Fathers’ Criminal Behavior and Involvement With Children: The Moderating Role of Family Relationships

Abstract

Objective

Research indicates that fathers’ criminal behavior can be problematic for children through multiple pathways, yet few studies have examined the effect of fathers’ kinship networks in this process. This study examines the association between fathers’ criminal behavior and involvement with their children and the extent to which a father’s relationships with individuals in his extended family network moderate this association.

Method

Hierarchical linear modeling was used to predict fathers’ involvement using data from a longitudinal intergenerational study of 335 children and 149 low-income, minority fathers. Measures included 8 father-involvement outcomes, a measure of fathers’ criminal behavior, and 2 moderator variables.

Results

High-quality relationships between fathers and their male relatives moderated the negative effect of criminal behavior on measures of fathers’ involvement. Criminal behavior was only associated with decreasing levels of father involvement when fathers had low-quality relationships with male relatives.

Conclusions

Strong and affirmative relationships—with male relatives specifically—may attenuate the adverse effects of antisocial and criminal behavior on fathers’ involvement in at-risk families. Implications for tailoring practice to improve relationships between fathers and male relatives and to enhance fathers’ prosocial involvement are noted.

Despite research assessing fathering from various perspectives (Cabrera & Tamis-Lemonda, 2013), there is still little known about the extent to which fathers differentially contribute to children’s well-being and the factors that promote and inhibit positive father involvement (Coley, 2001; Coley & Chase-Lansdale, 1999). Criminal behavior and its relation to father involvement is one area of research important to understanding the complexity of fathering. Research with fathers who are engaged in criminal activities has found that their children are at high risk for poor outcomes, including conduct problems, substance abuse, mental health disorders, and poor academic achievement (Johnson, 2009; Murray & Farrington, 2005). A growing body of evidence shows that criminal and antisocial behavior among fathers can be problematic for children through several pathways, including decreased father involvement (Fagan & Lee, 2012), ineffective parenting practices (Capaldi, Pears, Patterson, & Owen, 2003; Thornberry, Freeman-Gallant, Lizotte, Krohn, & Smith, 2003), and heritability (Jaffee, Moffitt, Caspi, & Taylor, 2003). In the current study, we focused on just one of these pathways—father involvement—to assess its association with criminal behavior. We used multidimensional measures of father involvement that go beyond constructs of paternal involvement with children that have focused primarily on contact, residence, and child support.

In addition to our focus on the link between fathers’ criminal activity and involvement with children, a central purpose of this study was to build and test models that include protective factors that may mitigate the negative association between criminal behavior and father involvement. As such, this study examined fathers’ criminal behavior and its association with a range of fathering activities, and it assessed moderating factors in that process—specifically, fathers’ relationship quality with and fathering support received from family members. Our aim is to contribute to the knowledge base, thereby promoting the development and testing of targeted preventive interventions that support positive fathering.

Background

Criminal and Antisocial Behavior

Much of the empirical evidence about criminal and antisocial behavior and its effects on children has focused on the transmission of such behavior from one generation to the next (Capaldi et al., 2003; Capaldi, Pears, Kerr, Owen, & Kim, 2012; Kim, Capaldi, Pears, Kerr, & Owen, 2009; Vaughn, Salas-Wright, DeLisi, & Qian, 2015) with limited information about its association with father involvement. Although some research suggests that increased criminal behavior among fathers is linked to lower levels of involvement with their children (i.e., less time spent living with them) and to an increased risk of living apart from them altogether (Jaffee, Caspi, Moffitt, Taylor, & Dickson, 2001), other research has shown no significant relation to father involvement (Coley & Hernandez, 2006). A related set of studies examining fathers’ antisocial behavior and child outcomes found that in cases of high levels of father–child contact, effective or high-quality parenting mediates the potentially adverse effect of fathers’ antisocial behavior on children’s outcomes (Thornberry, Freeman-Gallant, & Lovegrove, 2009a) and mitigates the intergenerational transmission of antisocial behavior from father to child (Thornberry, Freeman-Gallant, & Lovegrove, 2009b).

Evidence suggests that one possible pathway between earlier antisocial behavior and later father involvement is the stability of antisocial behavior over time such that children with behavioral difficulties may continue these patterns in adulthood (Fergusson, Horwood, & Ridder, 2005). Such patterns could result in inadequate parenting that negatively affects father involvement directly or leads to adverse romantic relationship outcomes—such as intimate partner violence—that in turn affect fatherhood (Jaffee, Belsky, Harrington, Caspi, & Moffitt, 2006). Fathers who experienced antisocial behavior during their own development could have problematic relationships with the mother of their child later in life, leading to lower levels and quality of father involvement.

A related pathway draws from the adolescence-limited antisocial behavior model (Caspi & Moffitt, 1995). In this case, adolescents engage in delinquent behavior for some time but then gradually return to conventional behavior as they transition to adulthood. This transition may be delayed if young men are arrested or incarcerated (Moffitt, 2015). For these young men who simultaneously enter fatherhood, their antisocial behavior may negatively influence their involvement with children and create instability in the family system.

Relationship Quality

The quality of relationships between mothers and fathers is a strong predictor of union outcomes—a central protective factor for children—and correlates positively with father involvement (Carlson, McLanahan, & England, 2004; Charles, Jones, & Guo, 2013). Higher quality relationships between parents are particularly salient, as they are associated with an increase in the amount of time fathers spend with their children and the quality of the father–child relationship (Cabrera, Ryan, et al., 2004; Sobolewski & King, 2005).

There is also some evidence that the quality of relationship between the father and the mother’s family, and the mother and the father’s family, is positively associated with father involvement (Krishnakumar & Black, 2003; Ryan, Kalil, & Ziol-Guest, 2008). Further, fathers’ family-of-origin relationships (i.e., quality of relationship with parents, caretaker changes, having a single and/or teenage mother, experiencing harsh or inconsistent discipline, and parental criminal history) may impact fathers’ involvement with their own children (Jaffee et al., 2001). Fathers’ relationships with their mothers—and getting childrearing assistance and emotional support from their mothers—can be important parenting supports for young fathers (Miller, 1994; Roy, Dyson, & Jackson, 2010). A father’s relationship with his own father may also influence the extent to which he is involved in his child’s life; men who had positive experiences with their own fathers may imitate and model this behavior with their own children (Beaton & Doherty, 2007). Qualitative research also shows that fathers with difficult developmental histories, including separation from their own fathers in childhood, may seek to repair those relationships in adulthood as part of their own engagement process with their offspring (Roy & Lucas, 2006).

Social Support

Social support has been emphasized as crucial to effective parenting (Belsky, 1984), and social networks have the potential to provide psychological and material resources to help people cope with stress (Cohen, 2004), improve mental and physical health (Lin, 1986), and reduce crime (Cullen, 1994). For some families, social support may be critical to enhancing father involvement and the quality of the father–child relationship. Indeed, recent research has found higher rates of social support to be associated with increased paternal involvement and engagement (Castillo & Sarver, 2012; Fagan & Lee, 2011; Roy & Vesely, 2010).

Support from the mothers of their children is hypothesized to positively influence father involvement, although research suggests variation in the precise pathway through which this occurs. Some investigators have found that positive coparenting relationships influence father involvement through parents’ ability to work well together as they jointly raise children and make parenting decisions (Carlson, McLanahan, & Brooks-Gunn, 2008; Sobolewski & King, 2005). Other evidence has suggested that mothers’ support of fathers may operate through a father’s perception that his partner views him as a competent father (Bouchard & Lee, 2000) or from a mother’s belief in the importance of a father’s caregiving status (McBride et al., 2005). Support from other people in a father’s social network (e.g., siblings, extended kin) has also been found to be predictive of greater father involvement (Fagan & Lee, 2011; Roy & Vesely, 2010). Confidant support from different people, including relatives, has also been associated with better parenting, especially among divorced fathers (Degarmo & Forgatch, 2012).

Demographic and Other Predictors of Father Involvement

Previous research on predictors of fathers’ involvement in the lives of their children has indicated that certain characteristics of fathers are associated with the quantity and quality of their involvement with children (Coley & Hernandez, 2006; Gorman-Smith, Hunt, & Robertson, 2012). Fathers’ income, employment status, and education level can influence their participation in raising children (Castillo, Welch, & Sarver, 2013). Studies have also found that relationship status can affect father involvement, with unmarried and nonresident fathers less likely to have regular and high-quality child contact than their married and cohabiting counterparts (Coley & Chase-Lansdale, 1999; Landale & Oropesa, 2001). The number of children can affect fathers’ involvement, as having more children may decrease the time and resources a father can allocate to each child (Carlson & Furstenberg, 2006; Kotila & Kamp Dush, 2012). Although findings are mixed, some research has found that gender and age can be a potential influence, with fathers more involved with sons than with daughters (Lundberg, McLanahan, & Rose, 2007; Manlove & Vernon-Feagans, 2002) and more involved with younger children than with older children (Cooksey & Craig, 1998; Kulik & Sadeh, 2015). Some studies have found that fathers are more engaged when the child is their biological offspring (Hofferth, 2006; Dunn, 2004), although other evidence suggests that the variation in paternal investment according to biological status is reduced once differences between fathers are accounted for (Hofferth & Anderson, 2003).

Current Study

Conceptual Framework

Our conceptualization of father involvement was informed by a synthesis of theoretical perspectives derived principally from the work of Lamb, Pleck, Charnov, and Levine (1987). They proposed three domains as the basis for father involvement: accessibility (father is physically available to child); engagement (father interacts with child through recreational [e.g., playing], educational [e.g., reading], and caregiving [e.g., dressing] activities); and responsibility (father takes on duties related to planning for the child, making financial contributions toward expenses, and influencing decisions about the child).

The models we drew from emphasize the ecology of multiple determinants of parenting. In these models, child characteristics, parental characteristics, and social and contextual factors all play a role and have both positive and negative influences on father involvement (Belsky, 1984; Lamb, Pleck, Charnov, & Levine, 1985). For example, Doherty, Kouneski, and Erickson (1998) developed a conceptual model that emphasized the role of individual (i.e., paternal, maternal, child), relational (i.e., the coparental relationship), and contextual factors (i.e., income, employment, social support). The authors based their model on an ecological framework that emphasized the importance of context—both environmental and interpersonal—in fathering, asserting that the process of fathering is even more contextually sensitive than mothering.

The broad conceptualization of fathering moves beyond previous research suggesting that fathering is primarily defined by financial support and residence with the child. Instead, the growing consensus is that fathering should be considered multidimensionally, with father involvement and parenting skills as two major components (Gorman-Smith et al., 2012). This lies at the center of our conceptual framework. As such, we have proposed a model predicting father involvement, with attention to the role of extended family and mothers in fathering behavior. This model, as seen in Figure S1 (available online), includes fathers’ criminal activity as a primary risk factor associated with lower levels of father involvement. Two sets of protective factors—relationship quality with family members and supportiveness for fathering from family members—moderate the negative effects of fathers’ criminal activity. That is, the negative associations between fathers’ criminal behavior and father involvement will be attenuated in the presence of high-quality relationships and high levels of supportiveness. These risk and protective factors occur within the context of other father and child characteristics that previous research has found to be important to father involvement.

Research Questions

This study examines fathers’ criminal behavior as a risk factor for reduced father involvement and the extent to which relationships with mothers and extended family moderate this association. We address the following research questions:

• Question 1—Is there a negative association between fathers’ criminal behavior and father involvement?

• Question 2—Does support and encouragement for being a father from (a) fathers’ relatives, (b) mothers, and (c) mothers’ relatives attenuate the negative association between fathers’ criminal behavior and father involvement?

• Question 3—Does the quality of the relationship between fathers and (a) fathers’ relatives, (b) mothers, and (c) mothers’ relatives attenuate the negative association between fathers’ criminal behavior and father involvement?

To address Question 1, each of the outcome measures (see the dependent variables section) was regressed on criminal behaviors, controlling for the complete set of covariates (father characteristics and child characteristics). To address Question 2, each of the measures of supportiveness (see the moderating variables section) and their interaction with criminal activity was added to each of the models from Question 1 to test whether support from different family members moderated relations between fathers’ criminal activity and father involvement (e.g., supportiveness from fathers’ male relatives × criminal behavior, supportiveness from fathers’ female relatives × criminal behavior, etc.). To address Question 3, each of the measures of relationship quality (see the moderating variables section) and their interaction with criminal activity were added to each of the models from Question 1 to test whether relationship quality with different family members moderated relations between fathers’ criminal behavior and father involvement.

Method

Data

The data for this study were from eight waves of the Chicago Youth Development Study (CYDS), a longitudinal investigation of delinquent behavior among African American and Latino males from inner-city Chicago, IL, neighborhoods (Tolan, Gorman-Smith, & Henry, 2003). Boys (N = 341) in Grades 5 and 7 at 17 public schools were first interviewed at Wave 1 as children. The boys were interviewed every 1 to 2 years (except for a 3-year gap between Waves 6 and 7) throughout development and into adulthood, resulting in eight waves of interviews. By the seventh and eighth waves of data collection, the study had shifted from a focus on delinquent juvenile behavior to paternity, father involvement, parenting, and child functioning. The current study includes those participants who had become fathers (N = 165), who were interviewed at least once during CYDS Waves 7 and 8, and who answered questions about each of their children separately for up to five children. By the time of Waves 7 and 8, the fathers were between the ages of 20.5 and 30.5 years. Although the focus of the current study is fathers from Waves 7 and 8, we took advantage of the longitudinal nature of CYDS to include relevant control variables from Waves 1–6 in our statistical models.

Study participants in Waves 7 and 8 were asked a series of questions about fatherhood, including their access to, involvement with, and influence on their children’s lives and important aspects of their children’s development. They were also asked to describe their relationship with different individuals and the extent to which these people provided support or encouragement for being a father. In accordance with the original aims of CYDS, fathers were asked to report on their own antisocial behavior and criminal involvement. Father involvement was only included in Waves 7 and 8, hence our use of those waves. The final analytic sample consisted of 149 fathers who reported on 335 of their own children. Fathers who were incarcerated and could not answer relevant survey questions (n =4) and those with missing data on predictor variables (n = 12) were excluded from analyses. This research was approved by the institutional review board of a large university in the Midwestern United States.

Measures

Dependent variables

We included a set of outcome measures based on previous research (Cabrera, Moore, et al., 2004; Cabrera & Peters, 2000). Fathers were asked a series of questions to assess their involvement in three areas (Lamb et al., 1987): accessibility (two items), engagement (three subscales), and responsibility (three subscales). Summary information on the measures follows with additional detail available elsewhere (Charles et al., 2016). We scored the items similarly to methods used by Cabrera, Ryan, et al. (2004) and provided reliability information about the measures using Cronbach’s alpha.

Father accessibility (α = .82) was based on two items:

• “How often can you see the child?” (reverse-coded Likert scale: 1 = any time I want to, 2 = once per day but not anytime I want to, 3 = once per week but not every day, 4 = between once per week and once per month, 5 = less than once per month, 6 = never); and

• “How often do you see the child?” (1 = never, 2 = rarely—maybe once or twice per year, 3 = sometimes—about once a month or so, 4 = often—not every day but several times per week or month, or 5 = every day).

Both continuous items were based on the mean of their respective scales.

Father engagement was assessed with three subscales: (a) recreation (six items; α = .81); (b) education (two items; α = .90); and (c) caregiving (five items; α = .86). Fathers were asked the number of times in the past month they had participated in activities with their child (e.g., play outside, read stories). Reverse-coded items were used to generate mean scores ranging from 1–5 (1 = at least once per day, 2 = a few times per week, 3 = a few times per month, 4 = rarely, or 5 = not at all in the past month).

Father responsibility was measured with three subscales: (a) planning (18 items; α = .93); (b) financial responsibility (eight items; α = .91); and (c) influence on child-related decisions (eight items; α = .90). For planning and financial responsibility, fathers were asked to identify who ensures that various things get done or that services are provided (e.g., keep safe, take to school, get groceries, arrange child care), and who usually pays for things. The 5-category response sets (1 = father, 2 = mother, 3 = both, 4 = someone else, or 5 = no one) were collapsed into single-indicator variables indicating any occurrence of a father’s help with planning and paying for things (1 = father or both parents contributed, 0 = mother, someone else, or no one contributed). The final measures (planning and financial responsibility, respectively) consisted of mean scores ranging from 0–1 indicating whether the father was involved in planning and making financial contributions. Fathers’ influence in making decisions related to their children’s care was based on the question “How much influence do you have in major decisions such as …?” The items (1 = none, 2 = some, 3 = a lot) were collapsed into an indicator variable representing any degree of influence the father reported having (0 = none, 1 =some or a lot of influence) about the child’s life (e.g., rules, education, health care). The final influence scale was based on a mean score ranging from 0–1. Additional information about the father engagement and father responsibility outcome measures is included in Appendix A (available online).

Criminal involvement

The primary predictor variable was criminal involvement, measured as a composite of 17 individual, self-reported illegal behaviors in the past year based on a modified version of the Self-Report of Delinquency Questionnaire developed for the National Youth Survey (Elliott, Huizinga, & Ageton, 1985). Illegal behaviors were categorized according to legal seriousness (Class C misdemeanor through Class 1 felony). Using an algorithm from previous research that accounts for frequency and seriousness of offense categories, an index of criminal involvement was constructed with five categories ranging from 0–4, with higher numbers indicative of more severe, frequent levels of involvement in illegal behaviors (Giordano, Cernkovich, & Pugh, 1986; Gorman-Smith, Tolan, Zelli, & Huesmann, 1996). Category definitions included

• nonoffenders, who reported committing no minor or major offenses during the past year;

• low-frequency minor offenders, who committed one or two minor offenses in the past year but no major offenses;

• moderate-frequency minor offenders or low-frequency less serious major offenders, who had some minor offenses (three) or no more than two less serious major offenses (e.g., hit someone, used drugs);

• high-frequency less serious major offenders or low-frequency more serious major offenders, who had committed 3–9 less serious major offenses or one or two more serious major offenses (e.g., attacked someone with a weapon); and

• very high-frequency less serious major offenders or high-frequency more serious major offenders, who reported 10 or more less serious major offenses or three or more serious major offenses.

Moderating variables

The first group of moderating variables assessed the extent to which fathers perceived supportiveness and encouragement for being a father based on the question, “In general, would you describe the following people as being very supportive of your being a (father/father-figure) to [child], somewhat supportive, do not care one way or the other, or try to prevent you from having a relationship with [child]?” Fathers answered this about (a) male relatives (e.g., father, grandfather, uncle); (b) female relatives (e.g., mother, grandmother, sister); (c) the mother of the child; (d) the mother’s male relatives; and (e) the mother’s female relatives. Supportiveness was used as an indicator variable (1 = very supportive, 0 = somewhat supportive, does not care, or tries to prevent the father from having a relationship with his child).

The second group of variables assessed relationship quality with various individuals in the father’s kin and social network based on the question, “In general, would you describe your relationship with the following people as excellent, very good, good, fair, or poor?” The individuals were the same as those for the supportiveness item. The item was also used as an indicator variable (1 = excellent or very good, 0 = good, fair, or poor). The supportiveness and relationship quality variables were obtained from the DADS Initiative (additional details are available online in Appendix A; Cabrera, Moore, et al., 2004; Cabrera & Peters, 2000) and the Early Head Start Research and Evaluation Project, 1996–2010 (U.S. Department of Health and Human Services, 2005).

Covariates

We included a control for fathers’ history of illegal involvement from childhood through late adolescence. To create measures of the level and direction of change, we conducted a multilevel linear growth analysis of criminal involvement scores over the first six waves of CYDS. Random effects were included for the intercept and for age (years) slope. Age was centered at 18 so the intercept effect represented the level of criminal behaviors at age 18. The model showed significant variation in both intercept and slope. Values for the random solution for the intercept and slope were used as covariates in subsequent models to control for individual variations in level (intercept) and direction/rate of change (slope) of criminal involvement through adolescence.

Father characteristics

Household income was assessed with nine categories: 1 = less than $5,000; 2 = $5,000–$9,999; 3 = $10,000–$14,999; 4 = $15,000–$19,999; 5 = $20,000–$24,999; 6 = $25,000–$29,999; 7 = $30,000–$39,999; 8 = $40,000–$49,999; and 9 = $50,000 or more. Education and work were based on school achievement (1 = less than 7 years of school; 2 = 7–9 years of school; 3 = 10–11 years of school; 4 = high school graduate; 5 = 1 year of college or business/technical school; 6 = some college; 7 = college graduate; and 8 = professional—MA, Med, MD, PhD) and current employment (1 = yes, 0 = no), respectively. Fathers’ race/ethnicity were collapsed into 1 = non-Hispanic Black and 0 = Hispanic or other categories. Relationship status was measured with 1 = married, 2 = cohabiting, and 3 = single.

Child characteristics

Child gender (1 =male, 0 =female) and age (years and months) were included. Fathers also reported each child’s biological status (1 = biological, 0 = nonbiological) and whether they were residing with the child (1 = resident, 0 = nonresident).

Analytic Strategy

Using a hierarchical linear model approach, we examined the effects of child- and father-level predictors on father-involvement outcome variables, accounting for the nesting of children within fathers. To test for the moderating effects of kinship networks on father involvement, we include the main effects of supportiveness and relationship quality and their interactions with recent criminal behavior. Specifically, we focused on the interaction between criminal behavior and dichotomous levels of supportiveness for fathering (very supportive) and relationship quality (excellent or very good) from each of five groups of individuals from the fathers’ kinship networks: male relatives, female relatives, the mother of his children, mothers’ male relatives, and mothers’ female relatives.

An interaction term for criminal behavior and each moderator (supportiveness, relationship quality) for the five groups was entered in separate models (e.g., criminal behavior × support from fathers’ male relatives, criminal behavior × support from fathers’ female relatives, criminal behavior × relationship quality with mothers’ female relatives, etc.). These models allowed us to assess the effects of criminal behavior on eight father-involvement outcomes for fathers at low versus high levels of support and relationship quality, respectively.

For each of the eight outcomes, two models were estimated: one that included an interaction term for criminal behavior × support (high vs. low) and another with an interaction term for criminal behavior × relationship quality (high vs. low). These 16 models were estimated five times—once for each category of people in fathers’ social networks. This resulted in 80 models, the interaction terms and slope estimates of which are presented in Table 2. (The complete set of models with main effects and covariates are not included for parsimony and because of our primary interest in the interaction effects assessed.)

Table 2.

Interaction Effects Between Criminal Behavior and Supportiveness From and Relationship Quality With Fathers’ Relatives, Mothers, and Mothers’ Relatives

Dependent Variable	Supportiveness						Relationship Quality
	CrimBeh × Support (z)	High = 1		Low = 0		ESr	CrimBeh × RelQual (z)	High = 1		Low = 0		ESr
		B	SE	B	SE			B	SE	B	SE
Fathers’ male relatives
Accessibility
Can see child	2.17*	0.11	0.07†	−0.09	0.08	0.12	2.41*	0.09	0.06	−0.19	0.11†	0.17
Does see child	0.51	0.02	0.04	−0.01	0.05	0.02	2.52*	0.04	0.04	−0.15	0.07*	0.13
Engagement
Recreation	−0.34	−0.01	0.07	0.03	0.08	0.02	2.06*	0.06	0.06	−0.20	0.11†	0.15
Education	−0.56	−0.08	0.08	−0.01	0.10	0.03	2.11*	0.01	0.07	−0.30	0.13*	0.15
Caregiving	0.62	0.08	0.07	0.02	0.08	0.03	2.59*	0.12	0.06*	−0.20	0.11†	0.17
Responsibility
Planning	0.92	−0.01	0.02	−0.03	0.02†	0.05	1.92†	−0.01	0.01	−0.07	0.03*	0.14
Financial responsibility	2.48*	0.02	0.02	−0.04	0.02*	0.14	1.99*	0.00	0.02	−0.06	0.03*	0.14
Influence	0.94	0.01	0.02	−0.01	0.02	0.05	2.46*	0.02	0.01	−0.06	0.03*	0.20
Fathers’ female relatives
Accessibility
Can see child	1.13	0.06	0.06	−0.06	0.10	0.07	1.15	0.06	0.06	−0.10	0.13	0.10
Does see child	−0.38	−0.003	0.04	0.02	0.06	0.02	1.47	0.02	0.04	−0.11	0.08	0.09
Engagement
Recreation	−0.89	−0.01	0.06	0.09	0.10	0.06	0.99	0.03	0.06	−0.11	0.14	0.08
Education	−0.84	−0.08	0.07	0.03	0.12	0.05	0.99	−0.02	0.07	−0.19	0.16	0.08
Caregiving	−0.99	0.03	0.06	0.14	0.10	0.06	1.12	0.08	0.06	−0.08	0.13	0.09
Responsibility
Planning	−1.03	−0.03	0.01*	−0.003	0.02	0.06	0.08	−0.02	0.01	−0.03	0.03	0.02
Financial responsibility	0.23	−0.01	0.02	−0.01	0.03	0.00	0.46	−0.01	0.02	−0.02	0.03	0.02
Influence	−0.51	−0.001	0.01	0.01	0.02	0.03	0.75	0.01	0.01	−0.02	0.03	0.07
Mothers
Accessibility
Can see child	2.89**	0.11	0.06†	−0.17	0.09†	0.17	1.69	0.10	0.06	−0.06	0.08	0.10
Does see child	1.59	0.03	0.04	−0.07	0.06	0.07	1.18	0.04	0.04	−0.04	0.05	0.05
Engagement
Recreation	−0.20	0.003	0.06	0.02	0.09	0.01	0.04	0.01	0.07	0.01	0.09	0.00
Education	−0.53	−0.07	0.07	−0.01	0.12	0.03	1.01	−0.01	0.08	−0.13	0.10	0.06
Caregiving	−0.23	0.05	0.06	0.07	0.09	0.01	0.46	0.07	0.07	0.02	0.09	0.03
Responsibility
Planning	0.80	−0.02	0.01	−0.04	0.02†	0.05	1.15	−0.01	0.02	−0.04	0.02†	0.07
Financial responsibility	1.94†	0.01	0.02	−0.04	0.02†	0.11	0.37	−0.003	0.02	−0.01	0.02	0.02
Influence	1.49	0.01	0.01	−0.02	0.02	0.07	0.84	0.01	0.02	−0.01	0.02	0.05
Mothers’ male relatives
Accessibility
Can see child	0.68	0.06	0.08	0.002	0.07	0.03	1.20	0.09	0.07	−0.02	0.08	0.07
Does see child	1.44	0.05	0.05	−0.04	0.04	0.06	1.06	0.04	0.05	−0.03	0.05	0.05
Engagement
Recreation	−0.92	−0.04	0.08	0.04	0.07	0.05	−1.33	−0.05	0.07	0.09	0.08	0.08
Education	−0.77	−0.10	0.09	−0.02	0.08	0.04	−0.76	−0.08	0.08	0.00	0.09	0.04
Caregiving	−0.17	0.04	0.08	0.06	0.07	0.01	−1.08	0.01	0.07	0.12	0.08	0.06
Responsibility
Planning	−0.07	−0.03	0.02	−0.02	0.02	0.02	1.13	−0.01	0.02	−0.03	0.02†	0.05
Financial responsibility	1.44	0.01	0.02	−0.02	0.02	0.07	1.79†	0.02	0.02	−0.03	0.02	0.11
Influence	0.40	0.01	0.02	−0.002	0.02	0.03	2.00*	0.03	0.02	−0.02	0.02	0.12
Mothers’ female relatives
Accessibility
Can see child	1.22	0.09	0.07	−0.02	0.07	0.07	1.55	0.12	0.08	−0.03	0.07	0.09
Does see child	0.85	0.03	0.05	−0.02	0.04	0.03	2.24*	0.09	0.05†	−0.06	0.05	0.10
Engagement
Recreation	−1.61	−0.08	0.08	0.08	0.07	0.09	−0.67	−0.02	0.08	0.05	0.07	0.04
Education	−1.32	−0.13	0.09	0.01	0.08	0.07	0.52	−0.01	0.09	−0.07	0.09	0.03
Caregiving	−1.03	0.00	0.08	0.10	0.07	0.05	0.31	0.08	0.08	0.05	0.07	0.02
Responsibility
Planning	0.04	−0.02	0.02	−0.02	0.02	0.00	0.94	−0.01	0.02	−0.03	0.02†	0.05
Financial responsibility	0.41	−0.003	0.02	−0.01	0.02	0.02	1.28	0.01	0.02	−0.02	0.02	0.07
Influence	0.23	0.005	0.02	0.00	0.02	0.01	2.01	0.03	0.02†	−0.02	0.02	0.12

Notes. Interaction effects from hierarchical linear models predicting father involvement (N = 149 fathers, N = 335 children). Supportiveness and encouragement from others for being a father: 1 = very supportive (high support); 0 = somewhat supportive, tries to prevent father from having a relationship with the child, or does not care one way or the other (low support). Relationship quality with others: 1 = excellent or very good (high relationship quality); 0 = good, fair, or poor. CrimBeh = criminal behavior; RelQual = relationship quality; B = unstandardized regression coefficient; SE = standard error; ESr = the estimated effect size using a correlation coefficient metric. The effect size was calculated as the difference in parameters converted to the correlation metric by dividing by the product of the standard deviations for criminal behavior and father involvement: ES_r = (b_High − b_Low)/(s_{CriminalBehavior} * s_{FatherInvolvement}). By convention, an effect size of r = 0.1 is “small” and r = 0.3 is “medium.”

^†p < .1.

^*p < .05.

^**p < .01.

We hypothesized that criminal involvement would be significantly negatively associated with father involvement for fathers with low levels of support and relationship quality, and that this negative association would be attenuated (i.e., non-significant) for fathers with high levels of support and relationship quality. To aid interpretation of significant interaction terms, planned effect estimates show the association between criminal behavior and father involvement at both levels of support and relationship quality. All models included main effects, interaction terms, and the full set of controls for fathers’ characteristics, including past criminal involvement and covariates at the child level. For demonstration purposes (available online in Appendix B), we specify one analytic model in hierarchical linear model notation with a focus on financial responsibility as the outcome and supportiveness from fathers’ male relatives as the moderator.

Results

Descriptive Statistics

Father and child characteristics

Table 1 provides descriptive information for the independent variables. The average household income was between $20,000 and $24,000. Approximately 32% of the fathers had less than a high school degree, and 71% reported currently working. Seventy-one percent of the fathers were Black, and the remaining 29% self-identified as Hispanic or another race/ethnicity. Nearly one fourth of fathers (24%) were married, 38% were cohabiting, and the remainder (38%) were single. Fathers had an average of two children, most of whom (52%) were male. The children were 4.7 years old, on average, and the majority (84%) were related to their fathers biologically; over half (54%) lived with their father.

Table 1.

Sample Characteristics

Variables	M ± SD or N (%)
Father characteristics
Household incomea, c	5.8	±	2.7
Educationa
High school degree or greater	102.0		68.5%
Less than high school degree	47.0		31.5%
Currently workinga	105.0		70.5%
Race/ethnicitya
Non-Hispanic Black	106.0		71.1%
Hispanic or other race/ethnicity	43.0		28.9%
Relationship statusa
Married	35.0		23.5%
Cohabiting	57.0		38.3%
Single	57.0		38.3%
Number of childrena	2.1	±	1.1
Child characteristics
Male childb	173.0		51.6%
Child ageb	4.7	±	3.4
Biological child of fatherb	280.0		83.6%
Child lives with fatherb	181.0		54.0%
Supportivenessd (very supportive)
Fathers’ male relativesa	85.0		57.1%
Fathers’ female relativesa	105.0		70.5%
Mothersb	243.0		72.5%
Mothers’ male relativesb	128.0		38.2%
Mothers’ female relativesb	142.0		42.4%
Relationship qualitye (excellent or very good)
Fathers’ male relativesa	114.0		76.5%
Fathers’ female relativesa	123.0		82.6%
Mothersb	230.0		68.7%
Mothers’ male relativesb	188.0		56.1%
Mothers’ female relativesb	171.0		51.0%
Current criminal behaviora
Nonoffender	39.0		26.2%
Low-frequency minor offenders	25.0		16.8%
Moderate-frequency minor or low-frequency less serious major offenders	42.0		28.2%
High-frequency less serious major or low-frequency more serious major offenders	15.0		10.1%
Very high-frequency less serious major or high-frequency more serious major offenders	28.0		18.8%

Note. M = mean; SD = standard deviation.

^aVariables measured at the father level (N = 149).

^bVariables measured at the child level (N = 335).

^cHousehold income scale from 1–9: 1 = less than $5,000; 9 = $50,000+.

^dSupportiveness scale: 1 = very supportive; 0 = somewhat supportive, tries to prevent father from having a relationship with the child, or does not care one way or the other.

^eRelationship quality scale: 1 = excellent or very good; 0 = good, fair, or poor.

Supportiveness and relationship quality

More than half (57%) of all fathers reported high levels of support from their male relatives with respect to encouragement for being a father (Table 1). Similarly, supportive relationships were reported among their female relatives (71%), and nearly three quarters (73%) reported getting support from the mothers of their children. Despite the large proportion of highly supportive mothers, fathers reported the lowest prevalence of supportiveness from mothers’ family members—mothers’ male relatives (38%) and mothers’ female relatives (42%). Fathers’ reports of relationship quality with relatives were similar to their reports of supportiveness. They indicated having excellent or very good relationships with their own relatives at higher prevalence rates than with mothers’ relatives—77% and 83% with fathers’ male and female relatives, respectively, compared to 56% and 51% with mothers’ male and female relatives, respectively. Nearly 69% of fathers reported having high-quality relationships with the mothers of their children.

Criminal behavior

Fathers’ self-report of criminal behavior indicates significant variation in the frequency and level of seriousness of past-year illegal activity. Over one fourth (26%) of the fathers reported no involvement in illegal behavior whatsoever (score = 0), and 17% reported low-frequency minor offenses (score = 1). The midrange of the scale (score = 2) identified 28% of fathers as being involved with moderate-frequency minor offenses or low-frequency less serious major offenses. Finally, 10% of fathers were involved with high-frequency less serious major offenses or low-frequency more serious major offenses, and 19% were involved with very high-frequency less serious major offenses or high-frequency more serious major offenses.

Moderating Effects

We now turn to the results of the analyses evaluating the moderating effects of relationship quality and support on the association between criminal behavior and father involvement. Table 2 presents the test of significance for the criminal behavior × relationship quality and criminal behavior × supportiveness interactions and their effect estimates with high and low levels of relationship quality and supportiveness.

Fathers’ relatives

Among fathers’ male relatives, there were significant criminal behavior × supportiveness interactions for two measures of father involvement: can see his child (z = 2.17, p < .05) and financial responsibility (z = 2.48, p < .05). For both outcomes, higher levels of criminal behavior were associated with lower levels of involvement among fathers with low support (can see his child: B = −.09, SE = .08; financial responsibility: B = −.04, SE = .02), though this negative association was statistically significant for financial responsibility only. Among those with high support, criminal behavior was not significantly associated with father involvement for either outcome. The significant interaction suggests that having highly supportive male relatives may act as a buffer against the negative association observed between criminal behavior and father involvement in cases where support for the father is low.

Figure 1 plots the interaction and simple slopes for the effect of criminal behavior on one of the father-involvement outcomes—financial responsibility—for two levels of support with fathers’ male relatives: high and low support. The figure shows that at low levels of support between fathers and their male relatives, criminal behavior was negatively associated with fathers’ financial responsibility (i.e., as criminal behavior increased, the lower fathers’ financial contributions were to children). At high levels of support, criminal behavior was positively associated with fathers’ financial contributions, although this association was not statistically significant.

Figure 1.

Association between fathers’ criminal behavior and financial contributions at high and low levels of support from fathers’ male relatives. Associations are graphed from the first panel of model results in Table 2 (predictors of financial responsibility with interaction effects between criminal behavior and supportiveness from fathers’ male relatives).

The most consistent pattern of findings among all the models was observed with significant interactions between criminal behavior × relationship quality among fathers’ male relatives. This was the case for seven father-involvement outcomes: can see his child (z = 2.41, p < .05); does see his child (z = 2.52, p < .05); recreation (z = 2.06, p < .05); education (z = 2.11, p < .05); caregiving (z = 2.59, p < .05); financial responsibility (z = 1.99, p < .05); and influence (z = 2.46, p < .05). Like supportiveness, among fathers with low relationship quality with their male relatives, criminal behavior was negatively associated with father involvement for four of the seven outcomes (does see child, education, financial responsibility, and influence) and marginally significant for three of the outcomes (can see child, recreation, and caregiving). Stated simply, among fathers who reported high relationship quality with male relatives, there is a buffering of the negative effect of criminal behavior on father involvement observed in lower quality relationships.

Mothers

In the case of mothers’ supportiveness for and relationship quality with fathers, we found one significant interaction: criminal behavior × supportiveness for can see child (z = 2.89, p < .01). Among fathers with low support from mothers, criminal behavior was marginally negatively associated with perceived ability to see the child (B = −.17, SE = .09, p < .10).

Mothers’ relatives

With respect to mothers’ relatives, we found significant interactions in only two cases: criminal behavior × relationship quality interactions for (a) mothers’ male relatives for influence (z = 2.00, p < .05); and (b) mothers’ female relatives for does see child (z = 2.24, p < .05). However, in the presence of high-quality relationships, the association between criminal behavior and father involvement was positive but nonsignificant in one case (influence) and marginally significant in the other (does see child). In the presence of low-quality relationships, the association was negative but nonsignificant for both outcomes.

Discussion

In fatherhood research much attention has been given to the role of paternal residence (Greene & Moore, 2000; Nepomnyaschy & Garfinkel, 2011) and the relationship characteristics of parents (Carlson & Magnuson, 2011). Although important, these factors overlook the role of fathers’ social context (e.g., relations with people in one’s extended network, engagement in high-risk behavior) and how social support and relationships may influence fathers’ involvement with their children. This study examined the relation of criminal behavior and father involvement and the moderating role of extended family and mothers in this association. Although previous evidence has suggested that criminal behavior is problematic for children through several pathways (Fergusson et al., 2005; Jaffee et al., 2006), there has been limited exploration of the role of individuals from fathers’ social networks in these processes.

This study’s findings suggest that the effect of criminal behavior on father involvement depends on the level of supportiveness and relationship quality a father has with individuals in his network. In particular, criminal behavior was negatively associated with father involvement when fathers had low-quality relationships with male relatives (e.g., uncles, brothers). Conversely, we observed no association between criminal behavior and father involvement for high-quality relationships with male relatives. This suggests that having strong and affirmative relationships with male relatives may buffer the negative effects of criminal behavior on father involvement, expressed through disengagement of fathers from their children. We found this result across seven father-involvement outcomes: accessibility (access to the child, contact with the child); engagement (recreation, education, caregiving); and responsibility (financial responsibility, influence). We found more modest results for fathers who reported high levels of support and encouragement for being a father from male relatives. That is, criminal behavior was negatively associated with financial responsibility (but not other outcomes) when fathers reported low levels of fathering support from male kin.

Our findings may be explained by previous research that suggests a father’s experience in his own family of origin has implications for interactions he has with his children (Beaton & Doherty, 2007). Evidence indicates that men who had highly involved, nurturing fathers are more likely to be involved with their own children through identification and modeling, thereby replicating the parenting they received (Belsky, 1984). Men who had uninvolved fathers may also be more likely to be involved with their own children, but that involvement is through compensatory processes in which they enact parenting differently from that of their own fathers (Belsky, 1984).

In the current study, it may be the case that fathers who had close relationships with their fathers (or with other salient male family figures) growing up continued to have similar bonds in young adulthood and during stages of early parenting with their own children. If these relationships have indeed continued, they may afford opportunities to engage in prosocial behavior (positive father involvement) through modeling despite other behaviors that are high risk and antisocial (criminal activity).

What is unclear is whether fathers in our study started out in less close, nurturing relationships with their own fathers (or other male figures) that changed over time and later became higher quality and supportive, or if they began this way. Another uncertainty is who in the male extended family network (e.g., father, uncle, cousin, sibling) the fathers were thinking of when reporting high-quality and supportive relationships. Qualitative research on low-income fathers’ kin networks suggests that kin systems can be diverse and complex, with wide ranges of male and female kin and nonkin playing a role in the social support network pertaining to family, parenting, and child-rearing (Roy & Vesely, 2010). Future research could explore the variation in kin networks to better understand which members of the family system are most significant for father involvement and in what ways.

The second set of findings pertains to fathers’ connections with the mothers of their children and with mothers’ relatives. Again, the relation of criminal behavior and father involvement depended on the extent of support and quality of relationships the father had with others. Specifically, as expected, criminal behavior was associated with lower levels of father involvement (access to child) when mothers were less supportive of fathers. Contrary to expectations, criminal behavior was also associated with higher levels of father involvement when mothers were more supportive of fathers. A similar pattern was found for the association between criminal behavior and fathers actually seeing their children in the presence of high-quality relationships with mother’s female relatives. In summary, this set of results found that two potentially interconnected groups—mothers and mothers’ female relatives—played a role in moderating the relation between criminal behavior and father involvement. Specifically, these individuals had an effect on fathers’ perceived access to children and how frequently they saw their children under certain relationship conditions. Unlike less direct measures of father involvement (e.g., influence on important decisions), it may be the case that mothers and their female relatives play gatekeeping roles and influence father–child accessibility (i.e., the actual time and contact that a child has with his or her father).

The finding about maternal gatekeeping is largely congruent with other research on the fundamental role that mothers and kin play in mediating father–child relationships (Doherty et al., 1998). For instance, fathers who live with or near various kin—including their mothers, grandmothers, or siblings—often rely on these individuals to care for their children while they work, run errands, or socialize (Hamer & Marchioro, 2002). Another explanation is that mothers and related kin may play gatekeeping roles that hinder father-involvement activities (Fagan & Barnett, 2003; McBride et al., 2005). Although much of the research in this area has been conducted with samples of nonresident father households, other evidence points to the role of maternal gatekeeping even when parents are married or cohabiting (Schoppe-Sullivan, Brown, Cannon, Mangelsdorf, & Sokolowski, 2008). This body of work generally finds that maternal perceptions and attitudes about fathering, as well as mothers’ actual behavior (e.g., active encouragement for fathering), shapes father-involvement behaviors. Although research on maternal gatekeeping is substantial, there has been less investigation of extended female kin in a mother’s network; this should be the focus of future research to enhance understanding of the breadth and depth of social networks in affecting fathering behavior.

Limitations

Our study does have limitations that are important to acknowledge. Given the multiple forms of father involvement measured, and the sources of supportiveness for fathering and relationship quality with family members, these analyses included a large number (80) of interaction effects. Although we do not believe the results are entirely a function of chance, these analyses should be considered exploratory, and these findings should be verified with another sample. Nonetheless, the likelihood that the number of significant interaction effects obtained can be explained by random chance alone (i.e., obtaining 12 or more effects significant at p < .05 from 80 models) is quite small at p = .0006.

Further, because of how questions were asked, we are unable to discern who in the extended family the respondent is referring to when asked about relationships with and support from male and female relatives. For instance, if the father reports high levels of support from male relatives, we do not know if he is referring to his own father, his grandfather, an uncle, or another male relative. The same is the case for female relatives. Measures that allow for differentiation of people in the network would improve our understanding of the role that particular family members play in fathers’ involvement.

Likewise, we are limited in understanding how high levels of relationship quality and supportiveness are actualized. How, for example, is supportiveness and encouragement for fathering expressed to fathers who endorse high levels of such behavior by people in their network? Is such behavior expressed through emotional encouragement, tangible contributions to the child on his behalf, cooperative coparenting, or other forms of support? Again, having more nuanced measures could allow for an enhanced understanding of the mechanisms through which relationships and support affect father involvement and, potentially, child outcomes.

The lack of representativeness of our sample presents another limitation (participants were predominantly African American and Hispanic from a low-income, inner-city community.) Therefore, we cannot generalize these findings to other groups. Lastly, it is necessary to highlight our inability to isolate parental involvement from potential genetic influences. Gene-environment interaction studies have identified a significant genetic component in antisocial behavior (DeLisi, Beaver, Vaughn, & Wright, 2009). Our study, however, could not empirically examine the role that genetics may have in antisocial behavior, parental involvement, and the larger social context in which the participants live.

Conclusion

Our empirical findings complement those of qualitative studies—social support from partners and family members is crucial to paternal involvement, especially for African American fathers from low-income neighborhoods (Summers, Boller, & Raikes, 2004)—and our findings about the link between mothers, mothers’ extended family, and father involvement corroborate previous work in this area. However, these findings were generally modest given the number of father-involvement measures available in our study. Our most robust pattern of results suggests something different: Male relatives from a father’s own family may be centrally important to involvement with his children. This was an unexpected finding given the large proportion of fathers from our sample who were at high risk of growing up in a single-mother household without their father (Hamilton, Martin, & Ventura, 2009). This said, empirical evidence has shown significant variation in father contact—particularly among nonresident fathers—suggesting that consistently declining contact with children over time does not tell the whole story. As has been the case in other social and behavioral health studies where male family networks or male peers have been identified as important links to health-promoting behaviors (e.g., Schneider, Michaels, & Bouris, 2012), it may be the case that male kin are playing a pivotal role in young fathers’ lives in ways that researchers are just beginning to understand. This may have future importance as researchers and practitioners work to build preventive interventions that promote child health through positive father involvement and extended family support.

The results from this study may inform current social work practice and future research. The prominent findings in this study suggest that fathers with high-quality relationships with their male relatives may act protectively against disengagement from their children despite fathers’ involvement in criminal activity. Social workers could use an expansive and holistic approach when engaging fathers and their children, with a particular emphasis on helping fathers identify and call upon salient male role models in their family network. This might include helping to “bridge” members of one’s family so fathers can connect to and benefit from the positive influence of male relatives. For fathers who are not already well connected to male kin, social workers could also help identify people in the fathers’ community who could play the role of familial mentor or fictive kin to provide support as the father pursues his fathering goals. We are not suggesting that fathers’ male kin are the only important family members when trying to increase paternal involvement. Mothers are central to this process, and as this study finds, they are particularly pertinent to father–child access through what is likely maternal gatekeeping behavior. However, father involvement and parenting programs could go beyond traditional father–mother–child focused strategies and attend to other members of the extended family who could have notable influence in the prevention of disengagement between fathers and their children.

Biographies

Pajarita Charles, PhD, is an assistant professor at the School of Social Work, University of Wisconsin-Madison. This manuscript was completed while she was at the School of Social Service Administration, University of Chicago.

Deborah Gorman-Smith, PhD, is interim dean and the Emily Klein Gidwitz Professor at the School of Social Service Administration, University of Chicago.

Michael Schoeny, PhD, is an associate professor in Community, Systems, and Mental Health Nursing at Rush University College of Nursing.

Laura Sudec, MA, is a clinical research manager in the Early Intervention Research Group at Northwestern University. She was a master’s student in the School of Social Service Administration, University of Chicago, when she contributed to this manuscript.

Patrick Tolan, PhD, is the Charles S. Robb Professor of Education at the University of Virginia Curry School of Education and Department of Psychiatry and Neurobehavioral Sciences.

David Henry (deceased), PhD, was a professor of psychiatry and psychology at the University of Illinois at Chicago College of Medicine and a professor of health policy and administration in the School of Public Health.