|Home | About | Journals | Submit | Contact Us | Français|
This study focuses on relations between fathers’ behavior in family context and children’s adjustment, including the roles of paternal depressive symptoms, paternal marital conflict behaviors, paternal parenting, and children’s emotional security. Participants included 235 families with a six year old child, with families followed longitudinally each year for three years. In terms of fathers’ adjustment, paternal problem drinking was related to paternal negative marital conflict behaviors and decreased positive parenting, which was associated with children’s externalizing and internalizing problems. Fathers’ depressive symptoms were directly related with children’s internalizing problems. Children’s emotional security was an intervening variable in relations between father’s behavior in family context and children’s development.
Past research has often neglected the influences of fathers on children’s development within multiple family contexts. It remains that fathers are under-represented in studies of child development and are rarely a distinct focus of study (Phares, Field, Kamboukos, & Lopez, 2005). Consequently, understanding the role of fathers as socialization agents in the development of their children, especially in the context of longitudinal model testing, has been neglected. Although support is growing for the proposition that fathers influence children as a function of multiple family processes, fathers are still seldom a focus of systematic study involving simultaneous consideration of several possible pathways (Phares, Lopez, Fields, Kamboukos, & Duhig, 2005). The present study contributes uniquely by focusing on relations between fathers in marital and family context and their children’s development.
Based on the developmental psychopathology perspective (Cohen & Cicchetti, 2006), Cummings, Goeke-Morey and Raymond (2004) outlined a framework for examining multiple pathways of influence by fathers in the context of marital and family functioning. This framework provides bases for conceptualizing how fathers influence children’s adjustment over time, including psychological functioning, marital conflict behaviors, and parenting. These pathways include children’s coping processes, focusing on children’s emotional security, as an intervening and explanatory variable for developmental pathways. Each of the pathways emphasized in the fathering framework are potentially influential in the development of children, but have seldom been examined in a comprehensive model. Past research also indicates children’s emotional security plays an intervening role in relations between marital conflict and children’s outcomes (e.g., Cummings, Schermerhorn, Davies, Goeke-Morey, & Cummings, 2006). Specifically, children’s security about family relationships underlies children’s responses to certain stressors, including marital conflict (Davies & Cummings, 1994). However, children’s emotional security has yet to be examined specifically in relation to fathers’ behaviors in the family context. The present study explores the role of emotional security and the influence of fathers in multiple family contexts on child adjustment.
While there has yet to be one comprehensive model examining the links between fathers’ behaviors and children’s development, links between paternal adjustment such as drinking and depressive symptoms, marital conflict, and children’s security in the marital relationship, have recently been the subject of study. For example, marital conflict may intervene in the relations between paternal psychological symptoms and child adjustment (Cummings, Keller, & Davies, 2005). Keller, Cummings, & Davies (2005) found that paternal problem drinking predicted increased negative marital conflict, which, in turn, related to children’s maladjustment. Moreover, longitudinal relations are reported between paternal (but not maternal) problem drinking, the quality of the parent-child and marital relationships, and externalizing and internalizing problems in children (Keller, Cummings, Davies, & Mitchell, 2008). Parental problem drinking has also been found to impact children’s adjustment particularly when other family stressors, including parental depression and family conflict, are present (El-Sheikh & Flanagan, 2001). In terms of depressive symptoms, relations have been reported between paternal dysphoria, paternal marital conflict, child emotional security and child adjustment (DuRocher-Schudlich & Cummings, 2007). Mother-child relationship qualities and emotional security have been implicated in pathways between depressive symptoms and child adjustment (Goodman & Gotlib, 1999), but little is known about these relations for fathers (Phares, 1992). The scant research leaves questions about possible differential predictions for internalizing and externalizing problems. For example, Cummings, Keller, and Davies (2005) only reported pathways between paternal dysphoria and children’s internalizing, but not externalizing, problems. This study addresses gaps by examining the effects of paternal psychological symptoms in the context of other paternal behaviors in the family system, including assessments of children’s emotional security.
A gap also exists in the study of fathers’ parenting behaviors in the family context, including children’s emotional security. While it is known that positive father involvement can have a positive impact on children’s development and the marital relationship (Cowan, Cowan, Pruett, & Pruett, 2006), and that marital conflict can impact parent-child relationships (e.g., Frosch & Manglesdorf, 2001; Sturge-Apple, Davies, & Cummings, 2006), little is understood about how paternal marital conflict behavior affects paternal-child relationships and child functioning (Crockenberg & Forgays, 1996). The fathering vulnerability hypothesis (Cummings et al., 2004) posits fathering is especially susceptible to the stresses of family problems. Moreover, past studies have rarely distinguished the effects of fathers’ marital conflict on fathers’ positive parenting (Eisenberg et. al., 2005; Eisenberg, Cumberland, & Spinrad, 1998). Therefore, consistent with evidence of fathers’ withdrawal from discordant marriages (Osborne & Fincham, 1996), one hypothesis is that reduction in positive fathering may be an especially significant negative influence in the context of marital conflict. Fathers experiencing stress and dissatisfaction in their marital relationships may be less loving and attentive to the needs of their children, or offer less emotional support and positive stimulation (Coiro & Emery, 1998; Osborne & Fincham, 1996), which may be associated with adjustment problems in children (Eisenberg et. al., 2005). To test these possibilities, the current model examines positive parenting as a variable of interest in relation to children’s internalizing and externalizing problems.
In relation to fathers and children’s emotional security in the family, emotional security theory (EST; Davies & Cummings, 1994) posits that children appraise family situations in relation to their own security about their family’s and personal well-being with the goal of preserving their own sense of protection and security within the family (Cummings & Davies, 1996). Therefore, in the presence of marital conflict or other family dysfunction linked with fathers, emotional insecurity may be associated with increased adjustment problems in children. Children’s emotional security about family relationships has been linked with risk for internalizing, and less consistently, externalizing problems (Davies & Cummings, 2006). Although there are bases for expecting links between paternal behaviors and children’s emotional security, little is known about relations between family stressors associated with fathers’ behavior in the family context, children’s emotional insecurity, and child adjustment. Moreover, questions remain about whether emotional security associated with fathers’ behavior mediates relations with externalizing as well as internalizing problems, given less consistent relations of emotional security with externalizing problems (see also Grych, Harold, & Miles, 2003).
To address these gaps in the literature, multiple pathways of influence by fathers on children are examined, including exploration of the intervening role of children’s emotional insecurity. Notably, this study focuses on behaviors tied to fathers as sources of stress for families and children. Our hypotheses and proposed directional pathways are based on the framework set forth by Cummings, Goeke-Morey, & Raymond in 2004. For example, alcohol problems are linked with fathers rather than mothers as influences on children. Keller and colleagues (2008) found that paternal problem drinking was associated with marital conflict and negative parenting, which then impacted children’s outcomes over time. However, this study examined mother and father behaviors simultaneously and excluded other variables, such as depression and children’s emotional security. The current study will build upon and add to such research.
In addition to the appropriateness of this model specifically for fathers, examining these multiple influences at the same time meant that model testing could not simultaneously test mothers’ influences due to limitations from a statistical perspective. Our objective of testing a more comprehensive, multivariate model of paternal socialization processes requires considerable statistical power. Thus, inclusion of maternal influences in the model was not possible due to statistical constraints and model oversaturation. Therefore, each of the variables is limited to father effects, including paternal marital conflict behaviors (Crockenberg & Forgays, 1996). The focus on advancing a model for fathering in the context of family stressors addresses a gap resulting from an overwhelming emphasis on mothers.
It is hypothesized that paternal psychopathology, in the forms of paternal drinking and depressive symptoms, will be associated with destructive paternal marital conflict behaviors and less positive parenting, with consequences for children’s adjustment over time. It is also hypothesized that children’s emotional security will act as an intervening variable in these longitudinal relationships. The hypothesis is that family malfunction associated with fathers’ difficulties leaves children less secure about the family and relationships within the family and thus more vulnerable to adjustment problems over time. In particular, it is predicted based on past research (Cummings et al., 2005; Davies & Cummings, 2006) that internalizing problems are more likely to be related to fathers’ difficulties in the family. However, there are also sufficient bases on the study of these various family processes to expect possible relations with externalizing problems. Given how little is known about the distinctive impact of fathers on children’s development in a broader family context, the differential study of internalizing and externalizing outcomes holds interest. Moreover, this model reflects key themes and directions for future research identified in recent reviews (e.g., Cummings et al., 2004). Therefore, the current study specifically advances theory and research on fathering in the marital and family context.
Participants were from a dual-site longitudinal study of community families living in South Bend, IN, and Rochester, NY, and the surrounding areas. Families were recruited through postcards and flyers sent to several local neighborhoods and schools, booths at community functions, and referrals from other participants. To obtain a sample representative of the geographical areas, we made targeted efforts to actively recruit participants through school districts, community agencies, and events tailored to families of low socioeconomic status and of racial and ethnic diversity. Families were eligible to participate if they had been living together for at least three years and had one child currently enrolled in kindergarten between the ages of 5 and 6. The sample consists of 235 families (106 boys; 129 girls) during wave 1, approximately 227 families during wave 2, and 215 families in wave 3. There were no significant differences between retained and dropped families in terms of age, time living together, race, or marital status or on measures of problem drinking, marital conflict, and child functioning. Families who did not participate in later waves had lower incomes and included less educated mothers and fathers. The majority of couples were married (88.1%), living together for 11 years on average at the beginning of data collection. Consistent with the surrounding community, the ethnic distribution of participants was as follows: 79.2% White; 15.3% Black, 3.8% Hispanic, .4 % Biracial, and the remaining were of other racial backgrounds. The 2000 U.S. Census Bureau across the two counties indicated 82% White, 13% Black, and 5% Hispanic children. The median family income was between $40,000 and $54,999 a year and mothers and fathers averaged about fourteen years of education.
Each family visited the lab two times a year for three years. The vast majority of the visits were scheduled within two weeks of each other. Fathers, mothers, and children came to the lab for the first visit; mothers and children came for the second visit.
Families were monetarily compensated for their participation. The current study relied upon parental reports of behavior and adjustment. Moreover, due to the significant statistical relations between mother and father reports, many of the variables included in the study are based on combined scores of mother and father reports (see below). The study was conducted under the approval and direction of the Institutional Review Boards (IRBs) at both data collection sites.
The Conflict and Problem Solving (CPS; Kerig, 1996) scale was completed by parents to assess behaviors during marital conflict. Scores were averaged across male and female reporters to create paternal marital conflict scores on each of the three subscales used: physical aggression, verbal aggression, and stonewalling (correlations for these combined variables are r =.53; .51; & .40, respectively, and were all significant at the .01 level). Our selection of these specific subscales was guided by prior work indicating that hostile and disengaged forms of conflict are destructive from the child’s perspective (Cummings & Davies, in press). In the current sample, reliability coefficients ranged from .58 to .87 for the three scales on each of the measures.
The Alabama Parenting Questionnaire (APQ; Shelton, Frick, & Wootton, 1996) is a parent-reported measure. The “positive reinforcement” scale from the APQ consisted of 6 items. Internal consistency coefficients for variables used in the current analyses ranged from .78 to .84. Scores were averaged across both partners to create a father score for positive parenting. The correlation between mother and father reports of paternal positive parenting was .62.
The Parental Acceptance-Rejection Questionnaire (PARQ; Rohner, Saavedra, & Granum, 1991) is also a parent-reported index. The Warmth and Affection subscale consisted of 20 items. Scores were created by averaging maternal and paternal reports for fathers’ parenting behaviors, given the high correlation of .64. The reliability coefficients for the current project at times 1 and 2 ranged from .89 to .94.
The Center for Epidemiological Studies (CES-D; Radloff, 1977) measure assessed fathers’ depressive symptoms. The CES-D is a 20-item measure. Fathers rated how frequently they have experienced symptoms on a scale from 0 (less than one day) to 4 (five or more days). Scores of 16 or higher indicate potential levels of clinical depression; 13.7% of fathers had scores of 16 or greater. Alpha coefficients within the current sample were .86 for men at Time 1.
Parents’ alcohol use by fathers was assessed by the Parental Alcohol Experiences (PAE; Windle, 1996) scale. The PAE is a 15-item inventory assessing alcohol dependence and abuse within community and clinical samples and taps into the social, legal, family, and work/school problems associated with drinking (Davies & Windle, 1997). Participants rated how often they (and their partners) experienced each of the symptoms within the past year on a scale from 1 (never) to 5 (frequently/more than 10 times). Because both self and partner reports seem to reveal a more accurate depiction of problem drinking compared to self-reports alone, scores for fathers were averaged across both parents’ reports of father drinking (r = .62). In this study, reliability coefficients for mother reports of father drinking and father self-report of drinking were .87 and .86, respectively.
The Security in the Interparental Subsystems Scale (SIMS-PR; Davies, Forman, Rasi, & Stevens, 2002) measures children’s emotional security about marital relations. The SIMS is a 27-item measure completed by parents that assesses children’s reactions to conflict, including emotional reactivity, behavioral dysfunction, involvement in conflict, and avoidance, as outlined by the emotional security theory (EST; Davies & Cummings, 1994). Scores for each scale used were created by averaging parent reports. Correlations between mother and father reports for each respective scale are, r = .28; .57; .36; .22, all significant at the .01 level. For this sample, internal consistencies ranged from .70 and .83.
The Child Behavior Checklist (CBCL; Achenbach, 1991) was completed by the parents at each time point to assess children’s internalizing and externalizing problems. Parents rate how often children exhibited aggressive and delinquent behavior on the externalizing scale and how much children exhibit withdrawn, somatic, and anxious or depressed behaviors on the internalizing scale. These scales have well-established psychometric properties. These scores were then averaged across parents for each scale due to the following significant correlations: aggression, r = .53; delinquent behavior, r = .42; withdrawn, r = .41; somatic reactions, r = .38; anxious/depressed, r = .36. In this study, reliability coefficients ranged from .84 to .90.
Table 1 presents means, standard deviations and intercorrelations between the primary variables of interest included in the tested model. Variables used for autoregressive effects are not included in Table 1 for reasons of space; however, the information for these variables is available upon request. Analyses were based on structural equation modeling in AMOS 4.01 (Arbuckle & Wothke, 1999), which employs maximum likelihood estimation to handle missing data. Autoregressive effects were also included in the model of interest. According to MacCallum and Austin (2000), failure to include autoregressive effects can lead to biased estimates of the observed effects. Therefore, in longitudinal designs, it is essential to control for a variable’s effect on itself over time. Finally, bootstrapping was used to test the significance of indirect effects relevant to the current study.
Factor analyses were conducted to reduce data into latent variables of interest. Latent variables were created for the following: paternal conflict behaviors; positive father parenting; children’s emotional security; child internalizing problems; and child externalizing problems. Variables tested in the current study assessing the same latent constructs are denoted in bold in Table 1. Paternal drinking behaviors and depressive symptoms were treated as manifest variables due to the insufficient number (e.g. two) of indicators available for forming stable latent constructs.
The model for this study tested the hypotheses that paternal adjustment (e.g. drinking problems, depressive symptoms) would be associated with more destructive paternal marital conflict behaviors and less positive paternal parenting behaviors, consequently affecting adjustment over time. Children’s emotional security was examined as an intervening variable in the relations between fathers’ behaviors and children’s adjustment (Figure 1). This model also controlled for earlier symptoms by including autoregressive effects (Figure 1). That is, each endogenous construct was controlled for by including the same construct one year earlier and all variables at the same time points were allowed to correlate. Results indicate that all behaviors by fathers and children remained stable over time. Moreover, men’s drinking at T1 predicted an increase of paternal destructive marital conflict at T2 (β = .13, p <.01). This increase in conflict was significantly related to decreased positive fathering (β = −.23, p <.01), which then significantly predicted children’s externalizing problems (β = −.15, p <.01). Father’s depressive symptoms at T1 were significantly related to children’s internalizing problems at T3 (β = .23, p <.01). Standard fit statistics indicated a fair fit for the data (χ2/df = 2.32; RMSEA = .75; CFI = .986; NFI = .976).
Analyses seemed to support emotional security as an intervening variable in the processes of paternal influence on children’s development. Again, paternal problem drinking predicted negative conflict behaviors by fathers one year later (β =.13; p <.01). Paternal marital conflict behaviors at T2 were related to less positive parenting at the same time point (β =−.23, p <.01), which was also related to emotional insecurity in children at the same time point (β =−.18, p <.01). Emotional insecurity predicted internalizing and externalizing problems at T3 (β = .23, p< .01, β =.16, p< .05; respectively; see Figure 1). Using bootstrapping, tests of indirect effects between father variables, emotional insecurity, and children’s adjustment were examined. The indirect effects of marital conflict on children’s externalizing and internalizing problems through emotional security were significantly different from zero (95% CI for externalizing = .002 to .107; 95% CI for internalizing = .007 to .100). The indirect effect from decreased paternal positive parenting to children’s internalizing problems was mediated by children’s emotional security (95% CI = −.010 to −.001).
The impact of fathers’ behaviors on family factors and children’s adjustment has been neglected in past research, especially in terms of relatively complex longitudinal model testing. The results of the current study indicated that paternal behaviors can play an important role in children’s adjustment over time. Using structural equation modeling and rigorous statistical tests to examine the longitudinal links between fathers’ behaviors and children’s adjustment, multiple pathways to internalizing and externalizing symptoms emerged. In addition, children’s emotional security acted as an intervening variable between fathers’ behaviors and children’s adjustment.
With regard to child adjustment, paternal drinking and paternal depressive symptoms may start a downward spiral of detrimental behaviors by fathers and negative reactions by children, eventually leading to externalizing and internalizing problems for children. Thus, guided by theory and past cross-sectional research, indirect relations over time between fathering in the family context and children’s adjustment were expected. A subset of the hypotheses about pathways were supported. For example, paternal drinking problems were directly related to fathers’ negative marital conflict behaviors. Destructive marital conflict behavior by dads was then associated with a decrease in positive parenting behaviors by fathers at the same time point. Lack of positive parenting was related to decreased emotional security in children, while emotional insecurity was also significantly associated with the development of externalizing and internalizing problems in children. However, hypotheses about mediating processes for relations between paternal depressive symptoms and child internalizing were not supported, although direct relations with internalizing symptoms were identified.
In relation to fathers’ psychological health, these findings further support the notion of problem drinking by fathers as potentially detrimental to the functioning of the family (Eiden, Chavez & Leonard, 1999; Keller et al., 2005). Fathers’ problem drinking was negatively associated with multiple sub-systems in the family over time, including marital and parent-child relationships. Paternal depressive symptoms were directly associated with internalizing problems in children over time. These findings add further support to the significance of paternal depressive symptoms in the context of a broader literature that predominantly focuses on maternal depressive symptoms (Davies & Windle, 1997; Silverberg, Marczak, & Gondoli, 1996).
Additionally, the results suggest that even sub-clinical levels of depressive symptoms and drinking problems in fathers can affect children over time in the presence of multiple family contexts (Davies & Windle, 1997; Radloff, 1977). This offers a broader perspective regarding the implications of community levels of fathers’ adjustment problems for child development (Connell & Goodman, 2002) and further supports the theories that one area of malfunction in the family may relate to distress among other sub-systems (Cox & Paley, 2003). More research is needed in both community and clinical samples to examine why these symptoms in fathers influence children’s development, as well as what can be done to prevent the negative effects of paternal adjustment problems on marital and parent-child relationships over time. More information about the possible genetic effects of parental symptoms on children is also needed. Some children may be genetically predisposed to certain behaviors due to their biological make-up, but such interactions are impossible to study with the current data (Foley et.al., 2001; DeLucia, Belz, & Chassin, 2001).
While many studies have examined relations between marital conflict and parenting (Schoppe-Sullivan, Schermerhorn, & Cummings, 2007; Sturge-Apple, et. al., 2006; Cowan, et. al, 2006), few have focused on the implications of paternal depressive symptoms, problem drinking, and marital conflict for fathers’ parenting behaviors. Also, most research in this area has examined the effects of conflict and/or psychopathology on the increase of negative parenting without considering the effect of these factors on positive parenting behaviors (Fauber, Forehand, Thomas, & Wierson, 1990). This project was unique in investigating whether increases in fathers’ psychological symptoms and conflict behaviors were related to a lack of positive interactions, rather than an increase in negative behaviors (Eisenberg, et. al., 2005). Support was found for the notion that due to mounting problems, fathers may withdraw their warmth and affection from their children and engage in less positive parenting, contributing to adjustment problems for children over time (Coiro & Emery, 1998; Osborne & Fincham, 1996). Paternal problem drinking and paternal conflict behaviors were significantly associated with reductions in the positive quality of interactions between children and their fathers, with this lack of positive fathering significantly relating to more behavioral and emotional problems for children. The findings of the current study also have implications for intervention and prevention projects aimed at improving relationships between fathers, families, and their children.
Along with support for the effects of fathers’ behaviors on children, there was also evidence indicating that children’s emotional security about the family intervenes between fathers’ detrimental behaviors and children’s outcomes. Children’s emotional security intervened in the relationships between fathers’ behaviors and their own development over time, even when controlling for earlier symptoms in both children and fathers. In particular, relations between marital conflict, parenting, and children’s internalizing and externalizing problems were mediated by children’s states of security. It seems that how secure a child feels about his or her family can change the impact that the father has on this child. These results are consistent with the literature examining children’s states of emotional security about the family and children’s reactions to tense family situations as intervening variables between family processes and children’s adjustment (Davies, Harold, Goeke-Morey, & Cummings, 2002).
In weighing the contributions of this study, the limitations also merit consideration. All measures were questionnaires, and the constructs of children’s emotional security and outcomes were created by parental reports of children’s behaviors. Therefore, although use of multiple reporters for child outcomes reduces potential bias, reporter bias and common reporter bias may be present. Also, children’s own reports of their levels of emotional security and adjustment problems may be better indicators of what is taking place in their developmental trajectories. Addressing these limitations are directions for future research.
In addition, although focusing on fathers’ behaviors was essential for the present theoretically-guided longitudinal tests, we recognize that it can also be seen as a limitation. Mothers are a key aspect of all family contexts; it is therefore a limitation for interpretation of the pathways identified that they were not included in analyses. Unmeasured maternal influences and their relationships with paternal variables may have contributed to the results. Mothers’ behaviors in the family context may further explain or moderate the relationships identified between fathers functioning in the family and their children’s development. Testing mothers’ and fathers’ behaviors in the same model is therefore a direction for future research. The limited assessment of gender differences in children is also a limitation and another possible future direction for research.
Despite these limitations, the current project offers valuable knowledge about the roles that fathers play in their children’s development. Such information can aid clinicians, psychologists, and policy makers in developing programs for fathers and children faced with problematic issues. The emphasis on fathers’ positive parenting, the significance of multiple fathers’ behaviors in the context of family for children’s emotional security, and the demonstration of multiple pathways for the influence of fathers’ behaviors on child development each provide possible avenues for advances in conceptualizing clinical treatment approaches for fathers, children and families. Finally, although this report focused on fathers’ adjustment problems and child development, this information may benefit both mothers and fathers by providing them with new knowledge and insights for encouraging a healthy and well-functioning family system.
This research was supported by the National Institute of Mental Health (Project R01 MH57318) awarded to Patrick T. Davies and E. Mark Cummings. We are grateful to the children, parents, teachers, and school administrators who participated in this project. Our appreciation is expressed to project staff, including Marcie Goeke-Morey, Amy Keller, Michelle Sutton, Alice Schermerhorn, Courtney Forbes, Courtney Henry, and the graduate and undergraduate students at the University of Notre Dame and the University of Rochester.
Publisher's Disclaimer: The following manuscript is the final accepted manuscript. It has not been subjected to the final copyediting, fact-checking, and proofreading required for formal publication. It is not the definitive, publisher-authenticated version. The American Psychological Association and its Council of Editors disclaim any responsibility or liabilities for errors or omissions of this manuscript version, any version derived from this manuscript by NIH, or other third parties. The published version is available at www.apa.org/journals/fam.
Patricia M. Schacht, University of Notre Dame.
E. Mark Cummings, University of Notre Dame.
Patrick T. Davies, University of Rochester.