Fathering and Adolescent Adjustment: Variations by Family Structure and Ethnic Background

Abstract

The current study investigated how fathering behaviors (acceptance, rejection, monitoring, consistent discipline, and involvement) are related to preadolescent adjustment in Mexican American and European American stepfamilies and intact families. Cross-sectional data from 393 7^th graders, their schoolteachers, and parents were used to examine links between different dimensions of fathering and adolescent outcomes. Following an ecological multivariate model, family SES, marital satisfaction, and mothers’ parenting were included as controls. In all contexts, fathering had significant effects on adolescent adjustment. Both mothers’ parenting and adolescent gender moderated the associations, and we uncovered some provocative nonlinear relations between fathering and adolescent outcomes. The importance of ethnicity and family structure in studies of fathering are highlighted.

The potential influence of fathers on child development has long been of interest to researchers, but we still know relatively little about this potential influence in different family types from various race/ethnic backgrounds. According to the U.S. Census Bureau, 5.3 million children in the United States live in a stepfamily, with about one third of all children expected to live with a stepparent (usually a stepfather) before the age of 18 (Amato & Sobolewski, 2004; U.S. Census Bureau, 2004). Although demographic trends indicate a rise in the number of stepfamilies (U.S. Census Bureau, 2004), much of the previous work linking fathering to child outcomes has focused on “intact” families composed of a coresident birthfather and birthmother. Moreover, there is scant research on stepfathering among ethnic minority families, notwithstanding the growing number of immigrant families in the United States (Hernandez, Denton, & Macartney, 2008) and, in turn, an increase in the number of Mexican American stepfamilies (Stewart, 2007). The current study addresses this gap in the literature by examining associations between fathering behaviors and adolescent outcomes in both intact and stepfamilies of European American (EA) and Mexican American (MA) descent.

Theoretical Orientation

The current study is guided by Doherty, Kouneski, and Erikson’s (1998) conceptual model outlining the potential influences of responsible fathering. Unlike some previous models, this framework includes fathering inside or outside marriage and regardless of co-residence with the child. This ecological model highlights individual aspects of the father, mother, and child, as well as mother-father relationship factors and variables from the larger contextual environment. In our proposed model (see Figure 1), we examine components of each of these factors. Specifically, for father and mother factors we examine parenting skills (i.e., acceptance, rejection, discipline, and monitoring), as well as father-adolescent involvement and mother-adolescent involvement. For child/adolescent characteristics we include adolescent gender, and for co-parental relationships we examine marital quality and marital status (intact versus step families). For contextual factors we examine socioeconomic status and ethnicity. An important feature of this theoretical framework is its recognition that these factors interact in a systemic fashion to jointly influence children’s outcomes. This focus on the multi-faceted nature of father’s influence on children’s developmental adaptation is a central and guiding theme of our inquiry. Below we examine components of the model in more detail in stepfather versus intact families.

Figure 1.

Theoretical Model

Parenting Skills in Intact / Stepfamilies and Adolescent Adjustment

Two aspects of parenting are examined in this study, including qualitative variations in parenting skills and the quantitative level of father involvement. Qualitative variations in parenting patterns have been found to reflect two primary dimensions of behavior, namely emotionality and control (Baumrind, 1991; Parke & Buriel, 2006). Emotionality refers to how warm and responsive the parent is toward his/her child and control refers to how restricting parents are of their children’s behavior. The current study examines acceptance and rejection as the emotionality dimension, and discipline and monitoring as the control parenting dimension. Research has found parenting characterized by high warmth and low rejection to be related to positive emotional, social, and cognitive development in children (Baumrind, 1991). Furthermore, mothers’ and fathers’ inconsistent discipline has been linked to externalizing behaviors in children (Dodge, Coie, & Lynam, 2006) and poor monitoring has been related to lower academic skills and peer acceptance, and higher rates of delinquency and externalizing behavior (Dodge et al., 2006).

Parents –fathers as well as mothers – vary not only in the quality of their parenting but in their degree of involvement with their children. Father involvement is positively associated with children’s peer relationships (Burns & Dunlop, 1998), and psychosocial adjustment (Flouri & Buchanan, 2002), as well as social, emotional, and cognitive development (Cabrera et al., 2000; Pleck, 2010). In a recent systematic review of 24 longitudinal studies involving 22,300 children, Sarkadi, Kristiansson, Oberklaid, and Bremberg (2008) found that nearly all studies reported a positive effect of paternal influence on children’s psychosocial adjustment.

Less research exists on the degree to which associations typically reported between fathering practices and child outcomes in intact families are evident in stepfamilies. Overall, research on parenting practices and child outcomes in stepfather versus intact families has found more negative adjustment among children in stepfamilies (Amato & Sobolewski, 2004). In a sample of 10–18 year old children living in step and intact families, children from stepfamilies reported higher levels of internalizing and externalizing behaviors than children from intact families (Hetherington et al., 1999). Additionally, children growing up in divorced/stepfamilies displayed higher levels of internalizing and externalizing behaviors, poorer academic achievement, more behavioral and emotional problems, and lower social competence than their counterparts who were raised in intact families (Amato & Sobolewski, 2004; Hetherington, 2006). Nevertheless the more a stepfather is involved with his stepchildren, the fewer behavioral problems the child has and the better the child does in school (White & Gilbreth, 2001).

Researchers have examined differences in parenting processes to help explain differences in child adjustment between stepfamilies and intact families. Studies have found that stepfathers, on average, are less involved and communicative with their stepchildren, provide less warmth and nurturance, and hold a less positive view of their relationships with their stepchildren than birthfathers who live with their children (Hofferth, Pleck, Stueve, Bianchi & Sayer, 2002). Results regarding parental control have been mixed. Some studies have found that stepfathers exert less control than birthfathers (Hetherington, 2006) while other studies have shown no differences in parental control between stepfathers and resident birth fathers (Thomson, McLanahan, & Curtin, 1992). Parental monitoring has been found to be lower in stepfather families than in two-parent biological families (Fisher, Leve, O’Leary, & Leve, 2003).

Contextual Factors: Fathering in Mexican American families and Adolescent Adjustment

Although researchers have recognized that there may be cultural differences in parenting behaviors, little research has been devoted to understanding the similarities and differences in associations between fathering practices and child outcomes in European American and Latino families (Cabrera et al., 2000). Compared to European American and African American fathers, Latino fathers spend more time and interact more with their children (Toth & Xu, 1999) and are more likely to monitor their children (Hofferth, 2003; Toth & Xu, 1999). However, Latino fathers show similar levels of affection and warmth towards their children as fathers of other ethnicities (Hofferth, 2003; Toth & Xu, 1999). As with Caucasian non-Hispanics, acceptance and warmth were associated with positive outcomes such as decreased conduct problems and depressive symptoms for Mexican American children (Gonzales, Pitts, Hill, & Roosa, 2000). Mexican American fathers have also been found to be involved in discipline, rule setting, and play activities at levels equal to European American fathers (Backstrom, 2004).

Despite high levels of involvement on the part of Mexican American fathers, their style of parenting tends to be harsher than European American fathers (Parke et al., 2004). However, in Mexican culture authoritarian parenting is considered normative and a valued socialization mechanism (Parke & Buriel, 2006). Thus, authoritarian practices have more variable and often neutral effects on Mexican American children (Hill, Bush, & Roosa, 2003; Ispa et al., 2004; Lindahl & Malik, 1999); in contrast among European American families, authoritarian parenting is linked to poor child adjustment (Baumrind, 1991). In a study comparing Mexican, Mexican Immigrant, and Mexican American to Caucasian non-Hispanic fathers, Mexican Immigrant fathers and Mexican American fathers reported being more authoritarian than Mexican fathers and Caucasian Non-Hispanic fathers (Varela et al., 2004). However, there were no differences between groups in the use of authoritative parenting. In another study, Mexican American parents were more controlling than the Mexican parents and exhibited less warmth and acceptance than European American parents (Luis, Varela, & Moore, 2008). Nevertheless, Mexican American adolescents are relatively well adjusted even when reporting higher levels of parental harshness (Hill et al., 2003).

Very little research has examined the associations between fathering in Mexican American stepfamilies and adolescent adjustment. A study examining parenting and adolescent self-esteem in Latino intact, stepfather, and single-mother families found that the link between parental psychological control and self-esteem was strongest in stepfather families, with higher levels of psychological control being related to lower adolescent self-esteem (Plunkett et al., 2007). The current study extends prior research by examining the link between fathers’ parenting and adolescent adjustment in step and intact Mexican American and European American families. In light of the findings discussed above, we predict that father acceptance, monitoring, consistent discipline, and involvement with their adolescent will be positively associated with positive adolescent adjustment, while father rejection will be negatively associated with positive adolescent adjustment but the links between rejection and adjustment will be stronger for European American than Mexican American adolescents.

Co-parental Relations: Mother parenting as a moderator of the associations between father parenting and adolescent outcomes

Comparatively little research has examined the possible moderating effect of mothering on the associations between fathering behaviors and child outcomes. Family systems approaches to parenting have found evidence that the association between one parent-adolescent relationship and adolescent outcomes may depend on the other parent-adolescent relationship (Sim, 2003). The parenting behavior of the mother could moderate the association between fathering and child adjustment for several reasons. First, parents who co-parent successfully may reinforce each other by parenting similarly, in effect, drawing upon similar parenting scripts. This may help create an especially orderly environment for the child, and may uniquely contribute to child functioning. Alternatively, parents may intentionally divide up or split parenting tasks so that a dimension like monitoring cannot be accurately represented without consideration of the spouse’s monitoring of the adolescent. Therefore, we test whether mothering moderates the association between fathering and child adjustment and hypothesized that adolescents will have better adjustment when both parents are consistent in discipline, acceptance, monitoring, and interaction and expect poorer adolescent adjustment when both parents show rejection.

Adolescent Characteristics: Adolescent Gender and Fathering

Prior research suggests that mothers and fathers of adolescents (1) differ in their level of parenting engagement, (2) parent their sons and daughters differently, and (3) influence sons and daughters in different ways. In adolescence, mothers engage in more shared activities with their daughters than with their sons, while fathers tend to be more engaged with their sons, have less contact with their daughters, and overall have more distant relationships with their children than mothers (Hosley & Montemayor, 1997). An adolescent’s closeness with or perceived acceptance from the same-sex parent has been found to correlate with higher self-esteem for daughters, and to a lesser degree sons (Burnett & Demnar, 1996). It has been argued that adolescence is a period of gender intensification when time with, and attention to, same-sex parents increases (McHale et al., 2004). Thus, we hypothesize that adolescent gender will moderate the association between fathering and adolescent outcomes, with fathering showing stronger associations with adolescent adjustment for sons than daughters.

Controls

Various contextual factors may impact the parent adolescent relationship and we control for socioeconomic status (SES) since fathering behaviors in both intact and stepfamilies have been linked with SES in past work. While lower SES fathers are more restrictive and punitive with their children (Parke & Buriel, 2006), and may show less involvement than higher SES fathers (Yeung, Sandberg, Davis-Kean, & Hofferth, 2001), the links between involvement and SES are weak and inconclusive (Pleck, 2010). Other evidence suggests that economic stress, which is often higher among poorer families, is positively related to harsh parenting (Parke et al., 2004). Similarly, Behnke and colleagues (2008) found parental stress to be associated with negative parenting behaviors for mothers. In addition to socioeconomic status, the current study also controls for marital quality. The marital relationship has been found to “spill over” into parent-child interactions. Low marital satisfaction has been linked with higher levels of distress in children (Fishman & Meyers, 2000). Thus, the current study controls for both socioeconomic status and marital quality.

Overview of the Current Study

The current study examines the links between reports of fathering and preadolescent adjustment across these four family types (step versus intact and Mexican American versus European American). The current study is guided by three hypotheses: (1) father acceptance, monitoring, consistent discipline, and involvement with their adolescent will be positively associated with positive adolescent adjustment, while father rejection will be negatively associated with positive adolescent adjustment but the links between rejection and adjustment will be stronger for EA than MA adolescents; (2) mothering will moderate the association between fathering and adolescent adjustment, with better adolescent adjustment expected when both parents are consistent in discipline, acceptance, monitoring, and interaction and poorer adolescent adjustment expected when both parents show rejection; and (3) adolescent gender will moderate the association between fathering and adolescent adjustment, with links between fathering behaviors and sons expected to be stronger than links between fathering behaviors and daughters. These last two hypotheses are assumed to be equally applicable to families regardless of family structure or ethnicity.

Method

Participants

Families were recruited from six school districts in two southwest U.S. metropolitan areas (Riverside/San Bernardino, CA and Phoenix/Tempe, AZ). Individual interviews with mothers, fathers, and adolescents were conducted when the target adolescent was in the 7th grade. The average age of the adolescent was 12.9 years (SD = .48). The families were of either Mexican American (MA) (n = 194) or European American (EA) descent (n = 199), with all three family members of the same self-identified ethnicity. The sample consisted of two-parent families, either “intact” (i.e., two birth-parents, n = 218; MA = 108, EA = 110) or “stepfather” (i.e., a birthmother and a stepfather, n = 175; MA = 86, EA = 89). Stepfather families were defined as those in which the target adolescent’s birthmother had been living with a man who was not the adolescent’s birthfather for at least the past year, and in which the target adolescent lived with the mother more than half time.

Adolescents, mothers and fathers were interviewed individually in their language of preference (57 percent of MA parent interviews and 12 percent of MA adolescent interviews were conducted in Spanish). Interviews lasted between 1 and 3 hours and used both self-administered and interviewer-led questions.

Sample Demographics

When initially interviewed, the mean age of MA mothers and fathers was 37 and 38 years and for EA mothers was 41 and EA fathers was 43. On average, EA mothers completed 14.12 years (SD = 2.27) and fathers completed 14.02 years (SD = 2.35) of school, while U.S. educated MA mothers completed 12.41 years of school (SD = 2.26) and fathers completed 11.62 years of school (SD = 2.30). Of those parents educated in Mexico, mothers completed 8.66 years of school (SD = 3.89) and fathers completed 8.97 years of school (SD = 4.30). EA families earned, on average, $87,000 per year, while MA families earned $48,000 per year. This income disparity between EA and MA families is also found in the population at large (U.S. Census Bureau, 2005). In terms of language, our MA sample tended to over-represent Spanish speakers.

Measures

Socioeconomic Status (SES)

An index of socioeconomic status was created by combining information from per-capita family income, father’s education, and father’s occupational status, with higher scores representing higher levels of SES.

Gender

Adolescent gender was represented by a dummy variable with males coded 0 and females coded 1.

Marital quality

Marital quality was assessed using 6 items focused on happiness or satisfaction from a longer inventory (Johnson, White, Edwards, & Booth, 1986), completed by both mother and (step) father (r = .42), and averaging them across the two reporters (α >.90).

Parent acceptance, rejection, and consistent discipline

An adapted version of the Child’s Report on Parental Behavior Inventory (CRPBI; Teleki, Powell, & Dodder, 1982) was administered to the adolescent and both parents. Adolescents reported on both parents, and spouses reported on each other. Acceptance (10 items), rejection (9 items), and consistent discipline (9 items) were each assessed on 3-point Likert scales. Adolescent and mother reports of fathering were correlated (r = .27 for acceptance, .27 for rejection, and .20 for consistent discipline) as were adolescent and father reports of mothering (r = .16 for acceptance, .18 for rejection, and .15 for consistent discipline), and scores were averaged across reporters to create a summary scale of each parent’s acceptance, rejection, and discipline (α > .80 for all scales).

Parental Monitoring

A six item scale adapted from Stattin and Kerr (2000) was used which gauges the adolescent’s perception of the qualitative aspects of their parents’ monitoring, as well as parental self-reports of their monitoring on 5-point Likert scales. The adolescent reported on both parents, and parents reported on themselves. Adolescent and father reports of father monitoring were correlated (r = .30) as were adolescent and mother reports of mother monitoring (r = .35), and scores were averaged across reporters to create a summary scale of each parent’s monitoring (α = .73 for father monitoring, .78 for mother monitoring).

Parent involvement

Parents and adolescents completed five questions assessing the frequency with which each parent engages in specific activities with the adolescent (Coltrane, Parke, & Adams, 2004). The adolescent reported on both parents, and parents reported on each other using a 5 point Likert scale. Correlation across reporters was moderate (r = .29 for father and adolescent reports, .26 for mother and adolescent reports), and reports were combined across reporters into a summary scale representing father-adolescent involvement, and a summary scale representing mother-adolescent involvement (α = .68 for father-adolescent involvement, .71 for mother-adolescent involvement).

Adolescent internalizing and externalizing behaviors

The complete Behavior Problem Index (BPI: NLSY, 1979) was administered to two of the target adolescent’s teachers. In the release forms signed by the adolescent’s parents and initialed by the adolescent, they were asked to provide the names of two math, social science, or language arts teachers who then completed the BPI. The reports from the two teachers were correlated for internalizing (r = .45) and externalizing (r = .60) and were averaged into a single scale for internalizing, and a single scale for externalizing. Both scales had acceptable reliability (α > .75).

Positive behaviors

Mothers and fathers both reported on their adolescent’s positive behaviors using a 10 item scale based on the work of Quint, Bos, and Polit (1997). Mother and father reports were combined into a single scale (reliability α > .80). To facilitate the imposition of equality constraints across paths, this scale was reversed in the analyses.

Adolescent sociometrics

Two teachers separately completed three items assessing the degree to which the adolescent got along with and was accepted by other adolescents. Items included “Would majority of kids in child's class say they really don't like this child” and “Would majority of kids in child's class say this child has a lot of problems?” Both reports were combined into a summary scale (reliability α = .77). To facilitate the imposition of equality constraints across paths, this scale was reversed in the analyses.

Adolescent depression

An adapted version of the Child Depression Inventory (CDI: Kovacs, 1992) was administered. The CDI is a standardized measure and has been widely used with large nationally representative, as well as cross-cultural samples, to gauge an adolescent’s report of his/her own feelings of sadness and depression. Prior work with both English and Spanish versions has yielded strong reliability and validity (Beck, Steer, & Garbin, 1988). Our adapted version contained eight items and had an alpha of .67 for adolescent self-report, with high scores representing more depressive symptoms.

Adolescent anxiety

The complete Revised Children’s Manifest Anxiety Scale (Reynolds & Paget, 1981; 1983) was completed by the adolescent. This scale includes seven items such as “In the past month you worried about what was going to happen.” The response choices were in a dichotomous “yes” or “no” format, with high scores representing more anxiety (alpha = .66).

Analysis Plan

We used Mplus Version 4 (Muthén & Muthén, 1998–2004) to estimate the model using full information maximum likelihood estimation. Missingness was less than 5% on all variables. To examine differences by family type, these analyses were conducted in a multiple-group framework, comparing four groups (i.e., European American intact families, European American stepfamilies, Mexican American intact families, Mexican American stepfamilies). Adolescent outcomes were regressed onto predictors in a path analysis framework, allowing simultaneous estimation of all paths, and making possible the equating of parallel paths from predictors to adolescent outcomes, offering a more parsimonious final model. When evaluating the fit of structural models to the data, we used the standard chi-square index of statistical fit that is routinely provided under maximum likelihood estimation of parameters. We also used two indexes of practical fit, the root mean square error of approximation (RMSEA; Browne & Cudeck, 1993) and the standardized root mean square residual (SRMR; Bentler, 1995). We chose these two indices because the RMSEA is a commonly used index based on the non-centrality parameter, while the SRMR provides an estimate of observed to predicted covariance and is less redundant with the RMSEA than other commonly-used fit indices like the comparative fit index (CFI) or Tucker-Lewis index (TLI) which are also based on the non-centrality parameter.

To test of our first hypothesis, we examined the associations between fathering dimensions and adolescent adjustment outcomes. Three models were run for each fathering dimension: one regressing the two adolescent self-reports (anxiety and depression) onto the fathering dimension, one regressing the parent report of adolescent positive behavior onto the same fathering dimension, and one regressing the three teacher reports of adolescent outcomes (internalizing, externalizing, sociometric status) onto the fathering dimension. We analyze data from all three reporters separately because they provide slightly different results, and results from exploratory factor analyses (not shown) suggested it would be improper to force information from the three reporters onto a common factor.

To test the second hypothesis of moderation by mothering on the association between fathering and adolescent outcomes, the product term ‘mothering × fathering’ was included in the models. Significant moderation was graphed using the unstandardized regression coefficients (Whisman & McClelland, 2005); these graphs are available upon request from the first author.

To test the third hypothesis of moderation by adolescent gender on the association between fathering and adolescent outcomes, the statistical interaction between adolescent gender and father parenting was included in the models. Because failure to specify an existing nonlinear term can lead to spuriously significant interaction terms (Cohen, Cohen, West, & Aiken, 2003), the squared term of fathering was also included. While this term indexed the degree to which there was a nonlinear association between fathering and adolescent outcomes (see Deater-Deckard & Dodge, 1997), we included it primarily as a control to increase our confidence in any significant higher-order interactions with mothering or adolescent gender. Mother parenting, marital quality, gender of adolescent, and family SES were also included as control variables in these analyses.

Using measurement parcels in model estimation. Our hypotheses related to the structural model, and prior work suggests that use of multi-item parcels as indicators for latent variables is defensible in such situations (Bandalos & Finney, 2001; Marsh & O’Neill, 1984). Use of parcels in these circumstances addresses rater effects and reduces the number of estimated paths in the model. A domain-representative approach to parcel construction treats information from each reporter as equally valid (or equally biased) and unit-weights the raters by distributing their information across the parcels (for an empirical example see Schofield et al, 2009). Following the procedures outlined by Kishton and Widaman (1994), domain representative parcels were created, which allowed rater-specific variance and variance common across raters to contribute to the latent factor. As a further test of this procedure, we first ran the models using parcels as just described. We then ran the models separately for each informant. The results from these preliminary analyses supported the results from the parcel model we present here.

Results

In the interests of space the descriptive findings are briefly summarized with detailed tables and analyses available from the first author. Adolescents in stepfamilies received less monitoring from both parents, less discipline and more rejection from mothers, and less acceptance from fathers. Adolescents were also less likely to be less involved with stepfathers.

Adolescents in stepfamilies were less well adjusted on various outcomes (adolescent positive behavior, teacher-rated internalizing, externalizing and sociometric status, and adolescent-rated anxiety and depression). After including four controls, (SES, adolescent gender, time living with stepfather, and relationship with biological father) the significant differences in adolescent outcomes between intact- and stepfamilies remained. There were no significant ethnic differences with regard to either adolescent outcomes or parenting after including controls.

To assess the hypotheses relating fathering and adolescent adjustment we started with an unconstrained model for each reporter of adolescent outcomes (i.e., adolescent-report, parent-report, teacher-report), and then imposed constraints on prediction paths across outcomes and groups. This enabled our final models for each fathering dimension to be relatively parsimonious, despite the presence of six adolescent outcomes across four groups. Tables 1–5 provide coefficients and fit indices from the final models for each fathering dimension. Control variables are listed first, followed by the fathering variable, with the interactions with fathering below. Indices of model fit are presented at the bottom of the table. The following results involve the associations between fathering and adolescent outcomes, having partialled out the effects of SES, marital quality, adolescent gender, and mothering.

Table 1.

Standardized Estimates and Standard Errors Across Groups for Father Acceptance

Predictor	Family Type	Adolescent report of Anxiety		Adolescent report of Depression		Parent report of Positive Behavior		Teacher report of Internalizing		Teacher report of Externalizing		Teacher report of Sociometrics
		EA	MO	EA	MO	EA	MO	EA	MO	EA	MO	EA	MO
SES
	I	-	-	-	-	-	-	-	-	-	-	-	-
	S	-	-	-	-	-	-	−.16(.06)	-	−.18(.07)	-	-	-
Adolescent gender (CG)
	I	.10(.04)	.10(.04)	-	-	−.11(.04)	−.12(.04)	−.15(.05)	−.16(.06)	−.18(.06)	−.25(.06)	−.22(.06)	−.23(.06)
	S	.09(.04)	.11(.04)	-	-	−.11(.04)	−.10(.04)	−.11(.04)	−.26(.08)	−.22(.05)	−.39(.08)	−.16(.04)	−.38(.08)
Marital quality
	I	-	-	-	-	-	−.09(.05)	-	-	-	-	-	-
	S	-	-	-	-	−.08(.04)	−.10(.04)	-	-	-	-	-	-
Mother acceptance (MA)
	I	−.09(.02)	−.10(.02)	−.11(.02)	−.11(.02)	−.10(.04)	−.11(.04)	−.09(.04)	−.10(.05)	−.10(.05)	−.09(.04)	−.09(.04)	-
	S	−.08(.02)	−.10(.02)	−.07(.01)	−.10(.02)	−.08(.03)	−.09(.04)	−.06(.03)	−.08(.04)	−.06(.03)	−.17(.05)	−.06(.03)	−.07(.03)
Father acceptance (FA)
	I	−.12(.04)	−.13(.02)	−.14(.03)	−.14(.03)	−.33(.07)	−.22(.05)	−.22(.05)	−.25(.06)	−.26(.06)	−.22(.05)	−.23(.05)	−.26(.06)
	S	.14(.05)	-	-	-	−.25(.05)	−.23(.05)	−.22(.05)	−.26(.06)	−.24(.06)	−.24(.06)	−.22(.05)	−.24(.06)
MA × FA
	I	-	−.09(.04)	-	-	-	-	-	-	-	-	-	-
	S	−.09(.04)	−.15(.06)	−.08(.04)	−.14(.06)	-	-	-	-	-	-	-	-
CG × FA
	I	-	−.12(.04)	-	−.13(.04)	-	-	-	-	-	-	-	-
	S	−.16(.06)	−.17(.06)	−.14(.05)	−.16(.06)	-	-	-	-	-	-	-	-
FA2
	I	-	-	-	-	-	-	-	−.12(.04)	-	−.11(.04)	-	−.13(.04)
	S	-	-	-	-	-	-	−.18(.06)	−.21(.08)	−.19(.07)	−.20(.07)	−.18(.06)	−.20(.07)

Note. EA = European American, MA = Mexican American, I = intact families, S = stepfamilies. Positive behavior and sociometrics reverse-coded. Adolescent model: χ²= 66.77, df = 91, RMSEA = .000, SRMR = .037; parent model: χ²= 91.30, df = 78, RMSEA = .013, SRMR = .050; teacher model: χ²= 76.66, df = 86, RMSEA = .000, SRMR = .040, p < .05 for all paths

Table 5.

Standardized Estimates and Standard Errors Across Groups for Father Involvement With Child

Predictors	Family Type	Adolescent report of Anxiety		Adolescent report of Depression		Parent report of Positive Behavior		Teacher report of Internalizing		Teacher report of Externalizing		Teacher report of Sociometrics
		EA	MA	EA	MA	EA	MA	EA	MA	EA	MA	EA	MA
SES
	I	−.19(.06)	-	−.23(.07)	-	-	-	-	-	-	-	-	-
	S	−.15(.05)	-	−.14(.04)	-	-	-	−.14(.06)	-	−.16(.07)	-	-	-
Adolescent gender
	I	.09(.04)	.09(.04)	-	-	−.14(.04)	−.15(.04)	−.20(.05)	−.21(.05)	−.30(.05)	−.27(.05)	−.33(.05)	−.31(.05)
	S	.08(.04)	.10(.05)	-	-	−.13(.04)	−.12(.04)	−.16(.04)	−.17(.05)	−.24(.04)	−.24(.04)	−.22(.04)	−.23(.04)
Marital quality
	I	−.12(.06)	−.11(.05)	−.15(.06)	−.11(.05)	-	−.14(.04)	-	-	-	-	-	-
	S	-	-	-	-	−.12(.04)	−.11(.03)	-	-	-	-	-	-
Mother involvement (MI)
	I	−.10(.04)	−.10(.04)	−.13(.05)	−.11(.05)	-	-	-	-	-	-	-	-
	S	−.09(.04)	−.13(.05)	−.08(.03)	−.12(.05)	-	-	−.12(.04)	-	−.13(.04)	-	-	-
Father involvement (FI)
	I	−.50(.18)	−.55(.20)	-	-	−.48(.17)	-	−.21(.06)	−.26(.08)	−.24(08)	−.23(.07)	−.22(.07)	−.26(.08)
	S	-	-	-	-	−.42(.14)	-	-	−.22(.07)	−.17(.05)	−.22(.07)	−.16(.05)	−.20(.06)
MI × FI
	I	-	-	.23(.09)	-	-	-	-	-	-	-	-	-
	S	-	-	-	-	-	-	-	-	-	-	-	-
CG × FI
	I	.50(.19)	.53(.20)	-	-	.43(.20)	−.28(.06)	.19(.08)	.22(.09)	.22(.09)	.20(.08)	.21(.08)	.23(.09)
	S	−.21(.09)	-	−.18(.08)	-	.19(.08)	−.51(.07)	-	.19(.08)	-	.19(.08)	-	.18(.07)
FI2
	I	−.16(.06)	-	−.19(.07)	-	-	-	-	-	-	-	-	-
	S	-	−.19(.07)	-	−.18(.06)	-	-	-	-	-	-	-	-

Note. EA = European American, MA = Mexican American, I = intact families, S = stepfamilies. Positive behavior and sociometrics reverse-coded. Adolescent model: χ²= 74.80,df = 85, RMSEA = .000, SRMR = .041; parent model: χ²= 85.37, df = 89, RMSEA = .000, SRMR = .038; teacher model: χ²= 75.66, df = 90, RMSEA = .000, SRMR = .040, p < .05 for all paths

Hypothesis 1 was that fathering would be associated with adolescent adjustment. Consistent with this hypothesis, father acceptance (Table 1) was negatively associated with parent and teacher reports of adolescent maladjustment (coefficients range from −.22 to −.33). In intact families, father acceptance was associated with lower levels of adolescent-reported anxiety and depression. Father rejection (Table 2) was positively associated with parent and teacher reports of adolescent maladjustment (coefficients range from .10 to .33). In intact EA families, father rejection was associated with higher levels of adolescent-reported anxiety and depression. Father discipline (Table 3) was negatively associated with adolescent and parent reports of adolescent maladjustment. Father monitoring (Table 4) was negatively associated with teacher reports of adolescent maladjustment. Father involvement (Table 5) was negatively associated with adolescent maladjustment, with the exception of adolescent-reported depression.

Table 2.

Standardized Estimates and Standard Errors Across Groups for Father Rejection

Predictors	Family Type	Adolescent report of Anxiety		Adolescent report of Depression		Parent report of Positive Behavior		Teacher report of Internalizing		Teacher report of Externalizing		Teacher report of Sociometrics
		EA	MO	EA	MO	EA	MO	EA	MO	EA	MO	EA	MO
SES
	I	-	-	-	-	-	-	-	-	-	-	-	-
	S	-	-	-	-	-	-	−.16(.06)	-	−.17(.07)	-	-	-
Adolescent gender
	I	.10(.04)	.10(.04)	-	-	−.17(.04)	-	−.16(.05)	−.17(.05)	−.29(.05)	−.22(.07)	−.27(.05)	−.28(.05)
	S	.09(.04)	.11(.05)	-	-	−.16(.04)	-	−.12(.04)	−.14(.04)	−.21(.04)	−.22(.04)	−.20(.04)	−.20(.04)
Marital quality
	I	-	-	-	-	-	−.11(.05)	-	-	-	-	-	-
	S	-	-	-	-	−.09(.04)	−.08(.04)	-	-	-	-	-	-
Mother rejection (MR)
	I	.11(.02)	.11(.02)	.13(.02)	.12(.02)	.27(.05)	.22(.06)	.11(.04)	.12(.04)	.13(.04)	.11(.04)	.12(.04)	.12(.04)
	S	.10(.02)	.13(.02)	.09(.02)	.12(.02)	.26(.06)	.18(.05)	.09(.03)	.10(.04)	.10(.03)	.10(.04)	.09(.03)	.10(.03)
Father rejection (FR)
	I	.75(.20)	-	.89(.23)	-	.20(.04)	.33(.06)	.11(.04)	.15(.05)	.12(.04)	.14(.05)	-	-
	S	-	−.26(.09)	-	−.24(.08)	.24(.05)	.22(.04)	-	.11(.04)	-	.11(.04)	-	.10(.03)
MR × FR
	I	-	.16(.04)	-	.17(.05)	-	.12(.05)	-	-	-	-	-	-
	S	-	.26(.07)	-	.24(.07)	-		-	-	-	-	-	-
CG × FR
	I	−.44(.20)	.29(.06)	−.52(.23)	.30(.06)	-	-	-	-	-	-	-	-
	S	.25(.05)	.28(.06)	.22(.04)	.26(.06)	-	-	-	-	-	-	-	-
FR2
	I	−.28(.07)	-	−.33(.08)	-	-	−.15(.08)	-	-	.15(.06)	-	.14(.05)	-
	S	-	-	-	-	-	-	-	-	-	-	-	-

Note. EA = European American, MA = Mexican American, I = intact families, S = stepfamilies. Positive behavior and sociometrics reverse-coded. Adolescent model: χ²= 87.79, df = 87, RMSEA = .010, SRMR = .049; parent model: χ²= 93.70, df = 76, RMSEA = .049, SRMR = .045; teacher model: χ²= 89.70, df = 76, RMSEA = .043, SRMR = .051, p < .05 for all paths

Table 3.

Standardized Estimates and Standard Errors Across Groups for Father Discipline

Predictors	Family Type	Adolescent report of Anxiety		Adolescent report of Depression		Parent report of Positive Behavior		Teacher report of Internalizing		Teacher report of Externalizing		Teacher report of Sociometrics
		EA	MO	EA	MO	EA	MO	EA	MO	EA	MO	EA	MO
SES
	I	-	-	-	-	-	-	-	-	-	-	-	-
	S	-	-	-	-	-	-	−.17(.07)	-	−.18(.07)	-	-	-
Adolescent gender
	I	.12(.04)	.11(.04)	-	-	−.11(.04)	−.12(.04)	−.16(.05)	−.17(.05)	−.28(.05)	−.27(.06)	−.26(.05)	−.26(.05)
	S	.11(.06)	.12(.05)	-	-	−.11(.04)	−.10(.04)	−.13(.04)	−.13(.04)	−.25(.05)	−.20(.04)	−.19(.04)	−.19(.04)
Marital quality
	I	−.12(.06)	−.11(.05)	-	−.12(.06)	-	−.12(.04)	-	-	-	-	-	-
	S	-	-	-	-	−.12(.04)	−.10(.03)	-	-	-	-	-	-
Mother discipline (MD)
	I	−.16(.04)	−.16(.04)	−.19(.05)	−.17(.04)	−.17(.04)	−.19(.04)	−.33(.06)	−.20(.04)	−.22(.05)	−.17(.04)	-	−.20(.04)
	S	−.18(.04)	−.20(.05)	−.16(.04)	−.19(.04)	−.22(.05)	−.18(.04)	−.18(.04)	−.32(.05)	−.19(.04)	−.31(.05)	−.18(.04)	−.29(.05)
Father discipline (FD)
	I	−.20(.05)	−.22(.06)	−.24(.06)	−.24(.05)	−.09(.04)	−.11(.04)	-	-	-	-	-	-
	S	−.11(.04)	−.23(.06)	-	−.14(.04)	−.09(.04)	-	-	-	-	-	-	-
MD × FD
	I	-	−.12(.04)	-	−.13(.05)	-	-	-	−.15(.07)	-	−.14(.06)	-	−.16(.07)
	S	−.10(.04)	−.21(.08)	−.09(.03)	−.19(.08)	−.12(.05)	−.19(.08)	−.10(.05)	-	−.11(.05)	-	−.10(.05)	-
CG × FD
	I	.10(.04)	-	.11(.05)	.12(.05)	-	-	-	-	-	-	-	-
	S	.09(.04)	-	.08(.03)	-	-	-	-	-	-	-	-	-
FD2
	I	-	-	-	-	.21(.06)	-	.10(.04)	.14(.05)	-	.12(.04)	.10(.04)	.14(.05)
	S	.17(.06)	.24(.08)	.15(.05)	.22(.08)	-	.21(.06)	-	.09(.03)	-	.09(.03)	-	.08(.03)

Note. EA = European American, MA = Mexican American, I = intact families, S = stepfamilies. Positive behavior and sociometrics reverse-coded. Adolescent model: χ²= 68.31, df = 88, RMSEA = .000, SRMR = .039; parent model: χ²= 65.66, df = 77, RMSEA = .000, SRMR = .041; teacher model: χ²= 65.02, df = 88, RMSEA = .000, SRMR = .034, p < .05 for all paths

Table 4.

Standardized Estimates and Standard Errors Across Groups for Father Monitoring

Predictors	Family Type	Adolescent report of Anxiety		Adolescent report of Depression		Parent report of Positive Behavior		Teacher report of Internalizing		Teacher report of Externalizing		Teacher report of Sociometrics
		EA	MO	EA	MO	EA	MO	EA	MO	EA	MO	EA	MO
SES
	I	-	-	-	-	-	-	-	-	-	-	-	-
	S	-	-	-	-	-	-	−.15(.06)	-	−.16(.06)	-	-	-
Adolescent gender
	I	.11(.04)	.11(.04)	-	-	−.08(.04)	−.09(.04)	−.17(.05)	−.24(.04)	−.29(.05)	−.21(.04)	−.27(.05)	−.18(.05)
	S	.10(.04)	.11(.04)	-	-	−.08(.04)	−.07(.03)	−.22(.04)	−.14(.04)	−.28(.05)	−.21(.04)	−.20(.04)	−.21(.04)
Marital quality
	I	−.15(.06)	−.14(.05)	-	−.15(.06)	-	−.15(.04)	-	-	-	-	-	-
	S	-	-	-	-	−.23(.06)	−.20(.05)	-	-	-	-	-	-
Mother monitoring (MM)
	I	−.11(.02)	−.13(.02)	−.13(.02)	−.15(.03)	−.35(.05)	−.21(.05)	−.11(.04)	−.15(.05)	−.13(.04)	−.13(.04)	−.12(.04)	-
	S	−.11(.02)	−.17(.03)	−.10(.02)	−.17(.03)	−.17(.04)	−.20(.05)	−.19(.05)	−.14(.04)	−.21(.06)	−.21(.04)	−.19(.05)	−.21(.04)
Father monitoring (FM)
	I	-	-	-	-	-	-	−.07(.03)	-	−.08(.03)	−.09(.03)	-	-
	S	-	-	-	-	-	-	-	−.09(.04)	−.08(.03)	−.09(.04)	-	-
MM × FM
	I	.06(.03)	.10(.04)	-	-	-	-	-	-	-	−.10(.04)	-	−.12(.04)
	S	-	-	-	-	-	-	−.09(.04)	−.14(.05)	−.10(.04)	−.13(.05)	−.09(.04)	-
CG × FM
	I	-	−.18(.04)	−.17(.04)	−.20(.04)	-	-	-	-	-	-	-	-
	S	-	−.22(.05)	−.16(.03)	−.22(.05)	-	-	-	-	-	-	-	-
FM2
	I	−.07(.03)	−.11(.05)	−.08(.03)	-	-	-	-	-	-	-	-	-
	S	.23(.08)	-	.20(.07)	-	-	-	.34(.08)	-	.37(.08)	-	.34(.08)	-

Note. EA = European American, MA = Mexican American, I = intact families, S = stepfamilies. Positive behavior and sociometrics reverse-coded. Adolescent model: χ²= 85.82, df = 86, RMSEA = .000, SRMR = .043; parent model: χ²= 72.25, df = 75, RMSEA = .000, SRMR = .043; teacher model: χ²= 59.70, df = 76, RMSEA = .000, SRMR = .051, p < .05 for all paths

Hypothesis 2 was that mothering would moderate the association between fathering and adolescent adjustment. Consistent with this hypothesis, the interactions between father and mother acceptance (Table 1) suggest that when both parents are high in acceptance, adolescents receive additional benefit with regard to anxiety and depression (coefficients range from −.08 to −.15). When both parents are high in rejection (Table 2), the association between father rejection and adolescent maladjustment is augmented (coefficients range from .09 to .12); however, this interaction is found only among intact families. When both parents are high in discipline (Table 3), the negative association between father discipline and adolescent maladjustment is augmented (coefficients range from −.09 to −.21); however, this moderation is absent among intact EA families. When both father and mother are high in monitoring (Table 4), the negative association between father monitoring and teacher-reported adolescent maladjustment is augmented; however, high monitoring by both parents is also associated with increased adolescent-reported anxiety among intact families. High levels of involvement by both parents (Table 5) is associated with higher levels of depression among adolescents in intact EA families.

Hypothesis 3 was that adolescent gender would moderate the associations between fathering and adolescent outcomes. Consistent with this hypothesis, although father acceptance was generally associated with lower levels of adolescent-reported anxiety and depression in intact families, it was associated with higher levels of anxiety for boys in EA stepfamilies. Associations between father acceptance and adolescent outcomes are higher for girls than boys among all groups except intact EA families. Although father rejection was generally associated with higher levels of adolescent-reported anxiety and depression in intact families, it is related to less anxiety and depression for males in MA stepfamilies. The significant interactions with adolescent gender suggest the effect of father rejection among intact EA families is weaker for girls. For EA stepfamilies and intact MA families, father rejection has a harmful effect for girls, but no effect for boys. Although father discipline is generally associated with lower levels of adolescent-reported anxiety and depression, it is unrelated to anxiety and depression for females in EA families, and is positively related to depression for females in EA stepfamilies. Although father monitoring is generally not associated with adolescent-reported anxiety and depression, it is negatively associated with depression for females and anxiety for females in MA families. Although father involvement is associated with less maladjustment, this relationship is generally present only for boys, or girls in EA stepfamilies.

Evidence for nonlinear effects emerged across all 6 dimensions of fathering. For father acceptance, the fathering squared term is often significant for teacher-reported outcomes (coefficients range from −.11 to −.21), suggesting that very high levels of father acceptance have an added protective effect beyond the main effect; this nonlinear effect is found among all groups except intact EA families. At very high levels, the association between father rejection and adolescent anxiety/depression attenuates for adolescents in intact EA families; a similar effect is found with regard to parent-reported positive behavior in intact MA families. With regard to teacher reported outcomes, the fathering squared terms suggest that for adolescents in intact EA families, the harmful effect of father rejection on externalizing and sociometrics augments at very high levels. At very high levels, the negative association between father discipline and maladjustment attenuates, and for teacher-reported outcomes actually reverses sign and becomes a positive association. Although very high levels of father monitoring are associated with less anxiety and depression among intact EA families, among EA stepfamilies very high levels of father monitoring are associated with more maladjustment. Very high levels of father involvement are associated with less anxiety and depression among adolescents from intact EA families and MA stepfamilies.

Discussion

Our results clearly demonstrate the heuristic value of a multivariate systems approach to understanding the effects of fathering and underscore the usefulness of the Doherty et al. (1998) theoretical framework for addressing these issues. The current study examined the links between fathering and adolescent adjustment among intact and stepfamilies in both European American and Mexican American families. Our work supports prior studies (Thomson et al., 1992), suggesting that parenting in stepfamilies is of lower quality, on average, as indexed by less involvement, monitoring, and acceptance than in intact families. Second, adolescents in stepfather families are at greater risk for adjustment problems than those living with two birthparents, which may be due to higher levels of stress felt among stepfamilies or differences in parenting (Hetherington, 2006). This finding is consistent with past research indicating more negative adjustment among children in stepfamilies (Amato & Keith, 1991). Most importantly, these general are evident in Mexican as well as European American families.

Evaluation of the Links Between Fathering and Adolescent Outcomes

Next we turn to the major focus of our analysis, namely evaluation of the process models. Results supported our hypothesis that father acceptance, monitoring, consistent discipline, and interactions with the adolescent would be positively associated with positive adolescent adjustment, while father rejection would be negatively associated with positive adolescent adjustment. These findings were found across both MA and EA step and intact families while controlling for SES, adolescent gender, marital quality, and mothers’ parenting behaviors, suggesting that fathering is associated with adolescent adjustment regardless of family type in both ethnic groups. In addition, we used multiple reporters (mother, father, and adolescent reports) of fathering – a methodological feature that not only increases confidence in our findings but avoids the colinearity problems characteristic of many prior studies of the links between father involvement and adolescent outcomes. These results are also consistent with previous research (Burns & Dunlop, 1998; Flouri & Buchanan, 2002; Toth & Xu, 1999) and support our theoretical position that it is not only father involvement but also the parenting practices employed by the father that influence child adjustment. Moreover, the results underscore the fact that despite descriptive differences in average levels of some parenting dimensions across ethnic groups, the process links between fathering and adolescent outcomes are remarkably similar, a sign of cross ethnic generalizability in these father–adolescent patterns.

We also predicted that MA fathers would show higher levels of involvement, warmth, and rejection than EA fathers. While MA fathers were more involved with their adolescents and were more rejecting, they did not differ from EA fathers with regard to acceptance shown to the adolescent. These findings are generally consistent with past research (Backstrom, 2004; Hofferth, 2003; Toth & Xu, 1999). However, once SES was added as a predictor, ethnic differences in involvement and rejection levels disappeared. Therefore, these ethnic differences may be attributable to differences in SES across EA and MA families. The stress experienced by families of lower SES may impact parenting behaviors more than ethnicity. The current study also supported our hypothesis that there would be no differences between MA and EA adolescent adjustment, which coincides with past research (Hill et al., 2003).

Two additional hypotheses guided our analyses. First, the current study found that fathers’ behaviors were associated with adolescent adjustment above and beyond mothers’ parenting behavior, which underscores both the unique contribution of fathers to adolescent adjustment and the methodological importance of controlling for mother effects in fatherhood studies (Pleck, 2010). Results partially supported our hypothesis that mothering would moderate the associations between fathering and adolescent adjustment. We hypothesized that adolescents would show better adjustment when both parents displayed positive parenting behaviors and poorer adolescent adjustment when both parents displayed harsh parenting. Moderation was found in the majority of models, and was found for all five fathering dimensions when using adolescent reported outcomes. This suggests that associations between fathering and adolescent outcomes may be best understood in the context of the parenting behavior of the mother. Our findings demonstrate that when both parents discipline together, the adolescent shows better adjustment than when only one parent disciplines. When both mothers and stepfathers show high acceptance, adolescents in stepfamilies report less anxiety, depression and risky behaviors. It is likely that high consistency in parenting and coparenting provides a foundation of support that promotes adolescent development (Weissman & Cohen, 1985) by making adolescents feel more secure in their relationships with their parents. In all but one model (father-adolescent involvement and adolescent self-reports of depression), results failed to find that one parent makes up for deficiencies of the other. Thus it is not that one parent compensates for the other, but rather that having two parents who employ the same parenting practices seems to be particularly beneficial for adolescents. The finding supports Doherty et al.’s (1998) model in recognizing the moderating role of mothers’ parenting on fathers’ parenting. While particular types of moderation would sometimes cluster by family type (e.g., 70 percent of the moderation by mother discipline occurred in stepfamilies) moderation by mothering overall was not limited to a single family type or ethnicity, suggesting modest generalizability of this effect. However, the data tentatively suggest that parental agreement/disagreement may be more critical in stepfamilies in light of the less scripted and more conflict prone nature of stepfamilies (Amato & Sobolewski, 2004). This finding has implications for intervention programs for stepfamilies and underscores the importance of a “united front “ as a goal for stepparent families as a way of reducing risk of negative developmental outcomes for adolescents in stepfamilies.

Monitoring offered an interesting pattern of results with regard to moderation by mothering. For teacher reported outcomes, the co-occurrence of high monitoring by both parents was related to better outcomes. Looking at adolescent reported outcomes, the co-occurrence of high monitoring by both parents was related to better outcomes in one instance (risky behavior among EA stepfamilies), a finding that is consistent with the more influential part played by cross parent consistency in stepfamilies. However the co-occurrence of monitoring was related to more anxiety and risky behaviors in intact families. The direction of effect, in this case is unclear. Perhaps adolescents who self-report high amounts of deviant behavior elicit more active monitoring on the part of their parents in an effort to control the adolescent’s undesirable behavior. Alternatively, a major goal of adolescence is autonomy and high levels of monitoring by both parents may be viewed as inappropriate and interfering with their autonomy striving goals and/or signal a lack of trust in their adolescents’ judgment. Since monitoring is conceptualized as a negotiated disclosure process between the parent and adolescent (Stattin & Kerr, 2000), it may be that too much monitoring effort may lead to less adolescent self disclosure, and, in turn, less effective monitoring on the part of the parents. Clearly both the quality of the adolescent –parent relationship as well as family type merit more attention to better understand the conditions under which monitoring is more or less effective.

As predicted by our next hypothesis, adolescent gender moderated the associations between father-adolescent involvement and adolescent outcomes but in sometimes unexpected ways. Specifically, the frequency of father-adolescent involvement as well as father discipline showed larger associations with adolescent-reported outcomes among males. On the other hand, fathers’ acceptance, rejection, and monitoring showed larger associations with adolescent-reported outcomes among females. The fact that this moderation appears only for adolescent-reported outcomes may suggest that at this point in development, adolescent reports are especially sensitive to gender related differences. The pattern of these gender differences suggest that statements regarding the greater salience of fathering for boys rather than girls may be too global. Instead, these results suggest a more nuanced perspective, with girls being more sensitive to fathers’ affect, and boys being more sensitive to discipline, interactions, and shared activities with father. This is consistent with the view that girls are more attuned to affective aspects of interpersonal relationships while boys are more instrumentally inclined (Golombok & Fivush, 1994). Finally, this finding supports the Doherty et al. (1998) model that recognizes the importance of child characteristics, such as adolescent gender, that is linked to variations in the father-adolescent relationship.

The nonlinear nature of the associations between fathering and adolescent adjustment found in the majority of models was unexpected. While some research has conceptualized the relation between harsh parenting and child outcomes as containing a nonlinear component (Deater-Deckard & Dodge, 1997), our work suggests that this nonlinearity may extend to other parenting dimensions as well. Although normative levels of father acceptance and involvement are associated with positive adjustment in adolescents, very high levels of father acceptance and involvement appear to be especially beneficial. This suggests that when fathers are high in adolescent acceptance or are more involved with their adolescent, the adolescent, in turn, is particularly well adjusted. Across reporters, consistent discipline became harmful at very high levels. This may be interpreted as evidence that while parental discipline is beneficial at moderate levels, fathers rated very high in consistent discipline may be harmful. While it may be that adolescents who externalize elicit more discipline from parents, these associations were also found for measures of adolescent internalizing. Finally, while very high levels of monitoring were especially helpful in intact families, very high levels of monitoring were harmful in stepfamilies where less trust and closeness between stepfathers and adolescents may lead to monitoring being viewed by adolescents as intrusive and interfering. Given the paucity of work showing nonlinear associations between fathering and adolescent outcomes, further research is needed to validate these nonlinear associations between fathering and adolescent adjustment.

Limitations, Conclusions, and Future Directions

Despite the clear pattern of our findings, several limitations should be noted. First, this study was cross-sectional, and therefore, conclusions about direction of effects cannot be established. While fathering may produce changes in adolescent outcomes, it is possible that externalizing, for example, may elicit negative fathering behaviors. Longitudinal and experimental/intervention studies are needed to better establish the direction of effects between fathering and adolescent outcomes. The plausibility that variations in parenting can produce changes in child outcomes is evident from recent parenting intervention studies (Cowan & Cowan, 2002) as well as interventions designed specifically for fathers (McBride & Lutz, 2004). However, more intervention studies with MA fathers as well as stepfathers are needed to clearly address the direction of causality issue. Second, although we included parental, adolescent, and teacher behavior outcome assessments, despite mean differences in reported positive and negative parenting across adolescents from EA or MA intact and stepfather families, longitudinal analyses would allow a better understanding of patterns of stability and change across time as a function of family type and ethnicity. Third we recognize the intra group variability among Hispanic Americans and underscore that our sample was restricted to Mexican American fathers. Examination of fathers from Hispanic groups who have immigrated from other Latin countries such as Puerto Rico, Cuba, Central and South American is necessary to establish the generalizability of our findings to other groups of Latino fathers. Our findings suggest that the links between fathering and child/adolescent outcomes may be similar across groups even though the overall levels of fathering behaviors may differ across different Latino subgroups.

This study highlights the importance of parenting quality and father involvement for healthy adolescent adjustment in different types of families. Although step and intact families vary in the quality of their parenting, the links between quality of parenting and adolescent outcomes are similar across family structure and ethnicity. These findings move us beyond the previous focus on white intact middle class families and also demonstrate the applicability of core findings to diverse family situations. In short, our findings confirm that fathering matters regardless of family type or ethnic background. Moreover, fathers make unique contributions to adolescent adjustment above and beyond mothers’ parenting.

We suggest that interventions and policies to improve the father adolescent relationship, especially in stepfather families, would be valuable and beneficial for adolescent adjustment. We advocate parenting interventions for families of all ethnic groups that include fathers and suggest that they focus on strengthening the father-adolescent relationship, instead of just overall parenting. Recognition of the uniqueness of ethnicity, however, will increase the success of these interventions and should inform policy decision making on behalf of families.