Mothering vs. Fathering? Positive Parenting vs. Negative Parenting? Their Relative Importance in Predicting Adolescent Aggressive Behavior: A Longitudinal Comparison

Abstract

To understand whose parenting (mothers vs. fathers) and which type of parenting (warmth vs. hostility) is more important in predicting adolescent aggression, this study applied dominance analysis to evaluate the relative importance of four different parenting dimensions (maternal hostility, paternal hostility, maternal warmth, and paternal warmth). Four-waves of adolescent-reported longitudinal data from the PROSPER project (N = 626, 52% adolescent girls, 89% White-rural, age 12 to 15) were used to investigate longitudinal change in the relative importance of these dimensions over time. Findings reveal that at most ages, maternal hostility was relatively more important than both paternal hostility and maternal warmth in predicting adolescent aggression among adolescent girls and boys. However, paternal parenting was more important for boys at specific ages. Findings are discussed in terms of implications for interventions and further research on parenting.

Introduction

Several theoretical models implicate parenting behaviors as salient factors influencing adolescent aggressive behavior problems (e.g., Reid et al., 2002). Among multiple types of parenting behaviors that have been examined, a substantial body of empirical research has demonstrated that parent affective quality (i.e., parental warmth, parental hostility; Lippold et al., 2014) is reliably and robustly related to adolescent aggressive behavior problems (e.g., Buehler et al., 2006; Fosco et al., 2014; Schlomer et al., 2015). During adolescence, the affective quality between parents and adolescents changes dramatically with increasing parental hostility and decreasing parental warmth (Brouillard et al., 2019; Ebbert et al., 2018). Aggressive behavior problems also change significantly during this age period, which has implications for both concurrent and adult outcomes (e.g., Gronik et al., 2022). Understanding how parent affective quality influences adolescent behavior problems longitudinally is necessary for understanding how parent-child processes influence adolescent behavior problems, such as aggression.

Little is known, however, about the relative importance of mothering vs. fathering on adolescent aggression given the tendency for researchers to focus on only one parent (typically mothers) or to combine mothering and fathering into an overall parenting score (Lamb & Lewis, 2004; Stolz & Barber, 2005). Likewise, little research has been conducted that directly compares parental hostility vs. warmth for predicting adolescent aggression (though see Conger et al., 1995; Luthar & Sexton, 2007; Waller et al., 2015). Investigating the relative importance between mothering and fathering and between parental hostility and warmth for predicting adolescent aggression can inform the tailoring of intervention programs to focus on the relatively more important dimensions of parenting for the outcome; consequently this knowledge can potentially make intervention programs more efficient and effective.

The purpose of this study was to compare the relative importance of mothering vs. fathering and parental hostility vs. warmth for predicting early- to mid-adolescent aggressive behavior problems. To make these comparisons, a dominance analysis approach was applied. Dominance analysis can determine the relative importance of different predictors by evaluating their R² contributions (i.e., percentage of variance explained) across a series of regression models (Azen & Budescu, 2003, 2006). This approach provides a more comprehensive and robust comparison of predictors’ relative importance compared to other analytic approaches (e.g., individual multiple regression models, Tonidandel & LeBreton, 2011). No prior studies have used dominance analysis to examine the relative importance of mothering vs. fathering and hostility vs. warmth for predicting adolescent aggression. More importantly, as suggested by family systems theory, families are comprised of multiple relationships and subsystems that are correlated and influence one another (Cox & Paley, 1997). From this perspective, it is difficult to fully understand the effects of one certain relationship (e.g., mother-child relationship) on children’s development without considering other possible relationships (e.g., father-child relationship). Dominance analysis allows us to examine the effects of one relationship while also considering others.

The “Importance” of Parenting in Predicting Adolescent Behavior Problems

To determine the importance of parenting in predicting adolescent behavior problems, researchers often use multiple regression models to examine the unique contribution of different dimensions of parenting (i.e., mothers vs. fathers, warmth vs. hostility) in predicting adolescent behavior problems (e.g., Benson & Buehler, 2012; Dallaire et al., 2006; Eamon, 2002). In these models, the construct that explains the larger, statistically significant, proportion of variance in the outcome is considered the relatively more important predictor. However, when predictors are highly correlated with each other, which is often the case for maternal/paternal parenting indicators, multicollinearity can make it difficult to draw firm conclusions from these models (Azen & Budescu, 2003, Johnson, 2000; Tonidandel & LeBreton, 2011). More specifically, if a predictor is highly correlated with another predictor in the model, its unique contribution to explaining variance in the outcome will be underestimated (Azen & Budescu, 2003, Johnson, 2000; Tonidandel & LeBreton, 2011). Importantly, if the predictors are highly correlated, it may not be meaningful to think of the change in one predictor while all other predictors remain constant, because a change in one predictor would most likely co-occur with change in other highly-related predictors (Gelman & Hill, 2006).

Given limitations of individual multiple regression models, dominance analysis, developed by Budescu (1993) and extended by Azen and Budescu (2003), is an advocated method that can better evaluate the relative importance of predictors on an outcome when the predictors are highly correlated (Tonidandel & LeBreton, 2011). Dominance analysis allows researchers to evaluate the contribution of each predictor by itself in predicting an outcome variable as well as the contribution of each predictor in the context of all possible combinations with other predictors. In other words, dominance analysis evaluates the relative importance of a predictor in comparison to another for predicting an outcome by comparing the variance explained by the predictors across a set of models (Azen & Budescu, 2003).

To evaluate the relative importance of predictors of an outcome, any number of predictors can be included as part of the model set and, once all models have been conducted, comparisons are made between specific pairs of predictors. The R² contribution of a predictor (e.g., maternal hostility) is then compared with another predictor (e.g., paternal hostility) across all relevant models to determine which predictor is relatively more important. In addition, when predictors reflect different dimensions of the same construct (i.e., mother/father warmth/hostility), higher-order relative importance generalizations, such as mothering vs. fathering and warmth vs. hostility, can be made via sets of paired comparisons. Relative importance is defined in terms of which predictor explains more variance in the outcome across all comparable models. Importantly, both the unique contribution and the shared contribution of one predictor on an outcome can be captured and directly compared against other predictors in parallel models (Stolz & Barber, 2005). Thus, the relative importance of a predictor can be captured more comprehensively and systematically by using dominance analysis compared to a single regression model. The current study employed this approach to answer which dimensions of parenting – mothering vs. fathering in the context of hostility and warmth behaviors separately and hostility vs. warmth in the context of mothering and fathering separately– is relatively more important in predicting adolescent aggression. We also examined which specific aspect of parenting (mother hostility, father hostility, mother warmth, or father warmth), on average, is the relatively most important predictor of adolescent aggression.

Mothering vs. Fathering in Predicting Adolescent Behavior Problems

There has been a shift in the parenting literature away from focusing on just one parent or combining parent reports to research that disentangles the effects of mothering and fathering (Lamb & Lewis, 2004; Stolz & Barber, 2005). Recent empirical studies indicate that both mothers’ and fathers’ parenting behaviors uniquely predict adolescent behavior problems (e.g., Hoeve et al., 2011; Jeynes, 2016; Murray et al., 2014; Van Heel et al., 2019) and that mothering and fathering sometimes have different effects on child outcomes. For example, several studies have examined the effects of mother and father parenting separately on youth outcomes, finding different results. Hoeve et al. (2011) examined whether parenting styles of fathers and mothers had different associations with adolescent behavior problems. They found that father’s neglectful parenting style has a long-term (i.e., 5 years) association with children’s delinquency, while this is not the case for mother’s neglectful parenting style. In addition, Van Heel et al. (2019) found that adolescents who experienced more maternal harsh punishment were more likely to perform more aggressive behavior problems, while this was not the case for paternal harsh punishment. These studies suggest that mothers’ and fathers’ specific parenting behaviors may have different effects on adolescent behavior problems. However, because in these studies the parenting dimensions are in different models, they do not shed light on the relative importance of mother parenting and father parenting on youth outcomes.

A handful of studies have examined whose parenting (mothers’ or fathers’) is relatively more important in predicting adolescent behavior problems. Findings from these studies have been mixed such that some studies found fathering is more important (e.g., Chang et al., 2003), while others found mothering is more important (Gryczkowski et al., 2010; Stolz & Barber, 2005). For instance, the study by Chang et al. (2003) used structural equation modeling to examine mothers’ harsh parenting and fathers’ harsh parenting together in predicting child aggression while simultaneously accounting for each other. They found that fathers’ harsh parenting had a stronger effect on child aggression compared to mothers’. The research by Stolz and Barber (2005), to the best of our knowledge, is the first empirical work using dominance analysis to assess the contribution of mother and father support and parental behavioral control on adolescent antisocial behaviors (e.g., lying or cheating, stealing). The authors found that maternal behavioral control compared to paternal behavioral control was relatively more important in predicting sons’ antisocial behavior. However, the relative importance of parental behavioral control for predicting girls’ antisocial behavior appeared to be equivalent between mothers and fathers.

Important gaps remain. The majority of existing studies have examined the relative importance of mothering versus fathering in predicting adolescent behaviors by comparing the unique contribution of mothering and fathering in one regression-based model. Yet relying on one regression model may pose bias, especially when mothering and fathering are highly correlated (Stolz & Barber, 2005). Very few studies have used dominance analysis, which may provide a more complete assessment of the relative importance of mothering and fathering. Further, no studies have specifically examined the relative importance of mothering and fathering in terms of hostility and warmth, which are important predictors of adolescent aggressive behaviors (e.g., Fosco et al., 2014; Schlomer et al., 2015). Therefore, the current study examines the relative importance of maternal vs. paternal hostility and warmth in predicting early- to mid-adolescent aggressive behavior problems using dominance analysis.

Positive vs. Negative Parenting in Predicting Adolescent Behavior Problems

Positive parent affective quality (e.g., warmth) predicts fewer adolescent aggressive behavior problems (Benson & Buehler, 2012; Pinquart, 2017), while parent negative affective quality (e.g., hostility, conflict) predicts more adolescent aggressive behavior problems (Benson & Buehler, 2012; Conger et al., 1995; Fosco et al., 2014; Klahr et al., 2011; Pinquart, 2017; Schlomer et al., 2015). According to a famous argument proposed by Baumeister and colleagues (2001) that bad is stronger than good across psychological phenomena, negative parent affective quality may be relatively more important in predicting adolescent aggressive behavior problems compared to positive affective quality. In the current study, this means that “bad” parenting (i.e., hostility) may have stronger effects on adolescent behaviors compared to “good” parenting (i.e., warmth). Though few studies empirically investigated this with a focus on positive and negative parent affective quality, the existing limited empirical work suggests that the strength of the effect of parent negative affective quality on adolescent aggressive behavior may be stronger than positive affective quality (Benson & Buehler, 2012; Pinquart, 2017). For example, Dishion and colleagues conducted several studies examining the effects of positive and negative affective parenting on children’s behavior problems (e.g., Waller et al., 2012; Waller et al., 2015). Broadly speaking, their results suggest that parental harshness has a stronger effect on children’s behavior problems compared to parental warmth. Longitudinal research by Benson and Buehler (2012) using four-waves of data (age = 12 to 15 years) revealed that when controlling for each other, family hostility (including parent-adolescent hostility) was a stronger predictor of higher subsequent adolescent aggressive behavior than family warmth (including parent-adolescent warmth) which negatively predicted subsequent adolescent aggressive behavior. Additional longitudinal research, by Luthar and Sexton (2007) investigated the relative importance of negative and positive affective parenting behaviors for children’s behavior problems (8 to 17 years of age). Similarly, they found that the unique effect of negative parenting on children’s behavior outcomes was most pronounced. This longitudinal research is meaningful for developmentally understanding the relationship between parenting behaviors and adolescent aggression. However, to our best knowledge, no studies have longitudinally explored the relative importance of negative versus positive parenting in association with adolescent aggressive behavior specifically and no studies have used dominance analysis to systematically evaluate the relative importance of the parenting characteristics on adolescent aggression. The current study extends the negative versus positive parenting literature by applying dominance analysis to four-waves of adolescent data to systematically compare the relative importance of negative parenting (i.e., parental hostility) versus positive parenting (i.e., parental warmth) on adolescent aggressive behavior problems. By doing so, the current study can also provide additional evidence on whether bad is stronger than good is supported for parent affective quality (Baumeister et al., 2001).

Adolescent Gender Differences in Aggressive Behavior Problems

Research on gender differences generally show that boys have more aggressive behavior problems than girls and aggressive behavior problems in boys are more stable than in girls over time (Broidy et al., 2003; Card et al., 2008). Further, some research indicates that parenting may have stronger effects on adolescent boys’ aggressive behavior than girls (Griffin et al., 2000). However, the effects of parenting on adolescent aggressive behaviors may differ based on both the gender of the parent and the gender of the child (Bolkan et al., 2010; Gryczkowski et al., 2010; Stolz et al., 2005). For example, the study by Bolkan et al. (2010) found that authoritarian mothering had stronger adverse effects on adolescent boys’ behavior problems than adolescent girls’. Given the potential for adolescent gender differences, this study further examined the relative importance of mother and father warmth and hostility in predicting adolescent aggressive behavior for adolescent girls and adolescent boys, separately.

Summary of Current Study

The current study used 4 waves of longitudinal data from the PROSPER project, spanning approximately age 12 years to age 15, to examine the relative importance of different dimensions of parenting (maternal warmth, maternal hostility, paternal warmth, paternal hostility) on adolescent boys’ and girls’ aggressive behavior problems. Dominance analysis (Azen & Buescu, 2003) was used to evaluate the relative importance of different parenting dimensions on adolescent aggressive behavior problems. Because the relationship between parents and adolescents changes substantially during adolescence (i.e., increasing negativity, decreasing positivity; Brouillard et al., 2019; Ebbert et al., 2018), this study further explored how the relative importance of parenting dimensions changes during early- to mid-adolescence by examining lagged associations (e.g., parenting at age 12 → aggression at age 13). Thus, our analysis enabled us to examine relative importance of each aspect of parenting developmentally across the early adolescent period.

Three research questions were examined: (1) Which one, mother or father affective quality, is relatively more important in predicting adolescent aggression? It was hypothesized that mothering would be more important than fathering (McHale et al., 2003). (2) Which one, positive or negative parent affective quality, is relatively more important in predicting adolescent aggression? It was hypothesized that negative parent-adolescent relationship quality (operationalized by parental hostility) would be more important in predicting adolescent aggression compared to positive parent-adolescent relationship quality (operationalized by parental warmth; Baumeister et al., 2001). (3) Which of the four possible parenting dimensions (maternal hostility, paternal hostility, maternal warmth, and paternal warmth) is the most important, on average, for predicting adolescent aggression? Deriving from hypotheses 1 and 2, it was hypothesized that maternal hostility would be the most important predictor of adolescent aggression. We tested our models separately at multiple waves (from age 12 to 15), which enabled us to answer these questions at multiple time points. Given a lack of prior research, we did not have specific developmental expectations regarding whether the relative importance of different parenting dimensions is different at different ages.

Method

Participants and Procedure

Data used in current study came from a randomly selected subset of the larger PROSPER study who, in addition to in-school assessments, completed five waves of intensive in-home interviews (N = 977 at the baseline, Lippold et al., 2014; Spoth et al., 2004). The PROSPER project is a large-scale effectiveness trial of a community-university partnership model to deliver evidence-based interventions. Twenty-eight rural communities were randomized into 14 intervention and 14 control units. All 6^th graders in the 28 community school districts were invited to participate in the school-based intervention. A subsample of this larger study was recruited for the in-home data collection that included in-home surveys from the child and their parents. The families were offered different types of incentives for their participation, which was decided by each local community-University team (for more details, see Spoth et al., 2007).

N = 977 adolescents and their parents participated in the in-home data collection at wave 1. Wave 1 data (Mean age = 11.29 years, SD = .49) were collected in the fall semester of 6^th grade and data from wave 2 to wave 5 (Mean age = 11.94, 12.95, 13.93, and 14.90 years, respectively) were collected in the spring semesters of 6^th, 7^th, 8^th, and 9^th grade, respectively. In this study we focused on data from wave 2 to wave 5, which all had one-year time intervals and were most comparable. Also, because we were interested in mothering and fathering, and to account for possible confounds due to single- vs. two-parent families, only cases with two-parent families were used as the current analytical sample. Thus, our final analytic sample consisted of N = 626 families at wave 2 from the in-home portion of the data collection (N’s at wave 3 to 5 were 563, 548, and 511, respectively). Both adolescent-report and parent-report data were collected during the in-home interview, but only adolescent-report data of both mothering behaviors and fathering behaviors as well as their own aggressive behaviors were used in the current study. For consistency and to minimize missing data from one or another parent, we rely on adolescent reports of their parents’ behavior. Further, given adolescents spend more time with their peers as they become older and parental knowledge and child disclosure of youth activities decreases during this time period (Keijsers & Poulin, 2013; Lam, McHale, & Crouter, 2014), parents may have less knowledge about aggressive behavior as youth become older. Indeed, adolescents report higher levels of aggression than their parents beginning in early-adolescence (Seiffge-Krenke & Kollmar, 1998; Yang et al., 2021). For these reasons, youth may be more accurate reporters of their aggressive behavior than their parents (Goodman et al., 2010; Lippold et al., 2011; Yang et al., 2021)

In the analytic sample, 52% of adolescents (N = 324) were girls and 89% of adolescents (N = 559) were White; 72% of their parents (N = 451) had at least some college education. Wave 2 comparisons between families with and without wave 5 data showed that families with missing data at wave 5 did not differ on the variables of interest at wave 2 and 82% of participants had at least 3 out of 4 waves of data. In addition, families with missing data at wave 5 did not differ on the adolescent race and parental education level at wave 2. As a result, missing data were treated as missing completely at random and listwise deletion was used in the absence of established imputation methods for dominance analysis (Schlomer, Bauman, & Card, 2010). The PROSPER project grant was approved by the IRB at the Pennsylvania State University. The current study was deemed exempt by the IRB at University at Albany, State University of New York because the study uses deidentified secondary data. Data used in this study are currently not publicly available. This study was not preregistered.

Measures

Independent Variables – Parent Affective Quality

Positive parent affective quality was operationalized as parent-adolescent warmth and negative affective quality as parent-adolescent hostility. Parent-adolescent warmth was measured by using three Likert-type items (1 = Always to 7 = Never), which were designed to measure youth perceptions of maternal and paternal warmth toward them (Lippold et al., 2014): 1) “let you know mother/father really cares about you;” 2) “let you know that mother/father appreciates you, your ideas, or the things you do;” and 3) “act loving and affectionate toward you.” Parent-adolescent hostility was measured using four Likert-scaled items (1 = Always to 7 = Never). These items were designed to measure adolescents’ perceptions of maternal and paternal hostility toward them: 1) “get angry at you;” 2) “shout or yell at you because mother/father was mad at you;” 3) “insult or swear at you;” and 4) “when you do something wrong, how often does your mother/father lose her temper and yell at you?” (Schlomer et al., 2015). All items were recoded so that higher values indicate higher levels of parenting. Adolescents completed these items for mothers and fathers separately. The average score of three parental warmth items and average score of four parental hostility items were used to represent the constructs of parent-adolescent warmth and parent-adolescent hostility. Alpha reliabilities of items for parent-adolescent warmth and hostility ranged from .83 to .98 across waves.

Dependent Variable - Adolescent Aggressive Behavior

Aggressive behavior among adolescents was measured using the Youth Self Report, which is a widely used child-report measure for assessing children’s internalizing and externalizing behavior problems (Achenbach & Rescorla, 2001). The 17-item aggressive behavior subscale was used to measure adolescent aggressive behavior problems. Adolescents were asked to rate how true each of the 17 items was for them within the past 6 months, such as physically attacking people, destroying things, and teasing a lot. Items were scored on a 0–2 scale (0 = Never true, 1 = Sometimes true, 2 = Always true). Alpha reliabilities of adolescent aggressive behavior problems items ranged from .86 to .88 across waves.

Control Variable - Intervention Status

A family-focused and school-based sequence of evidence based preventive interventions was implemented in PROSPER intervention communities (see Spoth et al., 2004). Community-university teams selected from a menu listing family-focused and school-based interventions (for more details, see Spoth et al. 2004). For the current analytic sample, 60% (N = 372) were in the intervention group, and 40% (N = 254) were in the control group. Intervention status was included as a control variable in all analyses to account for intervention participation.

Data Analysis

Dominance analysis (Azen & Buescu, 2003) was applied in the current study to examine the relative importance of mothering vs. fathering and hostility vs. warmth for predicting adolescent aggressive behavior problems. The SAS (SAS Institute Inc., 2014) macro code, developed by Azen and Buescu (2003) (retrieved from https://sites.uwm.edu/azen/damacros/), was adapted and used in this study to calculate the dominance analysis results. In total, 15 regression models were estimated to assess the contribution of four parenting dimensions: maternal hostility (x₁), paternal hostility (x₂), maternal warmth (x₃), and paternal warmth (x₄) in predicting subsequent adolescent aggressive behavior problems [four models with all possible single predictors (x₁, x₂, x₃, x₄) + six models with all possible two predictors (x₁x₂, x₁x₃, x₁x₄, x₂x₃, x₂x₄, x₃x₄) + four models with all possible three predictors (x₁x₂x₃, x₁x₂x₄, x₁x₃x₄, x₂x₃x₄) + one model with all four predictors (x₁x₂x₃x₄)]. In all 15 models, intervention status was included as a control variable. To address the first two research questions, two sets of paired comparisons between competing predictors were made based on subsets of the 15 total regression models (Azen & Buescu, 2003). For each paired comparison, the variance contributed by each predictor by itself is compared (x₁ vs. x₂). For example, the R² associated with maternal hostility (x₁) is compared to the R² associated with paternal hostility (x₂) in single-predictor regression models predicting adolescent aggression. Second, the variance explained by the competing predictors (x₁ vs. x₂) while controlling for another predictor (x₃) is compared. For hostility, the variance explained by maternal hostility (x₁) while controlling for maternal warmth (x₃) is compared to the variance explained by paternal hostility (x₂) while controlling for maternal warmth (x₃). The third comparison is similar to the second, except (x₄) is used in place of (x₃) as the covariate. For example, the variance explained by maternal hostility (x₁) while controlling for paternal warmth (x₄) is compared to the variance explained by paternal hostility (x₂) while controlling for paternal warmth (x₄). Last, the variance explained by x₁ and x₂ are compared in models that include both covariates (x₃, x₄). For hostility, the variance explained by maternal (x₁) vs. paternal hostility (x₂) would be compared between regression models that include both maternal warmth and paternal warmth (x₃, x₄) as covariates. Complete dominance (Azen & Buescu, 2003) is determined if one of the two competing predictors explains more variance in the outcome across all four of these comparisons. To address whether maternal vs. paternal affective quality is relatively more important for predicting adolescent aggression (Hypothesis 1), the dominance pattern for maternal hostility vs. paternal hostility is evaluated in conjunction with the dominance pattern for maternal warmth vs. paternal warmth. Similarly, to address whether positive vs. negative affective quality was relatively more important (Hypothesis 2) the dominance pattern for maternal hostility vs. maternal warmth is evaluated in conjunction with the dominance pattern for paternal hostility vs. paternal warmth. To answer which among the four parenting dimensions (i.e., maternal hostility, paternal hostility, maternal warmth, or paternal warmth; Hypothesis 3) is relatively most important, the overall average contribution of each parenting dimension to the R² across all 15 models is compared. The variable that has the largest average contribution to R² across all models is considered generally dominant over the others (Azen & Buescu, 2003) and the relatively most important parenting dimension.

Results

Preliminary Analyses

As shown in Table 1, among adolescent girls, the means for maternal hostility and warmth were larger than the means of paternal hostility and warmth across ages. A series of paired samples t-tests show that, on average, mother-daughter warmth and hostility were significantly higher than father-daughter warmth and hostility across ages (ps < .05). For adolescent boys, maternal warmth was statistically higher than paternal warmth at each age (see Table 1, ps < .001). However, there were no differences between mother-son hostility and father-son hostility at any age (ps > .05).

Table 1.

Means and Standard Deviations of Examined Variables across Ages

	Age 12	Age 13	Age 14	Age 15	Longitudinal correlationa
Variables	Mean (SD)	Mean (SD)	Mean (SD)	Mean (SD)	Range
Adolescent girls
Maternal hostility	2.01 (1.03)	2.08 (1.12)	2.24 (1.20)	2.35 (1.24)	.32 to .44
Paternal hostility	1.88 (1.06)	1.95 (1.14)	2.04 (1.20)	2.14 (1.22)	.32 to .41
Maternal warmth	6.00 (1.22)	5.94 (1.28)	5.75 (1.30)	5.59 (1.44)	−.28 to −.35
Paternal warmth	5.79 (1.53)	5.63 (1.60)	5.39 (1.67)	5.09 (1.78)	−.26 to −.29
Adolescent Aggression	.20 (.24)	.22 (.26)	.26 (.28)	.29 (.28)
	Range (Skewness)	Range (Skewness)	Range (Skewness)	Range (Skewness)
Adolescent Aggression	0 to 1.35 (1.80)	0 to 1.65 (1.97)	0 to 1.54 (1.64)	0 to 1.47 (1.32)
Adolescent boys
Maternal hostility	1.99 (.97)	2.14 (1.07)	2.13 (1.01)	2.13 (.98)	.22 to .32
Paternal hostility	1.99 (1.15)	2.20 (1.31)	2.23 (1.26)	2.19 (1.26)	.13 to .25
Maternal warmth	5.81 (1.25)	5.68 (1.33)	5.57 (1.41)	5.62 (1.27)	−.17 to −.28
Paternal warmth	5.40 (1.70)	5.27 (1.74)	4.95 (1.74)	4.95 (1.68)	−.21 to −.22
Adolescent Aggression	.17 (.23)	.20 (.26)	.22 (.25)	.22 (.27)
	Range (Skewness)	Range (Skewness)	Range (Skewness)	Range (Skewness)
Adolescent Aggression	0 to 1.29 (2.32)	0 to 1.47 (2.02)	0 to 1.59 (1.69)	0 to 1.65 (1.82)

Note.

^aRange of longitudinal correlations of parenting with adolescent aggression at a subsequent age. For adolescent boys, paternal hostility at age 14 was not significantly correlated with adolescent aggression at age 15. All other longitudinal correlations were statistically significant at level .01.

Mother-daughter and father-daughter warmth were significantly correlated with lower levels of subsequent adolescent aggression (rs = −.26 to −.35, p < .01), and mother-daughter and father-daughter hostility were significantly correlated with higher levels of subsequent adolescent aggression (rs = .32 to .44, p < .01). Consistently, mother-son and father-son warmth were significantly correlated with lower levels of subsequent adolescent aggression (rs = −.17 to −.28, p < .01), and mother-son hostility was significantly correlated with higher levels of subsequent adolescent aggression (rs = .22 to .33, p < .01). Father-son hostility was significantly related to more subsequent adolescent aggression (from age 12 to age 13: r = .25, p < .01; from age 13 to age 14: r = .24, p < .01), save for the longitudinal correlation from age 14 to age 15 (r = .13, p > .05).

Relative Importance of Mothering vs. Fathering in Predicting Subsequent Adolescent Aggression

To address Hypothesis 1 (i.e., mothering would be relatively more important for predicting aggression compared to fathering), maternal hostility was compared with paternal hostility and maternal warmth was compared with paternal warmth separately for girls and boys at each age examined (i.e., age 12 to 13, age 13 to 14, and age 14 to 15, respectively). Relevant results for testing Hypothesis 1 are included in Table 2 and are described below first for girls, and then for boys.

Table 2.

Comparisons between Mothering and Fathering in the Context of Hostility and Warmth Behaviors

			Additional variance contributed by
			Girls		Boys
Panel A:			Maternal hostility	Paternal hostility	Maternal hostility	Paternal hostility
Parental hostility behavior	Age 12 to age 13	Comp 1a	.162	.108	.056	.071
		Comp 2a	.102	.067	.027	.044
		Comp 3a	.105	.046	.024	.034
		Comp 4a	.092	.049	.022	.097
	Age 13 to age 14	Comp 1a	.119	.103	.124	.061
		Comp 2a	.054	.052	.067	.021
		Comp 3a	.075	.054	.068	.022
		Comp 4a	.053	.044	.065	.020
	Age 14 to age 15	Comp 1a	.242	.196	.050	.019
		Comp 2a	.140	.129	.018	.002
		Comp 3a	.184	.133	.023	.001
		Comp 4a	.139	.117	.018	0
Panel B:			Maternal warmth	Paternal warmth	Maternal warmth	Paternal warmth
Parental warmth behavior	Age 12 to age 13	Comp 1b	.075	.080	.035	.045
		Comp 2b	.015	.023	.006	.013
		Comp 3b	.034	.018	.008	.008
		Comp 4b	.013	.014	.005	.006
	Age 13 to age 14	Comp 1b	.077	.065	.071	.046
		Comp 2b	.012	.021	.014	.006
		Comp 3b	.026	.016	.007	.007
		Comp 4b	.010	.013	.014	.007
	Age 14 to age 15	Comp 1b	.114	.081	.043	.045
		Comp 2b	.012	.023	.011	.018
		Comp 3b	.047	.018	.002	.027
		Comp 4b	.001	.012	.011	.022

Note. Comp = Comparisons. Comp 1a = predictor itself, Comp 2a = predictor and maternal warmth, Comp 3a = predictor and paternal warmth, Comp 4a = predictor, maternal warmth, and paternal warmth. Comp 1b = predictor itself, Comp 2b = predictor and maternal hostility, Comp 3b = predictor and paternal hostility, Comp 4b = predictor, maternal hostility, and paternal hostility. Coefficients in bold are the larger of the comparison pair.

To test Hypothesis 1 among girls, four comparisons were initially made for maternal vs. paternal hostility (see Panel A in Table 2, Girls column). First, the amount of variance each predictor explained in aggressive behavior problems within single-predictor regression models (controlling for intervention status) was compared (Comp 1a). At age 12 to age 13, the maternal hostility R² was .162 and the paternal hostility R² was .108, indicating that maternal hostility explained more variance in girls’ aggressive behavior problems compared to paternal hostility. Second, two analyses were conducted to determine the variance explained by (1) maternal hostility while controlling for maternal warmth and (2) paternal hostility while also controlling for maternal warmth (Comp 2a). When comparing the ΔR²’s when adding maternal or paternal hostility between these two analyses, maternal hostility (while controlling for maternal warmth) explained more variance than paternal hostility (while also controlling for maternal warmth; .102 vs. .067 see Table 2). Third, a similar pair of analyses were conducted except paternal warmth was used as the covariate instead of maternal warmth (Comp 3a). In these analyses, maternal hostility explained more variance in girls’ aggressive behavior problems (.105) compared to paternal hostility (.046). Last, a fourth comparison was made by conducting two additional analyses wherein the variance explained by maternal hostility while controlling for maternal and paternal warmth was compared to the variance explained by paternal hostility while also controlling for the same covariates (Comp 4a). These analyses showed that maternal hostility (.092) explained more variance in the outcome compared to paternal hostility (.049). Thus, across all four comparisons between maternal and paternal hostility among girls at age 12 to age 13, maternal hostility explained more variance in aggressive behavior problems compared to paternal hostility. This same procedure was used to test the relative importance of maternal and paternal hostility from age 13 to 14 and from age 14 to 15 (see Panel A in Table 2, Girls column). In total, maternal hostility explained .2 to 6.0% more variance compared to paternal hostility across all comparisons and ages. This pattern of findings indicates that maternal hostility completely dominated (Azen & Buescu, 2003) paternal hostility for predicting adolescent girls’ aggressive behavior problems during early- to mid-adolescence.

When comparing the relative importance of maternal vs. paternal warmth among girls, the picture was less clear (see Panel B in Table 2 Girls column). For girls, across ages, neither maternal warmth nor paternal warmth emerged as completely dominant. In fact, among all comparisons across ages, maternal and paternal warmth were evenly split: maternal warmth explained more variance in 50% of the comparisons and paternal warmth explained more variance in the other 50%. Thus, in the context of parental warmth, mothering and fathering appear to be equivalent for predicting girls’ aggressive behavior problems during early- to mid-adolescence. In sum, these findings partially support Hypothesis 1 among girls: maternal hostility was relatively more important than paternal hostility but the importance of maternal and paternal warmth were essentially equivalent.

Among boys, comparing the relative importance of maternal vs. paternal hostility and maternal vs. paternal warmth followed the same analyses sequence and comparison procedure as described for girls. As can be seen in Table 2 (Boys column) from age 12 to age 13, fathering, whether paternal hostility or warmth, completely dominated mothering. That is, in all comparisons at age 12 to 13, paternal hostility and warmth explained more variance in adolescent boys’ aggression compared to maternal hostility and warmth (though there was one equivalent comparison, .008 vs. .008; see Table 2). However, from age 13 to age 14, mothering completely dominated fathering for both warmth and hostility, suggesting that mothering became more important at that age for predicting adolescent boys’ aggression (with the exception of one equivalent comparison, .007 vs. .007; see Table 2). Subsequently, from age 14 to age 15 maternal hostility completely dominated paternal hostility, while, during the same age period, paternal warmth completely dominated maternal warmth. Overall, this pattern of results indicates that among adolescent boys the relative importance of different parenting dimensions changed over time. Maternal hostility became more important for predicting boys’ aggressive behavior problems compared to paternal hostility over time, while at the same time paternal warmth became more important for predicting boys’ aggression than maternal warmth. In total, among boys, the hypothesis that mothering would be relatively more important for predicting aggression compared to fathering (Hypothesis 1) was not supported.

Relative Importance of Positive vs. Negative Parenting in Predicting Subsequent Adolescent Aggression

To address hypothesis 2 – that negative parent affective quality would be relatively more important compared to positive parent affective quality for predicting adolescent aggressive behavior problems – comparisons were made between maternal hostility vs. maternal warmth and paternal hostility vs. paternal warmth. The analyses and comparison procedure were the same as described for Hypothesis 1 and the results are summarized in Table 3 for girls and boys across age periods. Specifically, among girls, maternal hostility compared to maternal warmth explained more variance in aggressive behavior problems across all comparisons (Comps 1a-4a) ranging from .5% to 13.8 % more variance (see Panel A in Table 3, Girls column). These results indicate that, among girls, maternal hostility completely dominated maternal warmth and suggests that maternal hostility was relatively more important for predicting adolescent girls’ aggression compared to maternal warmth during early- to mid-adolescence. Comparisons between paternal hostility and paternal warmth among girls (Panel B in Table 3, Girls column) show that paternal hostility was only completely dominant over paternal warmth at ages 14 to 15. At other age periods, neither parental hostility or warmth was completely dominant over the other. These findings suggest that only at mid-adolescence (age 14 to 15) did paternal hostility became relatively more important than paternal warmth for predicting aggressive behaviors among girls. Prior to this age the relative importance of paternal hostility vs. warmth was mixed, though it is worth noting that in the majority of comparisons paternal hostility explained as much or more variance in girls’ aggressive behavior problems than paternal warmth (see Panel B in Table 3, Girls column). Taken together, these findings provide partial support for Hypothesis 2 among girls and suggest that parental hostility is, principally, relatively more important for predicting girls’ aggressive behavior problems than parental warmth.

Table 3.

Comparisons between Parental Hostility and Parental Warmth in the Context of Mothering and Fathering Behavior

			Additional variance contributed by
			Girls		Boys
Panel A:			Maternal hostility	Maternal warmth	Maternal hostility	Maternal warmth
Mothering behavior	Age 12 to age 13	Comp 1a	.162	.075	.056	.035
		Comp 2a	.065	.034	.006	.008
		Comp 3a	.105	.014	.024	.002
		Comp 4a	.061	.017	.004	.002
	Age 13 to age 14	Comp 1a	.119	.077	.124	.071
		Comp 2a	.031	.026	.063	.007
		Comp 3a	.075	.023	.068	.027
		Comp 4a	.028	.013	.063	.025
	Age 14 to age 15	Comp 1a	.242	.114	.050	.043
		Comp 2a	.083	.047	.031	.002
		Comp 3a	.184	.046	.023	.006
		Comp 4a	.077	.030	.026	.005
Panel B:			Paternal hostility	Paternal warmth	Paternal hostility	Paternal warmth
Fathering behavior	Age 12 to age 13	Comp 1b	.108	.080	.071	.045
		Comp 2b	.011	.023	.021	.013
		Comp 3b	.067	.019	.044	.012
		Comp 4b	.009	.009	.020	.007
	Age 13 to age 14	Comp 1b	.103	.065	.061	.046
		Comp 2b	.015	.021	<.001	.006
		Comp 3b	.052	.011	.021	.002
		Comp 4b	.013	.010	<.001	<.001
	Age 14 to age 15	Comp 1b	.196	.081	.019	.045
		Comp 2b	.037	.023	<.001	.018
		Comp 3b	.129	.013	.002	.008
		Comp 4b	.026	.012	<.001	.008

Note. Comp = Comparisons. Comp 1a = predictor itself, Comp 2a = predictor and paternal hostility, Comp 3a = predictor and paternal warmth, Comp 4a = predictor, paternal hostility, and paternal warmth. Comp 1b = predictor itself, Comp 2b = predictor and maternal hostility, Comp 3b = predictor and maternal warmth, Comp 4b = predictor, maternal hostility, and maternal warmth. Coefficients in bold are the larger of the comparison pair.

Among boys, maternal hostility completely dominated maternal warmth at age 13 to age 14 and at age 14 to age 15 (see Panel A in Table 3, Boys column). At age 12 to age 13, maternal hostility explained more variance than maternal warmth in 3 of the 4 comparisons. These results indicate that, with one exception, maternal hostility completely dominated maternal warmth for predicting boys’ aggressive behavior problems during early- to mid-adolescence. In contrast, paternal hostility vs. paternal warmth comparisons showed a different pattern across time (Panel B in Table 3, Boys column). At age 12 to age 13, paternal hostility completely dominated paternal warmth. No clear dominance pattern emerged at age 13 to age 14. However, at age 14 to 15, paternal warmth completely dominated paternal hostility. This pattern of results suggests that paternal hostility was relatively more important than paternal warmth for predicting boys’ aggressive behavior problems during early adolescence and then a transition occurred and paternal warmth became relatively more important by mid-adolescence. Overall, these findings provide partial support for Hypothesis 2 among boys with a caveat that paternal warmth appears to play an important role in adolescent boys’ aggressive behavior problems as they transition from early- to mid-adolescence specifically.

Relatively Most Important Parenting Dimension for Predicting Adolescent Aggression

To determine which parenting dimension was relatively most important for predicting adolescent aggression, the average contributions of each of the four parenting predictors (maternal/paternal hostility/warmth) were compared (Azen & Buescu, 2003). Averages reflect the mean R² contributions across all 15 models within each age period examined for each of the 4 predictors. Gender stratified results are shown in Table 4. Among girls, maternal hostility consistently explained more variance, on average, than paternal hostility, maternal warmth, or paternal warmth by at least 4.8% (.091 − .043 = .048) and as much as 6.4% (.091 - .027). These findings indicate that maternal hostility generally dominated (Azen & Buescu, 2003) the other three parenting dimensions in predicting girls’ aggressive behavior problems during early- to mid-adolescence. A similar pattern was found among boys: Maternal hostility explained more variance in boys’ aggressive behavior problems compared to paternal hostility, maternal warmth, and paternal warmth at age 13 to age 14 and age 14 to age 15. At age 12 to age 13, paternal hostility explained the most variance. These results indicate that among boys, parental hostility, on average, is the relatively most important parenting factor though there is change across time regarding which parents’ hostility is most important for predicting aggressive behavior problems. In total, Hypothesis 3 (maternal hostility would be most important) was supported among girls and partially supported among boys.

Table 4.

Overall Comparisons among Maternal Hostility, Paternal Hostility, Maternal Warmth, and Paternal Warmth

		Overall average explained variance by each predictor
		Maternal hostility	Paternal hostility	Maternal warmth	Paternal warmth
Girls	Age 12 to age 13	.091	.043	.027	.028
	Age 13 to age 14	.055	.043	.027	.024
	Age 14 to age 15	.129	.095	.041	.031
Boys	Age 12 to age 13	.022	.041	.011	.016
	Age 13 to age 14	.073	.021	.026	.014
	Age 14 to age 15	.029	.006	.014	.022

Note. Coefficients in bold are the larger of the comparison pair.

Results Summary

Table 5 provides a summary of the results and dominance interpretation of the 15 models and related comparisons described above. Generally speaking, with one exception (age 12 to 13 for boys), maternal hostility emerged as relatively more important for predicting adolescent aggressive behavior problems compared to paternal hostility. The relative importance of maternal vs. paternal warmth appeared more balanced; effectively equal at each time point for girls and alternating between paternal and maternal for boys. When examining hostility vs. warmth among mothers specifically, it was nearly universal across time points and genders that maternal hostility was more important than maternal warmth. For hostile vs. warm fathering the pattern suggests that paternal warmth and hostility are equally important for girls during early adolescence but paternal hostility may be more important during mid-adolescence. Among boys, the pattern essentially reversed where paternal hostility was most important at age 12 to 13, equal at age 13 to 14, and then warmth became more important at mid-adolescence (age 14 to 15). With all results and comparisons considered together, maternal hostility emerged as relatively most important for predicting adolescent aggression, though paternal hostility may have a role for boys during early adolescence.

Table 5.

A Summary of the Relative Importance of Parenting Behaviors

		Relatively more important parenting behavior
		Age 12 to age 13	Age 13 to age 14	Age 14 to age 15
Mother vs. Father Hostility	Girls	Maternal hostility	Maternal hostility	Maternal hostility
	Boys	Paternal hostility	Maternal hostility	Maternal hostility
Mother vs. Father Warmth	Girls	Equal	Equal	Equal
	Boys	Paternal warmth	Maternal warmth	Paternal warmth
Hostility vs. Warmth in Mothers	Girls	Maternal hostility	Maternal hostility	Maternal hostility
	Boys	Equal	Maternal hostility	Maternal hostility
Hostility vs. Warmth in Fathers	Girls	Equal	Equal	Paternal hostility
	Boys	Paternal hostility	Equal	Paternal warmth
		Most important parenting behavior in general
Mothers vs. Fathers vs. Hostility vs. Warmth	Girls	Maternal hostility	Maternal hostility	Maternal hostility
	Boys	Paternal hostility	Maternal hostility	Maternal hostility

Discussion

As many dimensions of parenting have been found to predict adolescent aggression (e.g., Bendezu et al., 2018; Pinquart, 2017; Van Heel et al., 2019), it is particularly crucial to understand the relative importance of these different dimensions of parenting. Knowing which parenting behaviors are relatively more important can point to areas where research should be emphasized. Armed with such findings, intervention programs could be more effectively built around the relatively more important parenting behaviors to improve their efficiency and outcomes. Indeed, knowing the relative importance of different parenting dimensions could be useful in developing policies that are maximally impactful. To understand the relative importance of mothering vs. fathering and positive parenting vs. negative parenting in explaining adolescent aggressive behavior problems, the current study employed dominance analysis to investigate whose (mothers vs. fathers) and which (negative vs. positive) parenting is relatively more important in explaining adolescent aggressive behavior problems. Four-waves of longitudinal data were used to reveal possible changes in the relative importance of different parenting dimensions during early- to mid-adolescence (i.e., from age 12 to age 15). Gender-stratified analyses were conducted to test the hypotheses that mothering would be relatively more important than fathering and that hostility would be relatively more important than warmth for predicting adolescent aggressive behavior problems. These hypotheses were partially supported among girls and received limited support for boys. Developmental patterns were detected, especially for boys, and the hypothesis that maternal hostility would generally dominate other parenting dimensions examined was supported.

Relative Importance of Mothering vs. Fathering

The hypothesis that mothering, compared to fathering, would emerge as relatively more important for predicting girls’ aggressive behavior problems was partially supported. Among all comparisons across all ages examined, maternal hostility completely dominated paternal hostility. However, no clear dominance pattern emerged for maternal warmth vs. paternal warmth among girls. The finding that maternal hostility was relatively more important for predicting girls’ aggression than paternal hostility is consistent with prior studies that also found mothering is relatively more important than fathering (Stolz et al., 2005; Keijsers et al., 2010; Van Heel et al. 2019). Nonetheless, it is also worth noting that this finding is inconsistent with at least one prior study that compared mothers’ vs. fathers’ harsh parenting and found that fathers’ harsh parenting had a stronger effect on child aggression in kindergarten compared to mothers (Chang et al., 2003). When considered together, the current and prior research suggest that the relative importance of the negative side of mothering versus fathering for children’s behavior problems may change over development. As children develop, the nature of parent-child relations also change, which may be especially true for girls since mother-daughter relationship quality is generally better than father-daughter relationship quality during adolescence (Branje et al., 2010). As a result of the closeness of the mother-daughter relationships, mothers’ hostility toward their daughters may be relatively more influential and girls may respond more negatively to mothers’ hostility. Thus, it is possible that the effect of mothers’ hostility on daughters becomes more important compared to fathers’ as daughters develop into adolescence. Additional studies are needed, however, to more closely track how changes in different aspects of parent-child relationships result in differences in the relative importance of mothering vs. fathering across development.

Among boys, the importance of mothering vs. fathering differed over time. From age 12 to 13, paternal behaviors (hostility and warmth) emerged as relatively more important than maternal behaviors, yet at age 13 to 14 this pattern switched and maternal behaviors were relatively more important. Then at age 14 to 15, maternal hostility dominated paternal hostility but paternal warmth dominated maternal warmth, suggesting that the relative importance of mothering vs. fathering may differ based on the type of parenting. Parents, especially fathers, are more involved with their same- than opposite-sex children (Raley & Bianchi, 2006) and therefore fathers’ parenting may be relatively more important for boys compared to mothers, at least during early adolescence. The characteristics of parental involvement differs between mothers and fathers (Lamb & Lewis, 2004), such that mothers tend to engage more in caretaking and routine family tasks while fathers are more likely to be involved in play and recreation. On one hand, maternal hostility may become the relatively stronger influence on adolescent boys’ aggression since hostility would be more incongruous with expected maternal caretaking-type behaviors. On the other hand, at the transition to adolescence, youth, particularly boys, experience increasing pressure to conform to traditional gender roles (Crouter et al., 2007). Fathers who participate in boys’ play and recreation activities may help boys generate attachment to their fathers and, as a result, better regulate aggressive behaviors (Fletcher et al., 2013). Consequently, the role of fathers’ warmth may be more important for boys over time than mothers’.

Relative Importance of Negative vs. Positive Parenting

Regarding the relative importance of positive parenting versus negative parenting, this study demonstrated that the effect of negative parenting (i.e., parental hostility) on adolescent girls’ aggression was stronger than positive parenting (i.e., parental warmth) but only for mothering, partly consistent with previous work (Baumeister et al., 2001; Benson & Buehler, 2012; Pinquart, 2017). Reflecting this overall pattern of findings, maternal hostility was found to be the most important parenting dimension for predicting adolescent girls’ aggression compared to the other three (i.e., paternal hostility, maternal warmth, and paternal warmth). One implication of these findings is that intervention strategies designed to reduce adolescent girls’ aggression may be most effective by focusing on mothers’ hostility compared to the other parenting dimensions examined. These findings also suggest that, among girls in particular, the tenet that “bad is stronger than good” may hold. Why should parental hostility play such a prominent role for predicting girls’ aggression? Baumeister and colleagues (2001) speculated an evolutionary argument that it should behoove organisms to preferentially attend to environmental threats to help ensure survival and therefore reproduction. Thus, negative interactions may be more salient than positive ones due to the embedded potential threat. Another explanation might be that the association between parental hostility and child aggression might actually reflect a gene-environment correlation (Burt & Neiderhiser, 2009). Yet another explanation is that negative information elicits stronger affective reactions than positive (though see Unkelbach, Koch, & Alves, 2019), thus leading to stronger impacts on behavior. The findings from this study are consistent with the general notion that bad is stronger than good, but the reasons for why are not well understood, can likely be explained at multiple levels of analyses (see Nesse, 2013), and should not be considered a universal “rule” but rather a trend that has nuance based on what is being studied and when, partially with regard to development.

In line with this thinking, the relative importance of parental hostility versus parental warmth among boys showed change over time: paternal hostility was relatively more important than paternal warmth during early adolescence, but in later adolescence, paternal warmth became relatively more important for predicting boys’ aggression. It is possible that during later adolescence, aggression is more strongly tied to youth autonomy (Moffitt, 1993). Paternal warmth may indicate more autonomy support, and henceforth help reduce aggression later in adolescence. Although there is some evidence that positive relationships with fathers may be critical for adjustment (Lam et al., 2012; Lippold et al., 2016), more studies across a longer developmental period are needed to help place this finding in context and replication is clearly needed. This finding also suggests that the bad is stronger than good principle may have some developmental nuance when it comes to parenting and aggression, especially between fathers and sons in middle adolescence and possibly thereafter. In middle adolescence, positive fathering behavior, which may be a better fit for boys’ developmental needs, may become relatively more important. This finding additionally implies that dealing with fathers’ hostility toward sons during early adolescence may be important for intervening in boys’ aggression. Likewise, promoting paternal warmth toward their sons may be beneficial during later adolescence.

Implications on Future Parenting Research

Recent research indicates that both mothering and fathering play a crucial role in children’s development (Cabrera et al., 2014). In response to this and to fill in the dearth of research on fathering, an increasing body of studies focus on understanding how fathers influence children’s development (Cabrera et al., 2018; Fitzgerald et al., 2021; Lamb & Lewis, 2004). The findings from the current study indicate that, overall, maternal hostility was the most important factor for predicting adolescent aggression. Stemming from these findings, a provocative question could be posed: Is this line of research on fathering really necessary? If fathering is shown to be less important than mothering, why study fathers at all? We contend that this study provides evidence that fathering is indeed important, even if, when examined globally, maternal behavior emerged as dominant. Specifically, our results showed that fathering is actually more important than mothering for adolescent boys’ aggression at some ages. Future researchers are encouraged to examine why fathering is relatively more important for adolescent boys than mothering at certain developmental times. For adolescent girls, future research is needed to understand what role fathers’ hostility plays in affecting their daughters’ aggression, as it was found to be less important in predicting aggression compared to mothers.’ Likewise, fathers’ and mothers’ warmth emerged to be equally important for girls’ aggression during early- to mid-adolescence, which suggests that future research is needed to evaluate the co-influence of mothers’ and fathers’ warmth in regulating their adolescent girls’ aggression.

Interpreting the Importance of Mothering vs. Fathering in the Context of Two Dimensions of Parenting

Across adolescent girls and boys, the hypothesis that mothering would be relatively more important than fathering for predicting early- to mid-adolescent aggressive behavior problems received partial to limited support. Given this finding, it is important to point out that hypotheses about the relative importance of mothering and fathering, using dominance analysis or otherwise, may inevitably be tied to the dimensions of parenting evaluated. In the current study, conclusions about the relative importance of mothering vs. fathering were tied specifically to the hostility and warmth dimensions (i.e., parental effective quality). Sweeping conclusions about the relative importance of mothering vs. fathering would require using a much larger array of parenting dimensions and systematic evaluation of each. To this end, the current data highlights a likely general finding that whether mothering or fathering emerges as more important for predicting a child outcome will depend on the type of parenting behavior examined. Further, the current results also suggest that the relative importance of mothering and fathering parenting dimensions will also be sensitive to developmental changes among children and adolescence, the gender of the child, and the outcomes examined. As a result, broad conclusions about mothering or fathering being more or less important for child development are likely untenable, as evidenced by the limited Hypothesis 1 support in the current study. Nonetheless, the current findings, particularly for girls, do suggest that there may be some categories of parenting for which maternal behavior may be relatively more important than paternal behavior. Only one aspect of negative parenting (hostility) was examined in the current study and additional research is needed to determine if other aspects of negative parent-adolescent affective quality show the same dominance pattern detected in this study for maternal hostility among girls.

Interpreting the Current Findings as Relative

Readers should also be cautioned that findings regarding the importance of one or another parenting dimension should not be interpreted as meaning that other dimensions of parenting are unimportant. All four of the parenting behaviors examined in this study are demonstrably salient and important factors impacting adolescent behavior problems (Pinquart, 2017). The importance of one parenting behavior compared to the other in current study should only be interpreted to a relative degree. For example, both paternal hostility and maternal hostility were important for predicting adolescent aggression; however, compared with paternal hostility, maternal hostility was relatively more important for predicting adolescent girls’ aggression. This example highlights another important consideration when interpreting these findings, that “importance” was defined in statistical terms as the amount and consistency of variance explained in the outcome via regression models. Defining what are important parenting dimensions will vary across contexts and individuals, as might be found in clinical or other applied settings, and the findings herein about what is relatively more important should be contextualized by those settings. Indeed, an additional interesting avenue of research would be to examine the relative importance of parenting dimensions in non-community-based samples.

Limitations

Several limitations of this study should be mentioned. First, because the findings from dominance analysis are regression-based, the generalizability of the results are equivalent to any regression-based analysis. However, this approach does not attend to the statistical significance of the predictors. A predictor may explain 1% more variance in an outcome compared to another but whether this 1% is statistically different than zero and whether 1% has practical meaning would need to be further examined. As described by Azen and Budescu (2003), additional simulation studies could provide helpful information. Nonetheless, it could be argued that demonstrating dominance of one predictor over another (even by 1%) indicates a highly reliable finding. Consistent findings across several analytic models suggest robust findings that may be more generalizable than single-regression results, even in the presence of statistical significance, which will be more sensitive to the variances of a single analysis.

Second, between-wave longitudinal comparisons were used in this study to determine the relative importance of parenting dimensions. This approach does not capture long-term trajectories of the relative importance of parenting dimensions. Understanding the long-term trajectories of the relative importance of parenting in predicting adolescent aggression, and potential change over time, is important for delineating the development of aggressive behavior problems. Additional relative importance studies using advanced longitudinal methodologies are needed.

A third possible limitation is that only youth reports were used to measure both adolescent aggression and parenting behaviors. During adolescence, however, adolescents likely know more about their own aggressive behaviors compared to their parents (Keijsers & Poulin 2013; Smetana et al 2009; Yang et al., 2021). As a result, adolescent reports of their own aggressive behaviors may more accurately reflect the actual prevalence of the behavior compared to parent report (see Yang, et al. 2021 for an overview). In addition, because the focus of this study was on predicting adolescent aggression from parenting behaviors, adolescent’s own perceptions of their parents’ behavior toward them is potentially a more salient and relevant predictor of their behavior compared to parents’ perception of their own parenting. Adolescent perceptions of parenting are suggested to be a better predictor of their adjustment (Shoval et al., 2013). Nonetheless, using the same reporter could inflate correlations due to method variance. If inflation were present, it would be equal across the relative comparisons made in this study and therefore would not influence the results or conclusions. Further, it is possible that differences in parent and adolescent perceptions of parenting could also be predictors of adolescent aggression (De Los Reyes & Kazdin, 2005; De Los Reyes et al, 2009; Goodman et al, 2010; Lippold et al, 2011, 2013, 2014). Examining the relative importance of differences in parent-adolescent perceptions of parenting is a fruitful area for future research.

Fourth, as one of our interests was to compare mothering and fathering, in order to account for the possible confounds due to single- vs. two-parent families, we only used two-parent families in the analyses. Caution is suggested when generalizing the current results to single-parent households. It might be that mothers’ (or fathers’) behaviors are always dominant in the single-parent households, depending on the household composition. Aggressive behavior problems also have several different subtypes, such as verbal, physical, and interpersonal, and researchers found that there are gender differences in different subtypes of aggression (Bjorkqvist, 2018). To fully understand the gender differences on relative importance of different dimensions of parenting behaviors for adolescent aggression, future research is needed to test this by measuring different subtypes of aggression. Finally, it should be noted that the majority of participants in current study were Caucasian. The findings in current study may not be generalizable to populations of different ethnicities. Future research using diverse samples is necessary to replicate the current study.

Conclusion

This study explored the relative importance of mothering versus fathering and positive parenting versus negative parenting in predicting adolescent aggressive behavior problems using a dominance analysis approach (Azen & Budescu, 2003; Budescu, 1993). The relative importance of mothering versus fathering and positive parenting versus negative parenting differs for boys and girls. Although maternal hostility was relatively more important than paternal hostility or maternal warmth in predicting adolescent girls’ aggression, paternal warmth emerged as potentially having an important role in predicting boys’ aggression during mid-adolescence. In addition, maternal hostility was the most important parenting dimension for girls; however, during early-adolescence, paternal hostility was relatively most important for predicting adolescent boys’ aggression. Taken together, this study indicates that maternal hostility played a central role in predicting adolescent girls’ aggression. This study also indicates that fathers may make crucial contributions to the development of adolescent boys’ aggression and more attention is needed on paternal parenting for adolescent boys’ aggressive behavior problems (Lamb & Lewis, 2004).