Paternal Identity, Maternal Gate Opening, and Fathers’ Longitudinal Positive Engagement

Abstract

Fathers’ positive engagement (FPE) benefits children’s development but some children receive greater FPE than others. Understanding why some fathers demonstrate greater FPE than others is critical to efforts to support FPE. However, studies of FPE often fail to account for the father’s residential context and changes in interparental relationships and FPE that occur as the child develops. This study examined the effects of paternal identity (i.e., status-level/role-level centrality) and maternal gate opening on FPE in diverse residential contexts from early to middle childhood. Using longitudinal multilevel modeling, this study analyzed data from 2,339 families in the Fragile Families and Child Wellbeing Study. Results demonstrated that maternal gate opening strengthened the association between paternal status-level centrality and FPE especially for fathers who were nonresident at childbirth. This protective effect did not change by child age. Furthermore, fathers who were resident at childbirth but nonresident in middle childhood increased in FPE over time. However, among fathers whose residential status shifted from nonresident to resident, those with low role-level centrality decreased in FPE over time. For fathers who were resident from birth to middle childhood, high maternal gate opening was associated with greater FPE, although this effect faded over time. Overall, these findings suggest the importance of both establishment of a strong paternal identity and maintenance of maternal support to promote long-term FPE.

Fathers’ positive engagement (FPE) is associated with better developmental outcomes for children, independently from mothers’ engagement. These outcomes include greater cognitive skills (Rollè et al., 2019), fewer physical and mental health problems (Allport et al., 2018), fewer internalizing or externalizing behavior problems, greater prosocial behaviors, greater academic success (Marsiglio et al., 2000), less risk of delinquency during adolescence and adulthood, and in the case of boys from low-income households, less risk of economic disadvantage during adulthood (Sarkadi et al., 2008). Benefits of FPE in child development have been reported consistently regardless of fathers’ residential status (for a review about nonresidential fathers, see Adamsons & Johnson, 2013; Amato & Gilbreth, 1999).

In recent years, many parents in the U.S. have their first child before or outside of marriage and 22% of children are born to unmarried parents (U.S. Census Bureau, 2020). Today, one-quarter of children (25.1%) do not live with their fathers (U.S. Census Bureau, 2020). However, even among divorced families, 75% of nonresident parents continue to have contact with children, and only 25% of children have an increased risk of disadvantaged developmental outcomes due to parent divorce (Greene et al., 2012). FPE is defined as a father’s direct interaction with children during caretaking, play, and leisure activities, which can stimulate children’s cognitive and social-emotional development (Pleck, 2010). Although resident fathers are more likely to be exposed to their family’s and child’s needs for parenting compared to nonresident fathers, a father’s residential status does not guarantee high FPE. Bocknek (2020) addresses that it is possible for a father to be physically present but psychologically absent or physically absent but psychologically present. FPE is an essential indicator of fathers’ psychological presence in his child’s life, because it is difficult for a child to perceive a father’s presence unless he invests time in childcare, educational, or play activities with his child.

Theoretical Background about Fathers’ Positive Engagement with Children

Identity theory illustrates that paternal identity is associated with FPE. According to identity theory (McCall & Simmons, 1978), the self consists of multiple identities, and identities have a prominence hierarchy. Although previous research on fathering has rarely differentiated, father identity can be divided into two levels: status-level identity and role-level identity (Kuscul & Adamsons, 2022). That is, status-level identity captures how much value a father puts into simply becoming or being a father. On the other hand, role-level identity captures paternal identity while performing a particular role (Kuscul & Adamsons, 2022). Kuscul and Adamsons (2022) suggest that fathers develop status-level identity first during the transition into parenthood and then role-level identity later. Also, a father’s status-level identity is expected to stay stable whereas role-level identity would change over time by child, father individual, or family development. With regard to fathering, status identity is assumed to be associated with global levels of presence while role identity is more likely to influence specific father involvement behaviors (Kuscul & Adamsons, 2022). When discussing father’s role identity, of course, fathers have multiple roles as a parent such as breadwinner, disciplinarian, role model, or mentor, but we focused on father’s role identity as a direct caregiver out of multiple fathering role identities in relation to FPE. Indeed, although other roles are consistent with traditional views on fathering, the most prominent change in fathering roles over the past 50 years is the increased expectation for fathers’ active involvement in direct care of infants and young children (Pleck & Pleck, 1997). Based on a recent study (Adamsons & Pasley, 2016), we hypothesized that both paternal status-level and role-level centrality would be positively associated with later FPE, but their longitudinal effects on FPE could be different over time. So far, there is little research about how distinct dimensions of paternal identity are related to FPE in diverse residential contexts.

Identity control theory (Burke, 2006) provides theoretical background to understand that the feedback loop with the significant other also influences the establishment and the enactment of parental identity (Adamsons, 2010). According to the identity control theory, each person has identity standards on one’s own roles as well as their partner’s. Identity standards on the partner’s roles are called counteridentity. If a new father’s identity standards and his child’s mother’s counteridentity about paternal identity are similar, the feedback from his child’s mother would confirm the father’s paternal identity and strengthen FPE. But if parents have disconfirming counteridentities, inconsistency between a parent’s own identity standards and actual behaviors would increase (Adamsons, 2010). Regarding parenting, there is reason to think that mothers’ counteridentity standards are particularly relevant to FPE, given that mothers are traditionally regarded as primary caregivers for children, with well-defined societal expectations for their roles. Indeed, previous research indicates that the effects of coparenting dynamics between mother and father on FPE are greater than their influence on mothers’ engagement (Akande & Heath, 2019; Roggman et al., 2012).

One way in which mothers’ counteridentity standards for fathers may manifest is via maternal gatekeeping, which is defined as “the degree to which mothers encourage or discourage fathers’ positive engagement in childrearing” (Schoppe-Sullivan et al., 2015, p. 167), and consists of gate opening (i.e., encouragement) and gate closing (i.e., discouragement). The concept of maternal gatekeeping began from maternal gate closing (e.g., Allen & Hawkins, 1999), but maternal gate opening has gained more attention as a protective factor. For example, mothers’ gate opening behaviors are directly associated with fathers’ greater engagement in childrearing (Fagan & Cherson, 2017; Zvara et al., 2013), and this positive association between maternal gate opening behaviors and FPE remains even after controlling for parents’ beliefs about the father’s role (Schoppe-Sullivan et al., 2008).

When focusing on dyadic processes in the establishment or the enactment of parental identities between coparents, identity theory has limitations especially when explaining (1) that the role expectation of fathering change over the life course according to the child’s developmental needs and (2) how contextual factors affect FPE. To account for these, our framework situated identity theory within Cabrera et al. (2007, 2014)’s heuristic model of FPE. Focused on the uniqueness of FPE compared to mothering, Cabrera et al. (2007) assumed that the role expectations for fathering can differ by child age, and that FPE would change over the life course following changes in children’s developmental needs. Later, Cabrera et al. (2014) put forth a more developed ecological model to explain FPE, including individual as well as family contextual factors. Thus, incorporating Cabrera et al. (2007, 2014)’s model helps create a more complete picture of FPE, particularly how it may evolve over time and in context.

This study has several unique contributions. First, this study extends the sample including resident-father families and nonresident-father families and it also extends the period of FPE to middle childhood. Previous research investigating the roles of paternal identity and maternal gatekeeping has tended to focus on resident-father families during the first five years after childbirth (e.g., Allen & Hawkins, 1999; McBride et al., 2005; Schoppe-Sullivan et al., 2015; Schoppe-Sullivan et al., 2008; Zvara et al., 2013). Considering the complex associations between fathers’ residential status, expectations for their parental responsibilities, and their access to children, it is worthwhile to investigate patters of FPE in diverse residential contexts for a longer period. Second, previous research has not taken into account flexibility in family structure over time, instead dividing fathers into two simplified groups (resident-father vs. nonresident-father families). As a result, our understanding of FPE in diverse residential contexts and how to best support fathers’ enactments of their roles in diverse residential contexts has been restricted. Lastly, previous research analyzing nonresident-father families has often focused only on maternal gate closing (Akande & Heath, 2019). Considering growing needs to understand fathers’ engagement in diverse family contexts (Roggman et al., 2012), and to take a strengths-based approach focusing on family resilience, we focus on the role of maternal gate opening in the association between paternal identity and FPE in diverse residential contexts.

The Present Study

This study examined the effects of paternal identity and maternal gate opening on FPE in diverse family contexts from early to middle childhood by analyzing longitudinal survey data from the Fragile Families and Child Wellbeing Study, which oversampled nonmarital births in large U.S. cities. In this research, we addressed four major questions: (1) How is paternal identity associated with FPE in different fathers’ residential contexts? (2) How is maternal gate opening associated with FPE in different fathers’ residential contexts? (3) Does child age moderate the association between paternal identity and FPE or the association between maternal gate opening and FPE from early to middle childhood? (4) Does maternal gate opening moderate the associations between paternal identity and FPE? In keeping with theory and past research (Adamsons & Pasley, 2016, Lamb, 2010), with regard to paternal identity, we hypothesized that (1) a father’s initial identity would be associated with their later FPE, (2) maternal gate opening would be positively associated with FPE, but (3) the direct associations between paternal identity/maternal gate opening and FPE would decline as their child aged considering the phenomenon of homeostasis in family systems theory (Minuchin, 1985), and (4) the associations between paternal identity and FPE would be strengthened by maternal gate opening.

Since it is unknown how these associations appear in different fathers’ residential contexts, we wanted to test the same models across different fathers’ residential contexts. With regard to fathers’ residential status, we classified it into four groups by pairing the father’s residential status at childbirth and at child age 9 years: (1) from nonresident to nonresident, (2) from nonresident to resident, (3) from resident to nonresident, and (4) from resident to resident. Across diverse residential contexts, we conducted similar longitudinal multilevel modeling including the same key predictors such as paternal identity and maternal gate opening with the same covariates. To account for flexibility in residential status between early and middle childhood in each group, we also included a covariate in each model reflecting stability or change in residential status in the intervening years. Again, our intention was to use fathers’ residential status as a proxy that provides more detailed information about the contexts in which FPE occurs and that may capture some variability in fathers’ access to children and day-to-day parenting responsibilities. We did not consider residential status as an indicator of higher quantity or quality of FPE.

Based on Cabrera et al. (2014), we included covariates with the assumption that father’s individual characteristics (race/ethnicity, age, education), child characteristics (temperament, gender, first child or not), and family economic contexts (father’s employment status, family poverty ratio) are associated with FPE. We assumed that the associations between the covariates and FPE might differ in fathers’ diverse residential contexts. According to identity theory (Adamsons, 2010), the covariate of mothers’ traditional gender role beliefs is also included as an indicator of mother’s identity on mothering and counteridentity on fathering.

Method

We report how we determined our sample size, and all measures in the study. This study is not preregistered.

Data and Analytic Sample

This study analyzed data from a subsample of 2,339 families whose fathers reported their residential statuses at least both at the 1^st wave and the 5^th wave (child age 9 years) in the Fragile Families and Child Wellbeing Study (FFCWS; https://fragilefamilies.princeton.edu/). For this research, IRB approval was not required because only publicly available data were analyzed. The FFCWS originally recruited 4,700 families including mothers and fathers at hospitals at childbirth in 20 cities across the U.S. Longitudinal data were collected from the child’s birth to age 1, 3, 5, 9, and 15 years old, but due to the attrition rates, the sample size decreased over time. In the follow-up assessments, the FFCWS contacted mothers and fathers through telephone surveys and visited their homes to gather information about demographic, socioeconomic, and psychological characteristics, and their children’s cognitive and social-emotional development (McLanahan & Garfinkel, 2000, 2012; Reichman et al., 2001).

Given that the FFCWS was designed to represent non-marital births in U.S. cities, the sample contains significant proportions of low-income families and families representing multiple racial/ethnic groups. In our sample, on the one hand, 1,020 fathers lived together with their child and their child’s mother from childbirth to middle childhood, but 703 fathers were resident at childbirth but nonresident with their child and the child’s mother in middle childhood. On the other hand, 154 fathers were nonresident at childbirth but resident with their child and the child’s mother in middle childhood, but 462 fathers were nonresident at childbirth as well as in middle childhood. Descriptive statistics of key variables are presented in Table 1 and demographic characteristics are presented in Appendix A.

Table 1.

Descriptive Statistics and Intercorrelations

	F3PE	F5PE	F9PE	PRCb	PSCb	MGO1	TGR	ChG	FirC	ChT1	Black	Hisp	Oth	FAgeb	FEdub	Pov1	Fue1	M (SD)
F3PE	1.00																	4.44 (1.40)
F5PE	.52***	1.00																3.67 (1.44)
F9PE	.34***	.40***	1.00															2.17 ( .84)
PRCb	.04	.06**	.05*	1.00														2.91 ( .30)
PSCb	.06*	.12***	.07**	.13***	1.00													11.21 (1.27)
MGO1	.22***	.17***	.13***	.05*	.05	1.00												3.73 ( .42)
TGR	−.02	−.04	−.01	.001	−.02	.02	1.00											1.98 ( .55)
ChG	.02	.02	.08**	−.02	−.001	−.04	−.01	1.00										.52 ( .50)
FirC	.05*	.03	.03	.02	.04	−.03	−.05*	−.001	1.00									.35 ( .48)
ChT1	−.04	−.06*	−.08**	−.03	−.06**	−.14***	.02	−.01	−.05*	1.00								2.96 (1.03)
Black	−.06**	−.10***	−.02	.03	−.06**	−.11***	−.08***	.002	−.11***	.10***	1.00							.47 ( .50)
Hisp	−.01	.01	−.06*	.05*	−.04	.06*	.14***	−.01	.03	−.01	−.55***	1.00						.25 ( .43)
Oth	.01	.01	.02	−.02	−.01	.02	.03	.03	.02	.01	−.19***	−.12***	1.00					.04 ( .20)
FAgeb	.02	−.02	.05*	−.06**	.01	.10***	.07***	−.01	−.24***	−.03	−.10***	−.10***	.04	1.00				28.12 (7.14)
FEdub	.05*	.06*	.15***	−.06**	.14***	.07**	−.07***	−.02	.07***	−.10***	−.14***	−.23***	.08***	.34***	1.00			2.22 (1.01)
Pov1	−.11***	−.09***	−.16***	.04	−.14***	−.12***	.10***	.01	−.12***	.10***	.25***	.13***	−.06**	−.26***	−.50***	1.00		3.00 (1.41)
Fue1	−.01	−.004	−.06*	.01	−.09***	−.10***	−.03	−.01	−.08***	.08***	.16***	−.04	.01	−.12***	−.23***	.25***	1.00	.18 ( .38)

^***p < .001

^**p < .01

^*p < .05.

Note. F3PE = FPE at child age 3. F5PE = FPE at child age 5. F9PE = FPE at child age 9. PRC = Paternal role-level centrality. PSC = Paternal status-level centrality. MGO = Maternal gate opening. TGR = Traditional Gender Role Beliefs. ChG = Child gender. FirC = First biological child. ChT = Child temperament. Black = Whether father’s race was Black. Hisp = Whether father’s race was Hispanic. Oth = Whether father’s race was others. FAge = Father’s age. FEdu = Father’s education. Pov = Family poverty. Fue = Father’s unemployment. The subscript b = at childbirth. The subscript 1 = at child age 1 year old.

Measures

Fathers’ positive engagement.

Fathers reported the frequency with which they participated in specific developmentally appropriate activities each week at child age 3, 5, and 9 years old. Based on Lamb et al.’s (1985, 1987) conceptualization of father involvement and Palkovitz’s (1997) extended definition of involvement, FFCWS developed the items for FPE as a part of the Developing a Daddy Survey collaboration (DADS; Cabrera et al., 2004). Given that developmentally appropriate activities change with child age, the items assessing FPE were child-age-specific (Cabrera et al., 2004). To check whether FPE at all three waves measured the same factor (overall FPE), we conducted a second-order confirmatory factor analysis with two levels of factor loadings: The first level showed the factor loadings of observed items at each wave for a single factor of FPE at the wave (α = .84 for 7 items at 3 years old, α = .83 for 12 items at 5 years old, and α = .86 for 10 items at 9 years old), and the second level showed the factor loadings of the latent variables of FPE at each of the three waves for overall FPE. The standardized factor analysis results showed high factor loadings. The factor loadings for the overall FPE from three latent variables of FPE at child age 3, 5, and 9 years old were .74 (p < .001), .90 (p < .001), and .56 (p < .001), respectively, supporting the notion that even though FPE was assessed using different items across three waves, these items appeared to be measuring the same construct (χ²(398) = 2467.00, p < .001, RMSEA = .05, CFI = .90, SRMR = .04). The original scale ranged from 0 to 7 days per week at child age 3 and 5 years old, and from 0 not once in past month to 4 every day at child age 9 years old. In this study, standardized scores were created after averaging all items at each wave.

Paternal status-level centrality.

At childbirth, fathers were asked how much they agreed with three items: “Being father is one of the most fulfilling experiences for a man,” “I want people to know I have a new child,” and “Not being a part of child’s life would be one of the worst things.” The scale for scoring each item ranged from 1 (strongly disagree) to 4 (strongly agree). We aggregated scores on the three items (α = .72). A higher score reflects higher fathers’ status-level centrality. A standardized score was computed for analysis.

Paternal role-level centrality.

For paternal role-level centrality, fathers’ response about how much it is important to him to provide direct care, such as feeding, dressing, and child care as a father was used in this study. The original scale ranged from 1 very important to 3 not important, but we reverse coded the item so a higher score indicated a father’s greater caregiving role-level centrality. In FFCWS, this item asked twice: once at baseline and again at the fifth wave (9 years old). However, the vast majority of fathers (88%) reported the same answer at both waves. In this study, we included this item measured at the baseline only to predict FPE from birth to 9 years old. A standardized score was used in analyses.

Maternal gate opening behavior.

For maternal gate opening behavior, the variable of maternal gatekeeping—encouragement (Fagan & Cherson, 2017) was used. Fathers reported how much mothers supported their parenting behaviors at each wave from child age 1. The following four items were used at each wave: ‘She respects the schedules and rules you make for (child)’, ‘She supports you in the way you want to raise (child)’, ‘When (mother) is with (child), she acts like the mother you want for (child)’, and ‘You and (mother) talk about problems that come up with raising (child)’. The original scale of each item ranged from 1 always true to 4 never true, but we reverse-coded to 1 never true to 4 always true and constructed maternal gate opening at each wave by averaging these items (α = .72, .71, and .85 at child age 1, 3, and 5 years old, respectively). To test the longitudinal associations between maternal gate opening behavior and FPE, we created standardized scores for maternal gate opening at child age 1, 3, and 5 years old and treated them as time-varying variables at time T-1 to predict FPE at time T.

Covariates

Fathers’ residential status.

We included the covariate of fathers’ residential status in each group’s analysis to account for fluctuations in family structures across waves. For the group from nonresident- to nonresident-father families, this variable indicates whether a father consistently lived apart from his child and the child’s mother across five waves (1) or not (0). On the contrary, for the group from resident- to resident-father families, this variable indicates whether a father lived with his family consistently across the five waves (1) or not (0). However, for the group from nonresident- to resident-father families and for the group from resident- to nonresident-father families, this variable indicates his concurrent residential status (1 = living together, 0 = not living together) at each wave, and it was treated as a time-varying covariate.

Traditional gender role beliefs.

Mothers reported traditional beliefs about gender roles in families at childbirth. This assessment was developed and incorporated as a part of the DADS collaboration (Cabrera et al., 2004). In FFCWS, three items were used: “It is much better for everyone if the man earns the main living and the woman takes care of the home and family”, “When there are children in the family, parents should stay together even if they don’t get along”, and “The important decisions in the family should be made by the man of the house” (α = .65). The scale for each item ranged from 1 strongly disagree to 4 strongly agree, indicating that a higher score meant mother’s more traditional gender role beliefs. A standardized score was computed for use in subsequent analyses.

Child gender.

Child gender reported by mother was coded as 0 for girls and 1 for boys.

First child or not.

Fathers reported the child’s birth order at the one-year follow-up. If the focal child was the father’s first biological child, it was coded as 1. If not, it was coded as 0.

Child temperament.

Each father reported their focal child’s temperament when the child was 1 year old using three items: “(Child) often fusses and cries,” “(Child) gets upset easily,” and “(Child) reacts strongly when upset.” Fathers scored each item from 1 (not at all like my child) to 5 (very much like my child). A mean score was created (α = .61), and the standardized score was used as a covariate.

Father’s race/ethnicity.

When the focal child was born, fathers reported their race/ethnicities using four categories. Based on this, we set White fathers as a reference group and created three binary variables (i.e., Black, Hispanic, and others (1 = yes, 0 = no)).

Father’s age.

Fathers reported their age at childbirth. This numeric variable was standardized and treated as a time-invariant covariate in our multilevel modeling.

Father’s education.

Father’s education was reported at each wave from 1 less than high school to 4 college or graduate school. Its standardized scores at child age 1, 3, 5 years old were treated as time-varying covariates.

Family poverty ratio.

In FFCWS, family poverty was constructed at each wave to indicate the ratio of household income to the poverty thresholds, considering household size and composition. Subsequently, it was transformed into a categorical variable. The original scale ranged from 1 0–49% to 5 300% or more, but we reverse coded it such that a higher score indicated more severe poverty. Standardized family poverty variables at child age 1, 3, and 5 years old were treated as time-varying covariates.

Father’s unemployment.

From child age 1 year to 5 years old, fathers were asked if they did any regular work for pay and received a regular paycheck during the previous week. We created a binary variable to indicate father’s unemployment status at each wave and treated it as a time-varying covariate at T-1 to predict FPE at time T.

Analysis Plan

To test the longitudinal associations of father identity and maternal gate opening with later FPE, we used two-level multilevel modeling for longitudinal data. In these data, repeated measures were nested within individual fathers. These are the basic equations for the longitudinal multilevel model to predict FPE:

$${\text{Level}1:\text{FPE}}_{ti}={\pi }_{0i}+{\pi }_{1i}{(ChildAge)}_i+e_{ti}$$

$$\begin{gathered} \text{Level}2:{\pi }_{0i}={\beta }_{00}+\sum _{q=1}^{Q_0}{\beta }_{0q}{X}_{qi}+\sum _{p=2}^P{{\pi }^0}_{pi}{a}_{ti}^p+r_{0i}\text{} \\ {\pi }_{1i}={\beta }_{10}+\sum _{q=1}^{Q_1}{\beta }_{1q}{X}_{qi}+\sum _{p=2}^P{{\pi }^1}_{pi}{a}_{ti}^p+r_{1i} \end{gathered}$$

In this model, subscript i indicates each individual and subscript t indicates measurement occasions. Let say π_0i refers to the initial FPE for a person i and π_1i refers to the change of FPE by child age. X_qi are time-invariant predictors and a^p are time-varying predictors. Then, β₀₀ is the average FPE across all individuals, β_0q is the direct effects of time-invariant predictors on FPE_ti, π⁰_pi is the direct effects of time-varying predictors on FPE_ti, and individual variability (r_0i). Regarding the change of FPE by child age (π_1i), β₁₀ is the direct effect of child age, β_1q indicates a moderating effect of a time-invariant predictor on the association between child age and FPE, π¹_pi indicates a moderating effect of a time-varying predictor on the association between child age and FPE, and r_1i is individual variation in the rate of FPE change. We call β_00, β_0q, β_10, β_1q, π⁰_pi, and π¹_pi as fixed effects and r_0i and r_1i as random effects.

The key predictors in this study were paternal identity and maternal gate opening behavior. Paternal status-level and role-level centrality were treated as time-invariant predictors. However, maternal gate opening behavior was treated as a time-varying predictor because this variable reflected the mother’s support for the father’s parenting rules and behaviors and was measured repeatedly since the focal child was 1 year old. In addition to these key predictors, we also included time-varying covariates (e.g., father’s unemployment status) and time-invariant covariates (e.g., father’s race/ethnicity, child gender) to reduce error in estimation. To handle missing data, maximum likelihood estimation was used. We analyzed the data in STATA 17.

Results

The correlations between key predictors and covariates are shown in Table 1. Paternal identity at childbirth was positively but weakly correlated with FPE in early to middle childhood (r = .04~.12). Correlations between maternal gate opening at 1 year old and FPE in early to middle childhood were higher (r = .13~.22) compared to the correlations between paternal identity and FPE. Paternal role-level and status-level centrality were weakly but positively correlated with each other (r = .13). FPE between age 3 and 9 was moderately correlated (r = .34), but FPE at wave T and the previous wave T-1 were strongly correlated (r = .40~.52). In terms of demographic characteristics, overall, fathers who had more education (r = .05~.15) and experienced less poverty (r = −.09~−.16) tended to have higher FPE in early to middle childhood.

Nonresident- to Nonresident-Father Families

The longitudinal multilevel results for the associations between paternal identity or maternal gate opening and FPE, before testing the interactions between paternal identity and maternal gate opening, are shown in Table 2. In Model 1 for the families whose fathers did not live together with their child and the child’s mother at childbirth and in middle childhood as well, neither dimension of paternal identity nor maternal gate opening were significantly associated with FPE. However, although the direct effects of paternal identity and maternal gate opening were not significant, when adding the interaction between fathers’ role-level centrality and maternal gate opening and the interaction between fathers’ status-level centrality and maternal gate opening to Model 1, the results demonstrated that the associations between fathers’ status-level centrality and later FPE differed by maternal gate opening (see Figure 1, and Appendix B). When maternal gate opening was high (M + 1SD), the simple slope of paternal status-level centrality on FPE was statistically significant (γ = .14, p < .05) but the simple slope of fathers’ status-level centrality became nonsignificant when maternal gate opening was average (M) or low (M − 1SD). That is, when maternal gate opening was at a high level, fathers with higher status-level centrality spent more time in FPE with children.

Table 2.

Longitudinal Multilevel Models Predicting Fathers’ Positive Engagement (N = 2,339)

	FPE
	Model 1: Nonresident → Nonresident (n = 462) Coeff. (S.E.)	Model 2: Nonresident → Resident (n = 154) Coeff. (S.E.)	Model 3: Resident → Nonresident (n = 703) Coeff. (S.E.)	Model 4: Resident → Resident (n = 1,020) Coeff. (S.E.)
Paternal Role-Level Centrality (F1)	−.03 (.06)	−.02 (.08)	.04 (.04)	.07 (.03) *
Paternal Status-Level Centrality (F2)	.04 (.05)	.11 (.08)	−.03 (.03)	.07 (.03) *
Maternal Gate Opening (M)	.03 (.05)	.02 (.07)	.09 (.03) **	.07 (.02) **
Child Age	.01 (.04)	−.07 (.05)	.20 (.03) ***	.01 (.02)
(F1) x Child Age	−.004 (.05)	.11 (.05) *	.02 (.03)	.01 (.02)
(F2) x Child Age	.08 (.04)	−.08 (.05)	.02 (.03)	−.02 (.02)
(M) x Child Age	−.04 (.05)	.09 (.05)	.004 (.03)	−.09 (.02) ***
Covariates
Residential Status a	−.44 (.10) ***	.60 (.15) ***	.76 (.07) ***	.24 (.11) *
Family Poverty	−.02 (.05)	−.05 (.06)	−.01 (.04)	.02 (.03)
Father’s Unemployment	.001 (.09)	−.01 (.13)	−.01 (.08)	.09 (.07)
Father’s Education	.13 (.05) *	.002 (.07)	−.02 (.04)	−.05 (.04)
Father’s Age at Childbirth	−.08 (.05)	−.18 (.07) *	−.04 (.04)	−.02 (.03)
Father’s Race/Ethnicity: Black	.10 (.22)	.20 (.30)	−.06 (.09)	.07 (.07)
Father’s Race/Ethnicity: Hispanic	.10 (.26)	−.09 (.33)	−.17 (.10)	−.12 (.08)
Father’s Race/Ethnicity: Others	.07 (.33)	−.45 (.61)	.06 (.24)	.08 (.13)
Child Gender (1 = Boy)	−.02 (.10)	.05 (.14)	.11 (.07)	.10 (.06)
First Biological Child or Not (1 = 1st child)	−.01 (.11)	.11 (.15)	−.001 (.08)	.17 (.06) **
Child Temperament	.13 (.05) *	.06 (.07)	−.01 (.04)	−.08 (.03 ) **
Mother’s Traditional Gender Beliefs	−.001 (.05)	.02 (.07)	.02 (.04)	−.13 (.03) ***

^***p < .001

^**p < .01

^*p <. 05.

Note. Italics indicate time-varying variables.

^a.The variable of fathers’ residential status in Model 1 and Model 4 indicates the stability of nonresident-/resident-father families across all five waves (1 = stably nonresident for Model 1; 1 = stably resident for Model 4), while for Model 2 and Model 3 this variable indicates fathers’ concurrent residential status at each wave (1 = living together for Models 2 and 3).

Figure 1.

Paternal status-level centrality, maternal gate opening, and fathers’ positive engagement among nonresident- to nonresident-father families

Note. A black line indicates a simple slope which is significantly different from zero while grey lines indicate nonsignificant simple slopes.

Nonresident- to Resident-Father Families

In Model 2 for the families whose fathers did not live together with his child and the child’s mother at childbirth but were living together in middle childhood, the intercept of FPE did not differ by paternal role-level centrality, paternal status-level centrality, or maternal gate opening (Table 2). However, paternal role-level centrality was significantly associated with the slope of FPE (Figure 2; γ = .11, p <.05). Simple slopes analysis showed that fathers with low role-level centrality had a significantly decreasing FPE slope as their child developed (p < .01), while the slopes of FPE were not significantly changed over time for fathers with average or high role-level centrality. When adding the interactions between paternal identities and maternal gate opening to Model 2 (Appendix B), the results indicated that maternal gate opening moderated the association between paternal status-level centrality and FPE (Figure 3; γ = .16, p < .05). Simple slopes analysis showed that paternal status-level centrality was positively associated with FPE when maternal gate opening was high (p < .05), while paternal status-level centrality and FPE were not significantly related when maternal gate opening was average or low.

Figure 2.

Child age, paternal role-level centrality, and fathers’ positive engagement among nonresident- to resident-father families

Note. A black line indicates a simple slope which is significantly different from zero while grey lines indicate nonsignificant simple slopes.

Figure 3.

Paternal status-level centrality, mothers’ gate opening, and fathers’ positive engagement among nonresident- to resident-father families

Note. A black line indicates a simple slope which is significantly different from zero while grey lines indicate nonsignificant simple slopes.

Resident- to Nonresident-Father Families

In Model 3 for the families whose fathers lived together with his child and the child’s mother at childbirth but were not living together in middle childhood, FPE significantly increased over time as their children developed (γ = .20, p < .001). Maternal gate opening was positively associated with FPE (γ = .09, p < .01) and this positive association between maternal gate opening and FPE did not change by child age (γ = .004, p > .05) (Table 2). When testing the interactions between paternal identity and maternal gate opening, there were no significant interactions found between them (Appendix B).

Resident- to Resident-Father Families

In Model 4 for the families whose fathers lived together with his child and the child’s mother at childbirth and in middle childhood, paternal role-level centrality (γ = .07, p <. 05) and paternal status-level centrality (γ = .07, p < .05) were positively associated with FPE. Maternal gate opening was also positively associated with FPE (γ = .07, p <. 01) (Table 2). But the association between maternal gate opening and FPE differed by child age (Figure 4). Simple slopes analysis demonstrated that higher maternal gate opening was significantly associated with FPE when the children were in early childhood (p < .001 at 3 years old; p < .01 at 5 years old). In middle childhood, the graph showed the decreasing FPE slope by maternal gate opening but simple slope analysis demonstrated that this slope was statistically not different from zero (p > .05 at 9 years old). There were no significant interactions between paternal identity and maternal gate opening when predicting FPE within this group.

Figure 4.

Maternal gate opening, child age, and fathers’ positive engagement among resident- to resident-father families

Note. Black lines indicate the simple slopes which are significantly different from zero while a grey line indicates the nonsignificant simple slope.

Discussion

The central objective of this study was to test relations of paternal identity, maternal gate opening, and their interaction with FPE from early to middle childhood. Results demonstrated the positive roles of paternal identities and mothers’ gate opening behaviors in supporting later FPE. However, the detailed interplay among paternal identity and maternal gate opening in predicting FPE varied by fathers’ residential contexts. The protective effect of maternal gate opening tended to be consistent across child age except for resident-to-resident father families, and this consistent protective effect was even more apparent among the fathers who had a high status-level centrality but did not live together with their child and child’s mother at childbirth.

For families whose fathers were nonresident at childbirth, understanding the dynamic between paternal identity and maternal gate opening was especially critical to explaining subsequent FPE. Specifically, both among nonresident-to-resident father families and among nonresident-to-nonresident father families, fathers who had a high status-level centrality—those who strongly valued fatherhood—were more likely to engage in FPE during later developmental periods when they were supported by a high level of maternal gate opening. It was interesting to note that the direct effects of paternal role-level and status-level centrality as well as maternal gate opening were not significant in predicting FPE among these families before including the interactions between paternal identity and maternal gate opening. However, when we added the interactions between paternal identity and maternal gate opening, the results exhibited significant interactions between paternal status-level centrality and maternal gate opening to predict FPE. Moreover, these significant associations did not differ by child age.

In previous research about nonresident fathers’ FPE, maternal encouragement at 3 years old predicted higher FPE at 5 years old, but maternal facilitation (i.e., seeking more FPE but with less support) at 3 years old predicted lower FPE at 5 years old (Fagan & Cherson, 2017). In our research, simple slopes analyses demonstrated protective effects connected to high levels of maternal gate opening on FPE from 3 to 9 years old for fathers who were nonresident at their child’s birth, which extends the findings of Fagan and Cherson (2017). As Edin et al. (2009) addresses, many nonresident fathers maintain a high level of involvement after a nonmarital birth. Building upon this, our findings shed light on the possibility that FPE among parents experiencing a nonmarital birth can be strengthened by both supporting fathers in developing a strong general sense of their value and by helping couples establish cooperative coparenting relationships characterized by maternal gate opening.

For fathers who lived together with their child during middle childhood, both paternal role-level centrality—how much fathers viewed active caregiving as central to their role—and paternal status-level centrality—how much fathers valued becoming a father— were important predictors of FPE. As we anticipated, fathers with higher status-level centrality engaged more when mothers showed greater gate opening behaviors. Interestingly, however, paternal role-level centrality was a significant predictor of FPE beyond the change in fathers’ residential status. For example, among fathers who were nonresident at childbirth but lived together with their family at middle childhood, fathers with low role-level centrality tended to show a significant decrease in FPE. The simple slopes results demonstrated that there was no change in FPE among fathers with high role-level centrality from early to middle childhood. Rather, the FPE of fathers with low role-level centrality significantly decreased as their children matured even though their residential status had changed from nonresident to resident. This result supports Bocknek’s (2020) ideas by demonstrating that fathers’ residential status does not guarantee the amount of FPE. Rather, in this research, it was paternal role-level centrality that predicted FPE.

This begs the question to us: why did these fathers have lower FPE after their reunion? One potential explanation is that fathers with low caregiving role-level centrality might have found their primary paternal role in other areas such as providing financial security with a higher salary. Previous research (Goldberg, 2015) demonstrated the effects of paternal identity and FPE were moderated by fathers’ residential status, but this study only tested status-level centrality and did not test role-level centrality. In Goldberg (2015), fathers’ residential contexts consisted of only three groups: constantly resident, constantly nonresident, and divorced/separated, and as a result, the effect of paternal identity on FPE was nonsignificant after separation in that study. So, our finding for families whose fathers changed their residential status from nonresident to resident makes a unique contribution. Our study brings a more interesting discussion on FPE by showing that paternal role-level centrality is a significant predictor of FPE, and that the effect of this paternal role-level centrality becomes greater as children develop even among families in which fathers shift to resident from nonresident.

For the fathers who were resident to resident from childbirth to middle childhood, the direct effects of paternal role-level centrality, status-level centrality, and maternal gate opening were significant in predicting FPE. This is consistent with previous research that reports positive associations between paternal identity or maternal gate opening and FPE among resident-father families (McBride et al., 2005; Schoppe-Sullivan et al., 2008; Zvara et al., 2013). Given that these previous studies collected data on FPE during infancy or early childhood, our research extends relevant discussions on FPE by analyzing FPE data collected from early to middle childhood. Previous research has reported significant interactions between paternal identity and maternal gatekeeping in predicting FPE in resident-father families (McBride et al., 2005). However, in this study, paternal role-level and status-level centrality as well as maternal gate opening all were positively associated with FPE. Whereas all direct effects of paternal identity and maternal gate opening were statistically significant on FPE, there was no significant interaction between paternal identity and maternal gate opening among these resident-to-resident-father families.

Although the findings of this study are consistent with previous research reporting the positive association between maternal gate opening and FPE (e.g., Schoppe-Sullivan et al., 2008; Zvara et al., 2013), we found that the longitudinal effect of maternal gate opening gradually decreased over time among resident-to-resident father families. This is somewhat different from the results found among other families. For example, there were no significant changes in direct and moderation effects of maternal gate opening among nonresident-to-nonresident or nonresident-to-resident father families. However, among these resident-to-resident fathers, our simple slopes results revealed that maternal gate opening became a nonsignificant predictor of FPE when children were 9 years old. Why did this happen? In resident-father families, there are arguably fewer family transitions and therefore less disruption to established patterns. Thus, although maternal gate opening might be important for promoting FPE when children are younger, its influence might fade as families function to maintain homeostasis (Minuchin, 1985) in the face of stability in family relationships.

By including diverse family contexts, this study contributes to the extension of our knowledge and understanding of the dynamics between paternal identity, maternal gate opening, and FPE. Of course, however, the results of this study need to be interpreted and applied with caution given its limitations. First, since this study analyzed secondary data, the measurement of some key constructs was limited or lacking entirely. For example, we did not have data on maternal gate closing to compare its role to that of maternal gate opening. In FFCWS, the mother-report item “You are critical of things the father does with the child” was not measured before child age 5. In addition, this item was measured only once, so there is a risk of treating it as if it is a time-invariant predictor of FPE when it is not truly time-invariant within the contexts of child development and fathers’ changing residential statuses. Future research should collect data on maternal gate closing and gate opening measured at the same time points to yield a better understanding of how maternal gatekeeping interacts with paternal identity to predict FPE across children’s development. Second, we were not able to use child-report or observational data. This study primarily utilized fathers’ self-report survey data. We chose to use father-report data because it offers a unique advantage in analyzing the dynamic between paternal identity and perceived maternal support in predicting FPE when the majority of previous research has relied on mothers’ reports only. In addition to having father-report data, we think it would be ideal to have observational data on the quality of FPE or children’s perspectives on the quality and closeness of the father-child relationship.

Despite several limitations, our results underscore that FPE is multiply determined by important factors including paternal identity, maternal gate opening, and the child’s development, which is consistent with identity theory (Adamsons, 2010) and models of fathering (Cabrera et al., 2014). Findings from this study highlight the importance of encouraging expectant and new fathers to forge a strong paternal identity and thus motivate their involvement in parenting. In addition, supportive coparenting relationships, characterized by maternal gate opening behavior, appear critical to FPE in early childhood, and persist in importance particularly for highly motivated nonresident fathers. More research with expanded measures of paternal identity, maternal gatekeeping, and father-child interactions and relationships is necessary to further reveal how the dynamics between paternal identity and maternal gatekeeping change with child age in relation to fathers’ parenting in families 6in diverse family contexts. Through such attempts, we hope to obtain a better understanding of the mechanisms underlying individual differences in FPE and determine how to support developmentally appropriate and high-quality involved fathering across children’s development.

Appendix

Table A.

Demographic Characteristics of the Sample (N = 2,339)

	N (%)		N (%)
Father’s Age (w1)		Mother’s Age (w1)
10’s	178 ( 7.6)	10’s	369 (15.8)
20’s	1,267 (54.2)	20’s	1,386 (59.3)
30’s	720 (30.8)	30’s	530 (22.7)
40’s +	174 ( 7.4)	40’s +	54 ( 2.3)
Father’s Education (w1)		Mother’s Education (w1)
Less than high school	682 (29.2)	Less than high school	693 (29.7)
High school or equivalent	774 (33.1)	High school or equivalent	700 (30.0)
degree		degree
Some college or technical	576 (24.6)	Some college or technical	616 (26.4)
schools		schools
College or graduate	307 (13.1)	College or graduate school	327 (14.0)
school
Father’s Race/Ethnicity (w1)		Mother’s Race/Ethnicity (w1)
White	547 (23.4)	White	592 (25.4)
Black	1,109 (47.4)	Black	1,046 (44.8)
Hispanic	590 (25.2)	Hispanic	604 (25.9)
Other	93 ( 4.0)	Other	92 ( 3.9)
Child Gender		Birth Order
Boy	1,225 (52.4)	First Child	1,519 (65.1)
Girl	1,114 (47.6)	Not First Child	815 (34.9)
Father’s Residential Status (w1)		Father’s Residential Status (w5)
Resident	616 (26.3)	Resident	1,165 (49.8)
Nonresident	1,723 (73.7)	Nonresident	1,174 (50.2)
Father’s Employment Status (w1)		Father’s Employment Status (w4)
Employed	1,943 (83.1)	Employed	1,683 (82.8)
Unemployed	394 (16.9)	Unemployed	349 (17.2)

Table B.

Longitudinal Multilevel Models that Include Interaction Terms Predicting Fathers’ Positive Engagement (N = 2,339)

	FPE
	Model 1: Nonresident → Nonresident (n = 462) Coeff. (S.E.)	Model 2: Nonresident → Resident (n = 154) Coeff. (S.E.)	Model 3: Resident → Nonresident (n = 703) Coeff. (S.E.)	Model 4: Resident → Resident (n = 1,020) Coeff. (S.E.)
Paternal Role-Level Centrality (F1)	−.02 (.06)	−.004 (.08)	.05 (.04)	.07 (.03) *
Paternal Status-Level Centrality (F2)	.02 (.05)	.09 (.08)	−.03 (.03)	.07 (.03) *
Maternal Gate Opening (M)	.01 (.05)	−.01 (.07)	.10 (.03) **	.07 (.02) **
Child Age	.002 (.04)	−.07 (.05)	.20 (.03) ***	.01 (.02)
(F1) x (M)	−.08 (.06)	−.04 (.08)	.01 (.03)	.004 (.02)
(F2) x (M)	.12 (.06) *	.16 (.08) *	.02 (.03)	−.02 (.02)
(F1) x Child Age	−.01 (.05)	.11 (.05) *	.02 (.03)	.01 (.02)
(F2) x Child Age	.07 (.05)	−.07 (.05)	.02 (.03)	−.02 (.02)
(M) x Child Age	−.04 (.05)	.08 (.05)	.006 (.03)	−.09 (.02) ***
(F1) x (M) x Child Age	.06 (.05)	.01 (.05)	.04 (.03)	−.01 (.02)
(F2) x (M) x Child Age	.02 (.06)	.04 (.07)	−.002 (.03)	−.001 (.02)
Covariates
Residential Status a	−.44 (.10) ***	.56 (.15) ***	.77 (.07) ***	.24 (.11) *
Family Poverty	−.02 (.04)	−.05 (.06)	−.01 (.04)	.02 (.03)
Father’s Unemployment	−.03 (.09)	−.01 (.13)	−.01 (.08)	.09 (.07)
Father’s Education	.13 (.05) *	.00 (.07)	−.02 (.04)	−.05 (.04)
Father’s Age at Childbirth	−.07 (.05)	−.18 (.07) *	−.04 (.04)	−.02 (.03)
Father’s Race/Ethnicity: Black	.13 (.22)	.12 (.30)	−.06 (.09)	.07 (.07)
Father’s Race/Ethnicity: Hispanic	.13 (.26)	−.16 (.32)	−.17 (.10)	−.12 (.08)
Father’s Race/Ethnicity: Others	.10 (.33)	−.56 (.60)	.06 (.24)	.08 (.13)
Child Gender (1 = Boy)	−.02 (.10)	.03 (.14)	.11 (.07)	.10 (.06)
First Biological Child or Not (1 = 1st child)	.00 (.11)	.08 (.15)	−.003 (.08)	.17 (.06) **
Child Temperament	.13 (.05) *	.06 (.07)	−.01 (.04)	−.08 (.03) **
Mother’s Traditional Gender Beliefs	.002 (.05)	.03 (.07)	.02 (.04)	−.13 (.03) ***

^***p < .001

^**p < .01

^*p <. 05.

Note. Italics indicate time-varying variables.

^a.The variable of fathers’ residential status in Model 1 and Model 4 indicates the stability of nonresident-/resident-father families across all five waves (1 = stably nonresident for Model 1; 1 = stably resident for Model 4), while for Model 2 and Model 3 this variable indicates fathers’ concurrent residential status at each wave (1 = living together for Models 2 and 3).