Predictors of Supportive Coparenting After Relationship Dissolution Among At-Risk Parents

Abstract

Supportive coparenting after relationship dissolution is associated with increased father involvement which can buffer against the negative effects of parental relationship dissolution. Low-income, at-risk families are much more likely to experience relationship dissolutions; hence, supportive coparenting after dissolution is particularly important in these families. We examined whether relationship (commitment and quality) and child (difficult temperament and gender) characteristics predicted initial levels of, and change in, supportive coparenting after relationship dissolution in the Fragile Families and Child Wellbeing Study (N = 1,603). We used structural equation modeling of latent growth curves to examine four time points collected at the focal child’s birth and first, third, and fifth birthdays. Ninety-percent of the mothers had nonmarital births, and about three-quarters had a high school diploma or less education. Overall, supportive coparenting decreased over time. Mothers in more committed relationships prior to the dissolution initially had significantly lower supportive coparenting. But over time, mothers who had been in more committed relationships increased in supportive coparenting. Mothers who had been in higher quality relationships prior to dissolution initially reported more supportive coparenting. At each time point, if a mother was romantically involved with a new partner, she reported significantly lower supportive coparenting compared to mothers who were single. With regard to child characteristics, mothers who reported their child as more difficult had significantly lower initial supportive coparenting. Similar results for fathers are discussed. Overall, the relationship characteristics of parents were important predictors of supportive coparenting both initially and over time.

Increasing evidence indicates that successful coparenting relationships are beneficial for children’s socioemotional development (McHale et al., 2002). These coparenting relationships are not merely characterized by the absence of destructive conflict, but involve active cooperation and communication between parents (Cohen & Weissman, 1984). Effective coparenting may be especially beneficial for the adjustment of children whose parents no longer share a romantic relationship (Ahrons, 1993). This is in part because cooperative coparenting relationships may be key to fostering positive nonresidential father involvement (Carlson, McLanahan, & Brooks-Gunn, 2008) and lower levels of interparental conflict (Maccoby & Mnookin, 1992), which have been linked to children’s behavioral outcomes, academic achievement, and psychological well-being (Amato & Gilbreth, 1999).

However, the contexts in which children experience coparenting by nonresidential parents have changed. Cohabitation has become increasingly common, and partly due to this increase, nonmarital childbearing reached historic levels in 2009, when the proportion of all births to unmarried women increased to 41%, the highest level reported in U.S. history. Nonmarital relationships are associated with multiple risk factors for families and children, including both economic and relationship instability (Edin & Reed, 2005). The relatively easy dissolution of nonmarital relationships compared to the costly and often lengthy divorce process places children at greater risk for cumulative effects of multiple family transitions. The more quickly parents dissolve their relationship, the sooner either parent can move into a new relationship, which itself will be rendered less stable by the presence of their child (Lichter, Qian, & Mellott, 2006). Due to multiple disadvantages, children born into these families are at the greatest risk for experiencing parental relationship dissolution (Bumpass & Lu, 2000) and may benefit the most from supportive coparenting.

Although some is known about correlates of coparenting for married parents (e.g., Lindsey, Caldera, & Colwell, 2005), relatively little is known about relationship or child characteristics that may shape the quality of coparenting relationships after dissolution, especially for never-married parents. It is critical to understand the factors that facilitate supportive coparenting in order to develop and implement effective programs that will ameliorate negative effects of family transitions. This study extends the literature on coparenting by focusing on predictors of and change in supportive coparenting in an at-risk sample of parents whose relationship has dissolved. Data come from The Fragile Families and Child Wellbeing Study, which follows 3,712 children born to unmarried parents (both nonresidential and cohabiting), and a comparison group of 1,186 children born to married parents. In particular, the present study examined associations between postdissolution supportive coparenting and relationship factors, child characteristics, and demographic variables.

The Context of Supportive Coparenting

A coparenting relationship exists when individuals have overlapping responsibility for parenting children, and its quality can be characterized by the extent to which parents support (or fail to support) each other’s parenting efforts (Feinberg, 2003). Key ingredients for a supportive coparenting relationship include joint investment in the child, valuing the importance of the other parent for the child’s growth and development, respecting the judgments of the other parent, and ongoing communication regarding the child’s needs (Cohen & Weissman, 1984).

Maintaining or forging a supportive coparenting relationship after the dissolution of a romantic relationship is especially challenging. Effective coparenting requires parents to set aside differences in their previous personal relationship, but often their relationship remains strained. Ahrons (1993) estimated that due to the volatile nature of this relationship after dissolution, less than half of divorced parents coparent effectively. Similarly, in their longitudinal study of coparenting, Maccoby, Depner, and Mnookin (1990) reported that only one fourth of couples were coparenting cooperatively in the second year post divorce.

The Role of Relationship Factors

Although the distinctness of romantic and coparenting relationships has been well established (Schoppe-Sullivan, Mangelsdorf, Frosch, & McHale, 2004), the quality of romantic and coparenting relationships are linked (Van Egeren, 2004). Schoppe-Sullivan, Mangelsdorf, Brown, and Sokolowski (2007) found that parents with higher prebirth marital quality formed more supportive coparenting relationships than those with lower prebirth marital quality. Maccoby et al. (1990) found that higher conflict at the initiation of a divorce was prognostic of less cooperative coparenting 18 months later.

Predissolution commitment levels may also affect the quality and course of coparenting. The investment model (Rusbult, 1980) suggests that the investment of resources is a marker of commitment. Partners invest resources, including time, money, and social/emotional support. With the passage of time, these investments grow and commitment increases. Supportive coparenting may suffer initially due to greater feelings of anger and loss after the dissolution of more committed relationships. However, high initial levels of conflict often decrease with time (Hetherington, 1999), creating the possibility for more supportive coparenting.

In contrast, the relationships of unmarried couples are characterized by lower levels of commitment (Nock, 1995) and greater instability (Lichter et al., 2006). Because of lower mutual investment, interparental conflict is likely to be lower immediately after dissolution, enabling less committed parents to initially coparent cooperatively. But, the lower level of relationship commitment may have a lingering negative effect—these parents may disengage over time, and supportive coparenting may decrease. When less committed relationships dissolve, there is no mandate to enter the family court system, and if couples do, it is most often to obtain a child support order. In these families, fathers may not remain involved in child rearing and as a result, coparenting can suffer (Carlson et al., 2008).

Another factor that may contribute to decreases in supportive coparenting postdissolution is repartnering. After a union dissolution, about 42% of women in a large Canadian sample repartnered, and former cohabitors did so more quickly than divorced individuals (Wu & Schimmele, 2005). As family ties unravel and new relationships are formed, maintaining effective coparenting becomes more difficult. Unmarried fathers in the third wave of the Fragile Families and Child Wellbeing Study reported less supportive coparenting when the mother of their child was in a new romantic relationship (Bronte-Tinkew & Horowitz, 2010).

The Role of Child Characteristics

Families function as systems, and the interdependence of these systems implies that all members evoke responses from each other (Thibaut & Kelley, 1959). As a result, child temperament and gender may influence coparenting. Research on the effects of child gender and temperament on supportive coparenting has been mixed. Some studies have found that parents of more difficult children report less supportive coparenting (e.g., Van Egeren, 2004), but others report no significant associations between child temperament and supportive coparenting (Schoppe-Sullivan et al., 2007). Similarly, although a few studies have reported more supportive coparenting in families with male children (e.g., Bronte-Tinkew & Horowitz, 2010), others have reported no differences in supportive coparenting by child gender (e.g., Lindsey et al., 2005).

The Present Study

We examined relationship and child characteristics as correlates of supportive coparenting after relationship dissolution in an at-risk population. Our study contributes to the literature by considering predictors of change in supportive coparenting over time as well as correlates of its initial levels, consistent with the notion that families are dynamic systems that constantly adjust to the needs of their members. The focus on supportive coparenting after the dissolution of nonmarital unions provides a unique look at a growing understudied population.

We hypothesized that supportive coparenting would decrease over time. We further hypothesized that higher predissolution relationship quality would be associated with more cooperative coparenting initially following dissolution, and a smaller decline in supportive coparenting over time. Because more committed relationships likely have more contentious dissolutions, we hypothesized that initially, more committed predissolution relationships would be characterized by less supportive coparenting. Over time, however, we expected relationships with lower initial levels of commitment to decrease in supportive coparenting, because parents may disengage from one another. Moreover, mothers that formed new romantic relationships were expected to report lower supportive coparenting compared to mothers who stayed single. Finally, we examined associations between infant gender and temperament and initial levels of and change in supportive coparenting over time; however, given the mixed findings of previous research we did not advance specific hypotheses regarding these associations.

Multiple demographic factors have been identified as correlates of supportive coparenting, including parental education, age, and employment status (Lindsey et al., 2005). Thus, these variables were included as control variables in the analysis.

Method

We used data from the Fragile Families and Child Wellbeing Study, a study of 4,898 mothers and fathers (n = 3,830) who had children (3,711 nonmarital and 1,187 marital) in the US between 1998 and 2000. Both mothers and fathers, when available, were interviewed in the hospital shortly after their child’s birth with follow-up interviews conducted when the child was one, three, and five years old. The Fragile Families and Child Wellbeing study suffered from selective participation and attrition of fathers from the sample (see Carlson et al., 2008 for a discussion). We had data from 79% of the fathers formerly partnered with the mothers in our sample. We relied primarily on the maternal data, but to test the sensitivity of our models to parental gender we reported sensitivity analyses using paternal reports of supportive coparenting as well.

To be in our sample, mothers (a) must have been dating, cohabiting, or married to the father at some point during the first three years of the child’s life (n = 4,312, loss due to restriction = 12.0%), (b) must have participated in at least one follow-up (n = 4,121, loss due to restriction = 4.4%), (c) must have experienced the dissolution of their relationship with the father by child age five (n = 1,907, loss due to restriction = 53.7%), and (d) must have had a child that saw their father at least once after the dissolution of their relationship (n = 1,603, loss due to the restriction = 15.9%). Dissolution at each wave was defined as reporting Separated/divorced, Just friends, or Not in any kind of relationship in response to the question: “What is your relationship with father now?” as well as reporting living with the father rarely or never. Over time, some mothers completely dropped from the sample. Of the mothers whose relationship dissolved by child age one, 2% of the mothers contributed one, 11% contributed two, and 87% contributed three observation points postdissolution. Of the mothers who dissolved by child age three, 9% contributed only one, and 91% contributed two observation points postdissolution.

Variables

Supportive coparenting

Supportive coparenting was measured using a 6-item scale, and was reported by both parents at child age one, three, and five if the father saw the child at least once since the previous survey. Questions measured key aspects of supportive coparenting, including interparental cooperation, communication, and the extent to which parents respected and valued each other’s parental roles (Cohen & Weissman, 1984). The same items have been used to measure supportive coparenting in other research utilizing these data (e.g., Carlson et al., 2008). Items included (a) “When (father) is with (child), he acts like the father you want for your child,” (b) “He supports you in the way you want to raise (child),” (c) “You can trust father to take good care of child,” (d) “He respects the schedules and rules you make for child,” (e) “You and father talk about problems that come up with raising child,” and (f) “You can count on father for help when you need someone to look after child for a few hours.” Response options when the child was one were 1 (rarely true), 2 (sometimes true), and 3 (always true). However, response options when the child was three and five included a never true option. These response options were recoded as 1 (never true/rarely true), 2 (sometimes true), and 3 (always true). The Cronbach’s alpha of the scale for mothers was .86, .88, and .87 and for fathers was .88, .87, and .77 at child ages one, three, and five respectively.

Relationship characteristics

The maternal reported continuum of predissolution commitment was a continuous measure of commitment whereby 0 = the couple was dating at the wave prior to the dissolution, 1 = the couple was dating and living together (cohabiting) at the wave prior to the dissolution, and 2 = the couple was married at the wave prior to the dissolution. At birth, the continuum of predissolution commitment was assessed from “Which of the following statements best describes your current relationship with your baby’s father? We are romantically involved on a steady basis; We are involved in an on-again and off-again relationship; We are just friends; We hardly ever talk to each other; We never talk to each other.” At each wave postbirth, living together was assessed by “Are you and (FATHER) currently living together all or most of the time, some of the time, rarely, or never?” Dating was defined as (a) romantically involvement on a steady basis or in an on-again/off-again relationship, (b) living together less than all or most of the time, and (c) not reporting marriage. Cohabitation was defined as (a) romantic involvement on a steady basis, (b) living together all or most of the time, and (c) not reporting marriage. Marriage was defined as reporting current marriage.

Relationship quality was measured from both parents at each wave and the measure of predissolution relationship quality was the quality of the relationship at the wave immediately preceding dissolution. Sample items included “He is fair and willing to compromise when you have a disagreement” and “He encourages or helps you to do things that are important to you.” Response options were 0 (never), 1 (sometimes), and 3 (often). The sum of all items was the scale score. The Cronbach’s alpha of the scale was .62, .63, and .62 for mothers and .72, .70, and .77 for fathers at the child’s birth, age one, and age three respectively.

At each wave, if the mother and father were not living together all, most, or some of the time, mothers were asked, “Are you currently involved in a romantic relationship with someone (other than the father of your child)?” Responses were coded 0 (no) and 1 (yes).

Child characteristics

Child gender was coded at birth as 0 (female), 1 (male). Difficult temperament was measured from mothers using selected items from the EAS (Emotionality, Activity, and Sociability) Temperament Survey for Children (Mathieson & Tambs, 1999) at age one and was the mean of: (a) “(He/She) often fusses and cries,” (b) “(He/She) gets upset easily,” and (c) “(He/She) reacts strongly when upset,” rated from 1 (not at all like my child) to 5 (very much like my child). The Cronbach’s alpha of the difficult scale was .59.

Control variables

Control variables were measured from mothers at birth unless noted. Parental age was measured in years. Parental race/ethnicity was measured as non-Hispanic/non-Black (0 = not non-Hispanic/non-Black, 1 = non-Hispanic/non-Black), non-Hispanic Black (0 = not non-Hispanic Black, 1 = non-Hispanic Black), and Hispanic (0 = non-Hispanic, 1 = Hispanic). Parental education was measured as Less than High School (0 = at least a high school diploma, 1 = high school dropout), High School (0 = less than a high school degree or some college or more education, 1 = high school graduate), and Some College (0 = high school graduate or less education, 1 = at least some college). Paternal employment was measured as 0 = father unemployed, 1 = father employed. The wave of dissolution was also coded.

Analytic Strategy

Because parents dissolved at different points across the four waves of data available for this project, data were pooled. For time-invariant variables, such as child gender and mothers’ education, we used the value at a single wave (either birth or child age one), or we used the value of the variable right before dissolution. However, for time-varying variables, we pooled our data such that short-term variables included data from the wave immediately following dissolution. Medium-term variables contained data from the second wave postdissolution, and long-term variables contained data from the third wave postdissolution (only available for those who dissolved between birth and child age one).

We used structural equation modeling of latent growth curves because we sought to capture and predict both initial levels of supportive coparenting, as well as the change in coparenting over time. To account for missing data, we used full information maximum likelihood estimation (Muthén, Kaplan, & Hollis, 1987). Each individual trajectory contained a unique intercept (α), a linear, time-varying slope (β), and time-specific measurement error (ε). Thus, the level one equation of the latent growth curve model was:

$$y_{it}={\alpha }_i+{\beta }_{i}t+{\epsilon }_{it}$$ (1)

where y represented supportive coparenting for individual i at time t, or within-individual change across time. Next, we incorporated time-variant covariates, or independent variables, representing whether or not the mother had a new partner at time t. Thus, equation 1 was modified as follows:

$$y_{it}={\alpha }_i+{\beta }_{i}t+{\gamma }_t{np}_{it}+{\epsilon }_{it}$$ (2)

γ_tnp_it represented the effect (γ) of a new partner (np) at time t for individual i. A significant γ would represent a deviation to the coparenting growth curve due to maternal report of a new romantic partner. Time-varying control variables were also entered into the equation at this step.

At level two of the latent growth model, random intercepts (α_i) and slopes (β_i) were a function of time-invariant variables that differed between individuals, rather than within. Time-invariant covariates representing the continuum of predissolution relationship commitment (c), relationship quality (rq), child gender (cg), and child difficult temperament (dt) were modeled at this level. The equations for the intercept and slope at level two were:

$${\alpha }_i={\alpha }_0+{\alpha }_1{c}_{i1}+{\alpha }_2{rq}_{i2}+{\alpha }_3{g}_{i3}+{\alpha }_4{dt}_{i4}+u_i$$ (3)

$${\beta }_i={\beta }_0+{\beta }_1{c}_{i1}+{\beta }_2{rq}_{i2}+{\beta }_3{g}_{i3}+{\beta }_4{dt}_{i4}+{\nu }_i$$ (4)

Thus, the intercept (α) and slope (β) were each directly regressed on the time-invariant independent variables. The time-invariant control variables were also entered at level two.

Models were estimated using AMOS. The two levels of the latent growth curve model were modeled simultaneously and overall fit statistics were evaluated. Following Kenny (2010) and Bollen and Curran (2006), we examined four fit indices include the chi-square (χ²). However, as noted by Kenny, the chi-square, which is supposed to be nonsignificant, is almost always statistically significant for models with more than 200 cases. Thus, we also relied on the Tucker Lewis Index, also known as the Non-normed Fit Index (NNFI), and the Comparative Fit Index (CFI). A value between 0.90 and 0.95 is acceptable for both of these indices, and above 0.95 indicates a good fit. We also reported the Root Mean Square Error of Approximation (RMSEA) for which an acceptable value is below 0.08. We fit both the unconditional latent growth curve model (the model with no covariates or controls) as well as the full model.

Results

Descriptive Statistics

Overall, the mean level of supportive coparenting was above the midpoint of the scale at each time point (see Table 1). At the wave immediately preceding dissolution, the majority of couples were dating, less than 15% were married, and the average level of relationship quality was high. About half the children were male and the average temperament was around the midpoint of the difficult temperament scale. By one year postdissolution, about a third of the mothers had a new partner, and by four years postdissolution, almost half the mothers had a new partner. On average, mothers were in their early twenties, and three-quarters had a high school degree or less education. A majority of mothers were non-Hispanic Black. Fathers were generally in their midtwenties, and almost 80% had a high school degree or less education. At birth, about three-quarters of fathers were employed. Roughly half of relationships dissolved by the time the child was one and an additional third dissolved over the next two years.

Table 1.

Descriptive Statistics (N = 1,603)

	M	SD	Range	% missing
Time-variant dependent variablesa
Coparenting in the short-term	2.28	0.62	1–3	10.67
Coparenting in the medium-term	2.17	0.66	1–3	13.13
Coparenting in the long-term	2.20	0.64	1–3	21.30
Time-invariant independent variables
Relationship Characteristics
Continuum of Pre-Dissolution Commitment	0.70	0.71	0–2	0.00
% Dating	44.48
% Cohabiting	40.86
% Married	14.66
Pre-dissolution Relationship Quality	6.17	1.63	0–8	0.69
Child Characteristics
Male	0.52		0–1	0.00
Difficult temperament	2.90	1.08	1–5	0.56
Time-variant independent variablesa
New partner in the short-term	0.32		0–1	0.75
New partner in the medium-term	0.42		0–1	0.16
New partner in the long-term	0.46		0–1	0.13
Time-invariant control variables
Mother’s age	23.68	5.52	15–43	0.00
Mother’s education				0.12
Less than high school	0.38		0–1
High school degree	0.35		0–1
At least some college	0.27		0–1
Mother’s race/ethnicity
Non-Hispanic/non-Black	0.17		0–1	0.00
Non-Hispanic Black	0.62		0–1	0.12
Hispanic	0.21		0–1	0.06
Father’s age	26.49	7.23	15–80	0.94
Father’s education				7.30
Less than high school	0.34		0–1
High school degree	0.43		0–1
At least some college	0.23		0–1
Father employed	0.76		0–1	5.99
Time-variant control variables
Dissolution occurred at age one	0.52		0–1	0.00
Dissolution occurred at age three	0.32		0–1	0.00
Dissolution occurred at age five	0.16		0–1	0.00

^aFor time-varying variables such as coparenting and whether or not the mother had a new partner, the percent missing reported at each time point was the percent of interviewed mothers missing. For example, the percent missing on coparenting in the medium-term was the percent of mothers interviewed two years after the dissolution of their relationship without a valid score on coparenting. For attrition rates, see the text.

Latent Growth Curve Results

The mother’s unconditional baseline model demonstrated adequate fit (N = 1,603, χ²(3) 30.52, p =.00, NNFI = .86, CFI = .93, RMSEA =.08); thus, we tested our full model. Results for the unconditional model indicated that the there was significant intercept and slope variance to be estimated. The full maternal model, without control variables and covariances for illustrative purposes, is shown in Figure 1. The model fit the data very well (N = 1,603, χ²(25) = 37.86, p =.05, NNFI =.98, CFI =.998, RMSEA =.02). Overall, supportive coparenting decreased over time.

Figure 1.

Latent growth curve model predicting the level and growth in supportive coparenting from time-variant and time-invariant independent variables. Parameter estimates reported in Table 2. Time-variant and time-invariant control variables, as well as the covariances between dependent variables, were excluded to simplify the presentation of the model.

Relationship characteristics

The continuum of predissolution commitment was significantly related to both the intercept and slope of supportive coparenting (see Table 2). Mothers in relationships that were higher in commitment had significantly lower supportive coparenting initially, but increased significantly in supportive coparenting over time. Mothers reporting higher quality relationships prior to dissolution also reported higher initial levels of supportive coparenting, but there was no association between predissolution relationship quality and change in coparenting. At each wave postdissolution, mothers who were romantically involved with a new partner reported lower supportive coparenting than mothers who were single.

Table 2.

Results From Supportive Coparenting Growth Models and Time-Invariant Child Characteristics, Time-Invariant and Time-Variant Relationship Characteristics, and Controls

	Coparenting
	Intercept (α)				Slope (β)
	Un-standardized	SE	Standardized		Un-standardized	SE	Standardized
Intercept	2.13***	0.07			−0.04	0.05
Time-invariant independent variables
Relationship Characteristics
Continuum of Pre-Dissolution Commitment	−0.08	0.02		−0.13***	0.03	0.01	0.24**
Pre-dissolution Relationship Quality	0.07	0.01		0.25***	0.00	0.00	0.04
Child Characteristics
Male	0.04	0.03	0.04		−0.02	0.01	−0.11
Difficult temperament	−0.03	0.01	−0.08*		0.01	0.01	0.07
Time-variant independent variables
New partner in the short-term	−0.13	0.03	−0.10***		—	—	—
New partner in the medium-term	−0.32	0.03	−0.25***		—	—	—
New partner in the long-term	−0.16	0.04	−0.12***		—	—	—
Time-invariant control variables
Mother’s age	−30.00	0.00	−0.02		0.01	0.00	0.30**
Mother’s education
Less than high school	0.13	0.05	0.14**		0.02	0.02	0.08
High school degree	0.06	0.04	0.06		0.00	0.02	0.01
At least some college (Excluded group)	—	—	—		—	—	—
Mother’s race/ethnicity
Non-Hispanic/non-Black (Excluded group)	—	—	—		—	—	—
Non-Hispanic Black	0.15	0.04	0.16***		0.06	0.02	0.03
Hispanic	0.04	0.05	0.04		0.04	0.02	0.17
Father’s age	0.01	0.00	0.07		0.00	0.00	−0.12
Father’s education
Less than high school	0.01	0.05	0.01		−0.02	0.02	−0.13
High school degree	0.07	0.04	0.08		−0.02	0.02	−0.12
At least some college	—	—	—		—	—	—
Father employed	−0.05	0.04	−0.05		0.03	0.02	0.13
Time-variant control variables
Short-term dissolution occurred at age one (Excluded group)	—	—	—		—	—	—
Short-term dissolution occurred at age three	0.06	0.04	0.04		—	—	—
Short-term dissolution occurred at age five	0.24	0.04	0.14***		—	—	—
Medium-term dissolution occurred at age one (Excluded group)	—	—	—		—	—	—
Medium-term dissolution occurred at age three	0.11	0.03	0.08***		—	—	—
N	1603

Note. Model fit statistics reported in the text.

^*p < 0.05.

^**p < 0.01.

^***p < 0.001.

Supportive coparenting growth curves for mothers whose relationship dissolved between birth and child age one are shown in Figure 2 by predissolution commitment and whether or not the mother had a new romantic partner. Mothers who had been dating began with the highest level of supportive coparenting, but declined the most rapidly. In contrast, mothers who were married began with the lowest level of coparenting, but increased over time such that married mothers had similar levels of coparenting to dating mothers by child age five. As illustrated, when the mother began a new romantic relationship, coparenting suffered and continued to decline more rapidly compared to mothers who remained single.

Figure 2.

Growth curves of the change in maternal reported supportive coparenting following a relationship dissolution by predissolution relationship commitment and whether or not the mother was dating someone new when the child was three. Figure based on latent growth curve model with grand mean centered continuous variables (except predissolution relationship commitment). The growth curve represents data from a mother whose relationship dissolved by child age one, and had the average level of relationship quality predissolution, a male child with the average difficult temperament, and was a non-Hispanic Black, high school graduate of the average age whose former partner was a non-Hispanic Black, high school graduate of the average age who was employed at the birth of the child.

Child characteristics

We found no significant associations between child gender and initial levels or change in supportive coparenting. However, mothers who reported that their child had a more difficult temperament also reported significantly lower initial supportive coparenting; temperament was not associated with change in coparenting.

Control variables

Maternal age was not associated with initial levels of supportive coparenting, but older mothers increased significantly in supportive coparenting across time. Mothers with less than a high school education reported significantly better coparenting initially than mothers with at least some college, but maternal education was not associated with change in coparenting. Non-Hispanic Black mothers initially reported significantly higher supportive coparenting as compared to non-Hispanic/non-Black mothers, but there were no other significant race/ethnicity difference in supportive coparenting. Paternal age, education, and employment status were not significantly associated with either initial levels of supportive coparenting or its change over time. Mothers whose relationship dissolved at child age one reported significantly lower supportive coparenting immediately after dissolution compared to mothers whose relationship dissolved at child age five, and two years after dissolution compared to mothers whose relationship dissolved at child age three.

Sensitivity Analyses: Paternal Reports of Supportive Coparenting

Thus far we have relied solely on maternal report of supportive coparenting. To test the sensitivity of our model to parental gender and to recognize that supportive coparenting has a dyadic component, we reran our structural equation model of latent growth curves with paternal reports of supportive coparenting, predissolution relationship quality, and race, using full information maximum likelihood estimation. The unconditional baseline model demonstrated adequate fit (N = 1,603, χ²(3) = 32.40, p =.00, NNFI = .91, CFI =.91, RMSEA =.08); thus, we tested the full model. The model fit the data very well (N = 1,603, χ² (25) = 29.15, p =.26, NNFI =.99, CFI =.999, RMSEA =.01). Results from this model are reported in an online supplement.

Consistent with the maternal model, paternal reports of supportive coparenting decreased over time. Further, the continuum of predissolution commitment was significantly associated with the intercept and marginally significantly associated with the slope of paternal reported supportive coparenting. Consistent with the maternal findings, fathers in relationships that were higher in commitment prior to dissolution initially reported lower supportive coparenting, but increased at a greater rate in supportive coparenting over time. Fathers who reported higher predissolution relationship quality also reported higher initial levels of supportive coparenting, as did mothers. However, unlike mothers, fathers’ higher predissolution relationship quality was also strongly associated with a decrease in supportive coparenting over time.

Consistent with the maternal model, at each wave postdissolution, if a mother was romantically involved with a new partner, fathers reported significantly lower supportive coparenting than did fathers who did not have a coparent involved in a new romantic relationship. Child characteristics were not associated with paternal report of supportive coparenting. In terms of the demographic variables, consistent with the maternal model, non-Hispanic Black fathers initially reported more supportive coparenting than did non-Hispanic/non-Black fathers. Initially employed fathers reported an increase in supportive coparenting over time. Also consistent with the maternal model, fathers whose relationship dissolved at child age one reported significantly lower supportive coparenting two years after dissolution compared to fathers whose relationship dissolved at child age three. Overall, the results of the sensitivity analyses suggested that both maternal and paternal reports of supportive coparenting were particularly sensitive to pre and postdissolution relationship contexts.

Discussion

Coparenting marked by support and cooperation is beneficial for child development (McHale et al., 2002) and preserves father involvement in child rearing even after the parents’ relationship ends (Carlson et al., 2008). Yet, this study was one of the first to focus on the evolution of the coparenting relationship postdissolution. Consistent with expectations, we found that among our sample of low-income parents, 90% of whom had a nonmarital birth, several factors were associated with supportive coparenting over time.

The relationship context of the couple prior to dissolution was associated with supportive coparenting. The complex associations hypothesized for predissolution relationship commitment were supported. Initially, couples who had been in more committed predissolution relationships reported lower supportive coparenting. As suggested by the investment model (Rusbult, 1980), more committed couples had invested more and had more to lose when their relationship dissolved; these couples often had a combined household and sometimes a legal commitment. These dissolutions were likely more contentious and thus supportive coparenting initially suffered. Lower relationship quality prior to dissolution could also be indicative of an impending hostile dissolution. Hence, couples with lower relationship quality prior to dissolution should also have reported poorer initial coparenting; our data supported this assertion.

Yet, greater predissolution commitment was associated with increased supportive coparenting over time and appeared to buffer parents against the overall coparenting decline. There are several possible explanations. Married couples, those most committed, must enter the family court system to dissolve their relationship, and are required to set up legal child support and custody arrangements including a visitation plan that can facilitate father involvement. Never married parents can also obtain a legal child support order, but legal child support and custody agreements are not necessary to dissolve these relationships, and many have informal arrangements (Nepomnyaschy & Garnfinkel, 2009). Further, fathers who were in more committed relationships may have been more involved with their child before their relationship dissolved, as they were living with their child. They had more time to interact with their child and coparent with the mother compared to fathers who were romantically involved with their child’s mother, but not living with the child. Finally, as time passes, the anger associated with dissolving committed relationships subsides (Hetherington, 1999). Indeed, fathers who were less satisfied with their relationship prior to dissolution also increased in supportive coparenting over time; the initial negative feelings about the relationship likely passed and fathers increased in coparenting. Thus, higher predissolution commitment, and for fathers, higher predissolution relationship quality, was associated with increases in supportive coparenting over time as initial feelings of anger subsided, the legal system supported those dissolving more committed relationships, and formerly more committed, co-residential parents drew on their previous coparenting experience to support their postdissolution supportive coparenting relationship.

A second factor that was strongly related to supportive coparenting was maternal relationship status. At each time point, when a mother was in a new romantic relationship, both maternal and paternal reports of supportive coparenting were significantly lower than when mothers were single. These findings replicate and extend Bronte-Tinkew and Horowitz (2010) such that even years after the dissolution occurred, new partners appeared to negatively impact both paternal and maternal reports of coparenting. Repartnered mothers may have been less motivated to coparent cooperatively with the father because they had a new romantic partner. Further, these new partners may also be spending time with her child. In a comparison of the father involvement of coresidential biological fathers and the involvement of coresidential nonbiological partners using data from the Fragile Families and Child Wellbeing Study, Bzostek (2008) found that mean levels of father involvement were identical for coresidential biological fathers and coresidential nonbiological partners. Thus, mothers were repartnering with new partners who also wanted to be involved in the life of their child. Claessens (2007), using a qualitative sub-sample of the Fragile Families and Child Wellbeing Study, also found that fathers reported that their relationship with their child’s mother deteriorated after she began a new romantic relationship. Because of the higher rates of multipartner fertility (Carlson & Furstenberg, 2006) and relationship dissolution (Lichter et al., 2006) among low-income couples, coparenting may be a dynamic process depending upon the mothers’ relationship status.

The association between child characteristics and supportive coparenting postdissolution was also examined. Child gender was not significantly related to supportive coparenting. However, initial levels of maternal reported supportive coparenting suffered when a child had a more difficult temperament. Schoppe-Sullivan et al. (2007) found that parents with poorer marital relationships and difficult infants had less effective coparenting relationships. We also examined demographic factors and found that few were associated with the level and change in coparenting over time, perhaps due to a lack of variability in the demographic factors in this homogeneous, at-risk sample.

This study was not without limitations. The Fragile Families data set did not include a measure of coparenting conflict; thus, we were not able to examine changes in conflict postdissolution, which are undoubtedly important for children’s functioning and could show different trajectories and patterns of predictors. The difficult temperament and relationship quality measures had somewhat low reliabilities, and their associations with supportive coparenting should be reexamined with more reliable measures. Another limitation is that all measures used in this study were self-report. Observational data on coparenting after a relationship dissolution has never been collected, as far as the authors know, but could shed further light on the family processes that surround the coparenting of children, and may not have the same biases in reporting. We only observed about half of our sample at three time points postdissolution, and the child’s age at the first wave after dissolution was significantly associated with coparenting, such that mothers whose relationships dissolved when their child was younger reported less supportive coparenting compared to those who relationships dissolved when their child was older. Longer time horizons may shed further light on the evolution of the coparenting relationship postdissolution. Finally, the results of this study are only generalizable to low-income parents who were ever romantically involved and then dissolved their relationship within the first five years of their shared child’s life. It may be the case that many of our findings are unique to this population. For example, dating a new romantic partner may not interfere with supportive coparenting among more highly educated, higher income parents who may have factors sustaining the supportive coparenting relationship such as income, court appointed mediators, mental health professionals, family, coworkers, and friends. Mothers did not report on supportive coparenting for fathers who had not seen their child; thus, we were unable to draw conclusions about supportive coparenting in dyads that included fathers who had not seen their child in several years. It is possible that fathers who did not see their child found other ways to supportively coparent with the mother of their child. Future research should replicate our results in more representative samples to examine how universal these associations are.

In conclusion, this study was one of the first to examine supportive coparenting over time after relationship dissolution. The mothers in this study continue to be at risk for having both another child with a new partner (Carlson & Furstenburg, 2006) and relationship dissolution (Lichter et al., 2006). The contexts of families, both prior and postdissolution, not only shape coparenting over the next several years, but the development of the child as well. The difficulties these parents face in maintaining supportive coparenting relationships and promoting their child’s wellbeing seem insurmountable when they include incarceration, intimate partner violence, and substance abuse. But some couples are making their coparenting relationships work despite these obstacles. Future research should continue to study these couples in order to support the development of interventions that will enable all parents whose relationships end to form effective, cooperative coparenting alliances that will translate into positive outcomes for their child. Well-designed interventions to positively impact coparenting among this at-risk population already exist (Cowan, Cowan, Kline Pruett, & Pruett, 2007). The key is making interventions to help couples coparent effectively post dissolution more widely available.