilities of relationship dissolution by fathers’ time off work and union type. Results are presented in Figure 1 (and detailed in Table A2). As shown in Figure 1, the predicted probability of dissolving a relationship between waves is lower for married couples in which fathers took time off work (P = .09) compared to married couples in which fathers did not take time off (P = .14; p < .05). In contrast, predicted probabilities of dissolution do not significantly differ by whether fathers took time off work for cohabiting or romantically involved couples. Additional analyses show that the average marginal effect of time off work on reducing the risk of relationship dissolution (Table A2) is 2.5 times higher for married couples (0.05) compared to unmarried couples (0.02). Yet, there is no significant difference in the apparent effect of taking time off work on relationship dissolution across union type (based on second differences – see Table A2). Thus, fathers’ time off work after a birth is generally associated with a lower likelihood of relationship dissolution across all union types, although we find some evidence that this association is more pronounced among married couples.

Figure 1:

Predicted Probabilities of Relationship Dissolution by Union Type and Fathers’ Time Off Work

Note: Estimates are taken from Model 4 of Table 2

Results focused on the association between length of time off work and relationship dissolution are presented in Table 3. As shown in Model 1, results from zero-order models suggest that relationship dissolution is less likely to occur when fathers take any length of time off work compared to when fathers do not take time off. When controls are included in Model 2, results support the first hypothesis by showing that taking one (OR = 0.83, p < .05) or two or more (OR = 0.75, p < .05) weeks off work is associated with a lower risk of dissolution compared to when fathers do not take time off.¹² However, Wald tests show that the associations between taking one or two or more weeks off work and relationship dissolution are not significantly different from each other (p = .29). In addition, results in Model 3 provide support for the second hypothesis; when relationship support is included in the model, the association between taking one week off and a lower likelihood of relationship dissolution is no longer statistically significant and the effect size for the association between taking two or more weeks off and a lower likelihood of relationship dissolution is reduced (again, Wald tests show no significant difference between taking one or two or more weeks off work). Formal mediation tests show that relationship support partially mediates these associations. Approximately 17% of the association between taking one week off and relationship dissolution (b = −.03, p < .05), and approximately 13% of the association between taking two or more weeks off and relationship dissolution (b = −.04, p < .01), is explained by relationship support.

Interaction terms are included in Model 4 of Table 3 to assess whether the association between length of time off work and relationship dissolution is moderated by union type. Predicted probabilities of relationship dissolution based on these estimates are presented in Figure 2. As shown in Figure 2, married couples in which fathers took two or more weeks off leave (P = .06) had a lower probability of experiencing dissolution between waves than married couples in which fathers did not take time off (P = .14; p < .01) and married couples where fathers took one week off (P = .10, p < .001). In contrast, predicted probabilities of dissolution do not appear to differ by length of time off work for cohabiting or romantically involved couples. Tests of second differences show that the negative effect of taking two or more weeks off work on relationship dissolution is generally stronger and more consistent for married couples compared to cohabiting or romantically involved couples (based on second differences – see Table A2).¹³ Thus, longer periods of time off work for fathers are associated with a lower likelihood of relationship dissolution for married couples than for unmarried couples.

Figure 2:

Predicted Probabilities of Relationship Dissolution by Union Type and Fathers’ Length of Time Off Work

Note: Estimates are taken from Model 4 of Table 3

DISCUSSION

Low rates of marriage, high rates of nonmarital childbearing, and high rates of relationship dissolution in socioeconomically disadvantaged U.S. communities have led scholars and policymakers to consider strategies to increase relationship stability among low-SES couples (Lichter et al. 2003; McLanahan and Percheski 2008; Wood et al. 2014). Equal sharing of paid and unpaid labor among parents has become increasingly important for maintaining stable, satisfying relationships (Carlson et al. 2016; Carlson et al. 2018; Schwartz and Gonalons-Pons 2016), yet maintaining work-family balance is particularly difficult for socioeconomically disadvantaged couples (Ciabattari 2007; Gupta 2007; Williams 2010). Many OECD nations have addressed issues related to family gender inequality by implementing public policies that allow fathers to take leave, as fathers’ leave-taking may encourage both father involvement and mothers’ attachment to the labor force (Boling 2015; World Bank Group 2018). Indeed, there is evidence that these policies encourage fathers to take leave (Escot and Fernández-Cornejo 2014; Mayer and Le Bourdais 2019). Yet, the United States government and U.S. businesses remain unique by failing to provide widespread access to parental leave, particularly for socioeconomically disadvantaged fathers (Petts et al. 2020; Klerman et al. 2012).

Building off of evidence from Nordic countries showing that paternity leave may reduce the likelihood of relationship dissolution (Lappegård et al., 2019; Olah, 2001; Viklund, 2018), the goal of this study was to consider whether fathers’ time off work following a birth is negatively associated with relationship dissolution in the U.S. Our results suggest that taking time off work is associated with a lower likelihood of relationship dissolution among socioeconomically disadvantaged parents. However, we only find consistent evidence of statistically significant associations for married parents.

First, we found evidence in support of our first hypothesis, which anticipated that fathers’ time off work after a birth would be negatively associated with relationship dissolution. Results indicated that parents were more likely to stay together if fathers took time off work following a birth. Taking time off work may signal fathers’ commitments to be more engaged parents and partners (Pragg and Knoester 2017; Rehel 2014). Demonstrating a commitment to be involved may increase feelings of equity, reducing the risk of relationship dissolution (Frisco and Williams 2003; Rehel 2014). Taking time off work may also enable fathers to become more comfortable and confident in parenting tasks, further strengthening parental relationships (Petts and Knoester 2020; Rehel 2014). Even though these periods of time off work are short by international standards, they may enable fathers to abide by the American cultural norms of being both a provider and an engaged parent, helping to improve relationship stability.

Second, we found evidence to support our second hypothesis that the associations between fathers’ time off work after a birth and relationship dissolution were partially mediated by fathers’ relationship support. Feeling supported may help reduce family strains as mothers may feel as though they can rely on fathers to offer their fair share of family contributions (Nomaguchi et al. 2017). Overall, fathers who take time off work following a birth may be more likely to shoulder more responsibility for unpaid labor, provide support to mothers, and increase feelings of equity within relationships, reducing the risk of relationship dissolution.

Finally, we find some evidence to support the third hypothesis. Specifically, we find evidence from interaction effects showing that married couples seem to be particularly less likely to experience relationship dissolution if fathers take time off work. Effect sizes also suggest that there are meaningful differences in the associations between fathers’ time off work following a birth for married couples compared to other couples. Nonetheless, the associations between relationship dissolution and time off work do not appear to significantly differ (statistically) by union type, for the most part. That is, tests of second differences suggest that the benefits of fathers’ time off work for parental relationships largely persist for all couples, regardless of union type. This finding is consistent with other research suggesting that the benefits of father involvement for parental relationships largely persist regardless of union type (Hohmann-Marriott 2009; Kalmijn 1999; Keizer and Shenk 2012; McClain and Brown 2017; Nomaguchi et al. 2017). Taking time off work – regardless of union type – may indicate a commitment to shared parenting and provide evidence of fathers’ support.

Additionally, we find evidence that the association between taking longer periods of time off work and relationship dissolution is more pronounced for married couples than for unmarried couples. Taking longer periods of time off may provide married fathers with more time to demonstrate their shared parenting commitments and relationship supportiveness (Kotsadam and Finseraas 2011; Petts and Knoester 2020; Rehel 2014). And, longer periods of time off work by married couples may be more supported by workplaces (Petts et al. 2020; Williams et al. 2013). In contrast, taking longer periods of time off appear to be less likely to reduce the likelihood of relationship dissolution in cohabiting and nonresident father families, which may be due to lack of institutionalized support for these couples, as well as nonresident fathers’ more limited access to provide care and support when they do not reside with their child and the child’s mother.

Regardless, despite its benefits, access to paternity leave is unlikely to increase for disadvantaged fathers without significant policy interventions. As demonstrated by other nations, a federally mandated, subsidized, well-paid leave policy that reserves a period of leave specifically for fathers seems to be the best option to encourage fathers’ leave-taking (Dearing 2016). The introduction of paid family leave policies at the state-level provide evidence that such policies can increase fathers’ leave-taking within the U.S. (Bartel et al. 2018). However, such options are limited in the U.S., leaving few options for socioeconomically disadvantaged parents to take extended periods of time off work following a birth (Klerman et al. 2012; Matos et al. 2017). Since 2005, the U.S. federal government has spent approximately $2 billion ($150 million annually) via The Healthy Marriage and Responsible Father Initiative (U.S. Department of Health and Human Services) on skills-building courses aimed at promoting marriage and engaged fatherhood with minimal success (Hsueh et al. 2012). Given evidence from this paper along with other research showing the benefits of paid paternity leave for fathers’ involvement in family life (Almqvist and Duvander 2014; Bünning 2015; Kotsadam and Finseraas 2011; Petts and Knoester 2019; 2020; Pragg and Knoester 2017), evidence increasingly suggests that redirecting these funds toward a federally subsidized paid paternity leave program may benefit families.

Although there are many strengths to this study, there are some limitations to note. First, these data do not contain information on how fathers are able to take time off work. Because American fathers often take short periods of time off work, having a more precise measure of length of time off (i.e., days instead of weeks) would also be helpful to policymakers in considering what types of leave policies may help to promote more stable parental relationships. Second, fathers were only asked about taking time off work for the focal child. It is possible that these fathers took (or chose not to take) time off for any subsequent births during the observed period (which may be associated with relationship dissolution), but this information is not available in the data. Third, although we speculate that the association between fathers’ time off work and relationship dissolution may be due in part to parents’ (especially mothers’) increased feelings of equity and reduced work-family conflict, we do not have information on these factors. We included relationship support as a proxy for these processes, but future research should examine specific measures of equity and fairness to more fully assess whether fathers’ leave-taking may promote egalitarianism. Fourth, this study utilizes data from a disproportionately socioeconomically disadvantaged sample. Although use of this sample is important in illustrating the potential benefits of time off work following a birth for this population (which has been largely understudied), future research should consider whether and how the association between fathers’ time off work and relationship dissolution may differ by SES.

Finally, although we account for selection effects to the extent that we are able, we are unable to account for selection effects stemming from unobserved factors (e.g., how fathers took time off work, workplace support, motivation to be an involved father) and from selection into employment. In addition to minimizing selection to the extent we are able, we considered whether the association between fathers’ time off work and relationship dissolution differed by whether fathers took paid or unpaid time off as a way to assess additional mechanisms that may be driving these results. Results suggest that fathers’ time off work was only associated with a lower likelihood of relationship dissolution if at least some of this time off was paid (Table A3). It could be that paid time off is particularly important for socioeconomically disadvantaged families, as paid time off allows fathers to be at home when their child is born while avoiding major financial setbacks. However, this could also provide evidence of selection, as fathers who are able to take paid time off may have more supportive jobs that help them to maintain more stable parental relationships. Unfortunately, we were not able to fully explore variations in paid and unpaid leave due to small cell sizes when analyzing length of time off work, but future research should continue to explore the possible consequences of fathers’ paid and unpaid time off work following a birth for families, while incorporating additional selection factors that might include indicators of the quality and flexibility of one’s job, for example.

Despite these limitations, we find that fathers’ time off work following the birth of a child is associated with more stable parental relationships among socioeconomically disadvantaged couples in the U.S. The economic and social constraints these parents face likely make it difficult to fulfill the dual responsibilities of breadwinning and caregiving, increasing the likelihood of union transitions (Hohmann-Marriott 2009; Lichter et al. 2006; Williams 2010). As such, increasing access to paid paternity leave in the U.S. may provide socioeconomically disadvantaged families with structural supports that foster family stability.

Table A1.

Results Comparing Estimates from Models using Various Techniques to Account for Selection

	Unweighted Event History Model		Weighted Event History Model		Selection Model
Variable	B	SE	B	SE	B	SE
Panel A: Main Effects
Took Time Off	−0.18*	0.09	−0.23*	0.09	−0.11*	0.05
Length of Time Off
One Week	−0.15	0.09	−0.20*	0.09	−0.04	0.03
Two or More Weeks	−0.29*	0.12	−0.31*	0.12	−0.09*	0.04
Panel B: Interactive Effects
Took Time Off	−0.56**	0.20	−0.52*	0.20	-	-
Length of Time Off
One Week	−0.38	0.21	−0.35	0.21	-	-
Two or More Weeks	−1.04***	0.25	−0.99***	0.25	-	-
Cohabitinga	0.36	0.22	0.39	0.23	-	-
Romantica	0.71**	0.23	0.80**	0.24	-	-
Took Time Off x Cohabiting	0.41	0.24	0.37	0.24	-	-
Took Time Off x Romantic	0.52*	0.25	0.42	0.25	-	-
One Week x Cohabiting	0.23	0.25	0.19	0.25	-	-
Two or More Weeks x Cohabiting	0.86**	0.30	0.83**	0.30	-	-
One Week x Romantic	0.28	0.26	0.19	0.26	-	-
Two or More Weeks x Romantic	1.24***	0.33	1.13**	0.34	-	-

Unweighted event history model estimates are taken from discrete-time event history models that do not account for selection. Weighted event history model estimates are replicated from Model 2 (Panel A) and Model 4 (Panel B) of Tables 3–4 (but regression coefficients are displayed instead of odds ratios to allow for more direct comparisons). Selection model estimates are taken from selection models predicting whether couples ever experienced dissolution. Propensity score models are used to estimate the influence of taking time off and augmented inverse propensity weighted estimates are used to estimate the influence of length of time off.

^aCoefficients for union type differ in models including time off work and length of time off. The coefficients presented are from models including time off work.

^*p < .05.

^**p < .01.

^***p < .001.

Table A2.

Probability of Relationship Dissolution by Fathers’ Time Off Work and Union Type; Marginal Effects of Time Off Work and Differences in Effects Across Union Type

Panel A: Time Off Work	No Time Off	Took Time Off	Difference (AME of time off)	Contrasts by Union Type
a Married	0.14	0.09	0.05*	-
b Cohabiting	0.19	0.17	0.02	-
c Romantic	0.25	0.24	0.02	-
Panel B1: Length of Time Off Work	No Time Off	1 Week Off	Difference (AME of time off)	Contrasts by Union Type
a Married	0.14	0.10	0.04	-
b Cohabiting	0.19	0.17	0.02	-
c Romantic	0.26	0.23	0.03	-
Panel B2: Length of Time Off Work	No Time Off	2+ Weeks Off	Difference (AME of time off)	Contrasts by Union Type
a Married	0.14	0.06	0.08***	c
b Cohabiting	0.19	0.17	0.02	-
c Romantic	0.26	0.28	0.02	a
Panel B3: Length of Time Off Work	1 Week Off	2+ Weeks Off	Difference (AME of time off)	Contrasts by Union Type
a Married	0.10	0.06	0.04***	b, c
b Cohabiting	0.17	0.17	0.00	a
c Romantic	0.23	0.28	0.05	a

AME model estimates are used to obtain first and second differences (i.e., whether time off work is associated with relationship dissolution within each union type and whether the effect of time off work on relationship dissolution differs across union type). The difference column presents first differences and the contrasts column presents second differences (when statistically significant at p < .05).

Table A3.

Results Comparing the Associations between Paid Time Off, Unpaid Time Off, and Relationship Dissolution

	Unweighted Event History Model		Weighted Event History Model		Selection Model
Variable	B	SE	B	SE	B	SE
Panel A: Main Effects
Took Time Off
Some paid time	−0.31**	0.10	−0.34**	0.10	−0.10**	0.03
Only unpaid time	−0.09	0.10	−0.11	0.10	−0.04	0.03
Panel B: Interactive Effects
Took Time Off					-	-
Some paid time	−0.78***	0.22	−0.74**	0.21
Only unpaid time	−0.23	0.22	−0.19	0.22
Cohabiting	0.38	0.22	0.40	0.23	-	-
Romantic	0.72**	0.23	0.82**	0.24	-	-
Some paid time x Cohabiting	0.52*	0.26	0.49	0.26	-	-
Only unpaid time x Cohabiting	0.14	0.26	0.10	0.26	-	-
Some paid time x Romantic	0.75**	0.28	0.64*	0.28	-	-
Only unpaid time x Romantic	1.17	0.27	0.09	0.28	-	-

Fathers are classified as taking “some paid time” if they reported receiving at least one week of paid leave. Fathers are classified as taking “only unpaid time” if they reported not receiving any weeks of paid leave. Of the fathers who took leave, 76 (5%) reported taking both paid and unpaid time. These fathers are classified as “some paid time” here. All control variables are included in the models presented (but not shown to conserve space), and the modeling approach is identical to the approach presented in Table A1. Differences between paid and unpaid time off are statistically significant in Panel A and for married couples in Panel B (p < .05).

^*p < .05.

^**p < .01.

^***p < .001.