Are Discrepancies Between Father and Adolescent Perceptions of Harsh Parenting and Conflict Associated with Adolescent Mental Health Symptoms?

Abstract

Though differences in informant perceptions of family processes are associated with poorer health, few studies have examined discrepancies between father- and adolescent-report of family phenomena and their impact on adolescent mental health. This study examined how father and adolescent-reported parenting and the differences in their perceptions is related to adolescent mental health. Participants were 326 father–adolescent dyads (Fathers: M_age = 41.2; Adolescents: 7th grade students, M_age = 12.0, 48.5% female). Overall, analyses revealed significant main effects of father and/or adolescent report of father–adolescent conflict and harsh parenting on adolescent internalizing and externalizing symptoms. Analyses revealed two instances in which discrepancies between father- and adolescent-report of family phenomena was related to adolescent mental health. Given the mixed nature of the findings based on the outcome reporter, the current study discusses implications for discrepancy research and future directions to better understand discrepant perceptions as useful information on their own. The parent clinical trial is registered at ClinicalTrials.gov (Identifier: NCT03125291, Registration date: 4/13/2017).

Introduction

Parent and adolescent differences in perceptions and reporting of family phenomena remains a complex issue in the fields of psychology, human development, and family studies. Differences in a parent’s experience and report of parenting and family interactions versus their child’s may generate false links between constructs, obscure important family processes, and obfuscate potential risk factors for child behavioral health (De Los Reyes et al., 2019). For example, discrepancies between a parent’s perceptions of their adolescent’s internalizing symptoms may thwart identification of mental health need (Bajeux et al., 2018; Orchard et al., 2019). Similarly, discrepancies between parent and adolescent on parenting behaviors may complicate understanding of parenting’s associations with developmental and health outcomes such as attention deficit/hyperactivity disorder, oppositional defiant disorder, and conduct disorder (Dirks et al., 2012). Indeed, the question of which informant to use when assessing youth mental health and experiences is often debated (De Los Reyes et al., 2019). In recent years, the favored solution has been to collect data from multiple informants and coalesce parent and youth perspectives using inventive methods (Laird & De Los Reyes, 2013). However, few studies examine the discrepancies themselves and what they portend. Additionally, fathers are rarely included studies of parenting despite their central role in youth parenting and socialization. This study examines how discrepancies between father and adolescent perceptions of harsh parenting and family conflict were associated with adolescent mental health symptomatology and if these associations varied based on adolescent sex.

Among parenting and family processes, harsh parenting and parent–adolescent conflict are particularly salient predictors of later mental illness and maladjustment (Gonzales et al., 2017; Lunetti et al., 2022). Harsh parenting, or coercive behaviors and negative emotional expressions toward children, has long lasting negative impacts on adolescent health and development (Pinquart, 2021). Encompassing both verbal aggression (e.g., name calling, yelling, shaming) and physical aggression, harsh parenting is associated with greater internalizing and externalizing symptoms (Deardorff et al., 2013; Pinquart 2017), conduct problems (Bauer et al., 2022; Caples & Barrera, 2006), and antisocial behavior (Afifi et al., 2019; Burnette et al., 2012) among adolescents. Conflict between adolescents and their parents is also widely understood to predict negative adolescent outcomes (Burt et al., 2007; Klahr et al., 2011) including internalizing problems (Caples & Barrera, 2006; Kuhlberg et al., 2010), conduct problems (Caples & Barrera, 2006), low self-esteem (Caples & Barrera, 2006; Li & Warner, 2015), and adolescent risk-taking (Qu et al., 2015; McCormick et al., 2016). Indeed, as with other dyadic relationships, some father–adolescent conflict is normative (Weymouth et al., 2016). Low to moderate levels of parent–adolescent conflict is also helpful for adolescent development and provides opportunities for adolescents to learn key communication, adaptive coping skills, and interpersonal, social emotional, and conflict resolution (Branje et al., 2009; Branje, 2018). The current study examines the detrimental aspects of conflict and its association with adolescent mental health symptom.

Although the broad impacts of critical socialization experiences are well understood, scholars continue to struggle with how to contend with discrepant perceptions and reports of these experiences (Laird & LaFleur, 2016). Indeed, differences in adolescent and parent report of the same phenomena are widely documented (e.g., Abar et al., 2015; De Los Reyes et al., 2019). Certainly, these discrepancies are due to multiple factors including differences in reporting patterns (Tein et al., 1994), social desirability bias among parents (De Los Reyes & Kazdin, 2005), and assumed measurement error (Moens et al., 2018). Additionally, discrepancies between reporters are largely a function of variation in perceptions and experiences between adolescents and their parents (De Los Reyes et al., 2013; Manongdo & Ramirez Garcia, 2007).

Increasingly, evidence demonstrates that parent-child discrepancies in reporting are meaningful in and of themselves (Abar et al., 2015). For example, when adolescents reported more parental rejection or negative reactions in response to adolescent anger compared to parent report, adolescents were more likely to have higher self-reported externalizing symptoms (Dimler et al., 2017; Jager et al., 2016). When parent and adolescent reports of parenting behaviors and parent-child interactions are widely discrepant, adolescents tend to engage in more delinquent behavior (De Los Reyes et al. 2010), exhibit more aggressive behavior (Dimler et al., 2017), and have greater internalizing problems (Jager et al., 2016). There are some instances where parent–child differences portend positive outcomes for the adolescent. For instance, discrepant perceptions of mother–adolescent relationships in which adolescents report more positivity than mothers are related to lower adolescent internalizing problems (Reidler & Swenson, 2012). Therefore, understanding discrepant perceptions and the direction of these discrepancies can provide meaningful information about adolescent mental health and wellbeing.

History of Discrepancy Analysis

Investigating differences between reporters has undergone many changes in the last five decades. When researchers began to examine reporter discrepancies, Hotelling’s (1940) t test procedure was traditionally employed, which tests of the differences between the multivariate means of different samples. Decades later, Meng et al. (1992) demonstrated the use of the Fisher’s r-to-z transformation to compare correlation coefficients of two different reports through a series of examples. The most commonly used methods typically used variations of the difference scores model (i.e., subtracting one reporter’s score from the other’s) to test reporter discrepancies (see Laird & De Los Reyes, 2013). However, the difference scores model is severely limited by the fact that the two reporters’ scores are measuring the same construct and therefore should be highly correlated (Edwards, 1994). Thus, comparing difference scores or correlations coefficients of the same survey instrument is not adequate for testing differences.

Alternatively, researchers recommend polynomial regression analyses as a more appropriate way of interpreting discrepancies between reporters, with the addition of interaction terms to measure the magnitude and direction of informant discrepancies (Edwards, 1994; De Los Reyes et al., 2019). Using this method, discrepancies are no longer conceptualized as simple “differences.” Rather, discrepancy is captured through the way the association of the father’s report of phenomena is (or is not) moderated by the child’s report of the same phenomena or vice versa. This method can more accurately make association claims due to its ability to add constraints to regression coefficients. The current study utilizes this method of polynomial regression analyses with interaction terms based off of the work from Laird & De Los Reyes (2013).

The Father–Adolescent Dyad

The father–adolescent dyad is a complex ecosystem, with each individual’s actions and perceptions playing both a direct and interactive role in the others’ experience of relationship quality and own mental health symptoms (Edwards et al., 2001). Just as a father’s parenting influences adolescent mental health, so does adolescent symptomatology impact father parenting and reactions (Markel & Wiener, 2014). Recognizing the bidirectional nature of family processes and adolescent development, the current study focused on fathering as the antecedent of outcomes in order to examine potential impacts of discrepant perceptions of fathering. Although there are well-established links of the effects of harsh parenting and conflict on adolescent outcomes, there is much less evidence on whether discrepancies between parent and adolescent reports of these constructs are linked to adolescent mental health problems. Most studies that examine discrepancies between reporters focus solely on mothers’ parenting behaviors (De Los Reyes et al., 2010; Laird & LaFleur, 2016). Among the ones that include both father and mother parenting, fathers comprise only 8–33% of the sample (Abar et al., 2015; Chen et al., 2017; Kim et al., 2016; Moens et al., 2018; Tein et al., 1994). Yet, fathers and their parenting behaviors are also central to the child and adolescent experience (Adamsons & Johnson, 2013; Meuwissen & Carlson, 2018). For instance, fathers who reported positive involvement behaviors, such as warmth/responsiveness and parental monitoring, had adolescents that reported significantly fewer internalizing and externalizing symptoms (Temmen & Crockett, 2021). Fathers also bring their own unique socialization goals and behaviors to the parenting process (Baker et al., 2018; Hill & Lynch, 1983). For example, consistent with gender intensification theory, parents spend more time with their same-sex child following puberty during which adolescents are being implicitly and explicitly socialized to sex-specific norms and goals (Crouter et al., 1995). Indeed, there is evidence of this influence of fathers on sons and mothers on daughters in myriad ways. Poor father-son relationship quality is associated with detrimental child outcomes, including increased school delinquency and substance use (Liu, 2004) and decreased self-esteem (Keizer et al., 2019). Similarly, mother–daughter conflict is associated with increased delinquent behaviors and cigarette use (Liu, 2004).

Current Study

Currently, there is limited research on father–adolescent discrepancies and this study seeks to expand on the limited evidence in the field. The current study aims were two-fold. The first was to examine the association between discrepancies in father–adolescent perceptions of harsh parenting and conflict with adolescent mental health symptomatology. Prior research shows that mother–child discrepancies in parenting are associated with poorer child health. Thus, the same was expected in this study of father–child discrepancies. It is hypothesized that greater father–adolescent discrepancies would be associated with greater internalizing and externalizing symptoms. The second was to examine whether and how these relations varied for male versus female adolescents. Gender intensification theory posits that socialization goals, behaviors, and impacts differ for fathers raising boys versus girls. As such, it was hypothesized that the impact of father–son discrepancies would significantly differ from that of father–daughter discrepancies. However, due to limited research on fathering, the current study did not make specific directional hypotheses. The current study was the first to examine discrepancies in perceptions of harsh parenting and conflict between fathers and their male children versus fathers and their female children.

Methods

Participants

The current study sample was drawn from a randomized control trial of a brief, family-focused substance abuse prevention program. The study included three cohorts of families recruited from three middle schools in the falls of 2015, 2016, and 2017. In order to be eligible, families must have had a 7th grade student and agreed to be randomized into intervention or control group; the adolescent was not in a self-contained class for emotional or cognitive impairment; and the participating caregiver(s) needed to live with the 7th grader for at least half of the month. In total, the study team contacted 2069 families with a child enrolled in the 7th grade; of those 60.3% (n = 1248) were eligible for the study. Among eligible families, 53.1% (n = 663) enrolled in the parent study. The current study used a subsample of families who had a male caregiver participating in the study, resulting in a final study sample of 326 male caregivers (M_age = 41.2 years, SD = 7.61 years, Range = 23–69 years) and their 7^th grade children (M_age = 12.0 years, SD = 0.50 years, Range = 11–13 years; 48.5% female). Almost all male caregiver participants identified as either a “father” (84.7%) or “stepfather” (15.3%). Thus, the male caregiver participants are referred to as “fathers” hereafter.

The study sample was racially and socioeconomically diverse. Fifty-three percent of fathers self-identified as Hispanic, 9.5% as mixed Hispanic, 16.3% as non-Hispanic White, 10.1% as non-Hispanic Black, 5.2% as non-Hispanic Native American/Indian, and 6.4% as another non-Hispanic ethnoracial group. The majority of families in the current study were from two-caregiver households (84.7%). Median annual household income was $40,000–50,000 (Range = Less than $5000-Over $100,000).

Procedure

Participants were recruited through three Title 1 middle schools in the greater Phoenix, Arizona area. Families with students enrolled in the seventh grade in the participating schools were sent letters inviting them to partake in the study. Caregivers provided written informed consent for their own participation and for their child’s participation. Adolescents provided written informed assent. Adolescents and caregivers independently completed a survey interview at the start of the academic year. All participants were given the option of completing study procedures in English or Spanish. More than half of participating fathers completed procedures in English (59.8%, n = 195). Almost all adolescents completed procedures fully in English (99.4%, n = 324). One adolescent completed procedures in Spanish and one in English and Spanish. Survey measures were originally sourced in English, translated into Spanish, and back-translated by program staff who are trained and employed specifically to assist in translation services. Measurement equivalence of the studied variables across language were established previously (Achenbach & Rescorla, 2015; Gonzales et al., 2000; Nair et al., 2009). All study procedures were approved by the University’s Institutional Review Board.

Measures

Harsh parenting

Perceptions of fathers’ harsh parenting was assessed using items originally developed for the Children’s Report of Parenting Behavior Inventory (Schaefer, 1965). Adolescents and fathers responded to 7 items that tapped into how often fathers displayed harsh, demeaning, or punitive parenting practices (e.g., “Your (father) screamed at you when you did something wrong,” “You got angry when (adolescent) was noisy around the house.”) These were measured on a 5-point Likert-type scale (1 = Almost never/Never to 5 = Almost always/Always). Cronbach alphas (α) for the adolescent report and the father report were 0.69 and 0.57, respectively. The items are highly skewed (skewness ranged 0.40–5.94 for father report; 0.39–3.38 for child report) which had negative effects on item alpha (Greer et al., 2006). We thus use confirmatory factor analysis (CFA) using maximum likelihood estimation with robust standard errors (i.e., MLR) to examine internal consistency of the construct, as Yang & Green (2011) showed that CFA is a better method of taking consideration of underlying item structure (e.g., linear vs. nonlinear; categorical vs. continuous, normal vs. skew) and testing unidimensionality. The one-dimensional CFA fit adequately for father report data [χ²(13) = 6.49, p = 0.93] and child report data [χ²(13) = 17.74, p = 0.17].

Father–adolescent conflict

Father and adolescent participants reported how often they experienced 13 common conflictual interactions with one another on a 5-point Likert-type scale (1 = Almost never/Never to 5 = Almost always/Always). Interactions ranged from frustrations (e.g., “You and your (father) got annoyed with each other”) to serious arguments (e.g., “You and (adolescent) had a serious argument or fight.”) between dyads (Ruiz et al., 1998). α’s for the adolescent report and the father report were 0.88 and 0.91, respectively.

Internalizing and externalizing problems

Adolescents completed the Youth Self Report (YSR) and fathers completed the Child Behavior Checklist (CBCL; Achenbach & Rescorla, 2001). T-scores from the internalizing problems and externalizing problems subscales were used as outcome variables for the current study. The YSR and CBCL have demonstrated strong reliability and validity across diverse samples (Achenbach & Rescorla, 2015). α’s for the adolescent report of internalizing and externalizing problems and the father report of internalizing and externalizing problems were 0.90, 0.87, 0.85, and 0.90, respectively.

Covariates

Covariates include father report of their preferred language (English vs. Spanish), household income, their age, and ethnicity. Given well-documented associations between parental depressive symptoms and child’s mental health symptoms (Grills & Ollendick, 2002), fathers report of the Center for Epidemiological Studies Depression scale (CESD; Radloff, 1977) was included as a covariate. α’s for the father report on CESD were 0.85.

Analysis Plan

First, the level of (dis)agreement was examined between adolescent and father for family process variables with bivariate correlations between adolescent-report of family process variables (harsh parenting and father–adolescent conflict) and father-report for the same variables for the full sample, for father–daughter dyads, and for father–son dyads. Correlations were investigated to see if they differed significantly between the father–daughter dyads and the father–son dyads using the Fisher’s r-to-z transformations.

Second, this study sought to examine how discrepancies between adolescent’s and father’s perceptions of family process were associated with (1) adolescent-reported internalizing symptoms, (2) father-reported internalizing symptoms, (3) adolescent-reported externalizing, and (4) father-reported externalizing symptoms in separate models. Non-linear regressions were conducted following previously documented methods of examining discrepancy effects via interaction terms (Laird & De Los Reyes, 2013):

$$Y=b_0+b_{1}A+b_{2}P+b_3{A}^2+b_4{P}^2+b_{5}AP+e$$ (1)

As discussed by Laird & De Los Reyes (2013), Eq. 1 estimates the linear effects of adolescent (b₁) and parent reports (b₂), the quadratic effects of adolescent (b₃) and parent reports (b₄), and the interaction between adolescent and parent reports (b₅). Father- and adolescent-reported hash parenting and conflict were mean-centered. The linear effects of adolescent and parent reports of family process (i.e., harsh parenting or father–adolescent conflict) represent linear associations between family process and adolescent symptomatology. The interaction terms tests whether the association between father-reported family process and mental health symptomatology is moderated by the adolescent’s perception of the same family process. For example, in the regression model with harsh parenting as the independent variable, a significant interaction term will denote that the association between father-reported harsh parenting and adolescent mental health symptoms varies as a function of adolescent perceptions of harsh parenting (Laird & De Los Reyes, 2013). This process was repeated for the model with father–adolescent conflict as the independent variable. The higher order terms were included in the model (i.e., quadratic terms) to account for complexity in the association between family process and mental health symptoms. Doing so ensures that the interaction term is not falsely representing a potential quadratic association (Laird & De Los Reyes, 2013). Including both the quadratic and interactive term also ensures that associations are not deemed nonsignificant when relations do in fact exist (Ganzach, 1997). The primary focus of the present study was to examine significant interaction terms, which represent significant effects of discrepant perceptions on adolescent outcomes. In order to understand the nature of significant interactions, post-hoc tests were conducted of the simple slopes for the effect of father report of parenting to adolescent symptoms at high (+1 SD) and low (−1 SD) levels of adolescent report of parenting. The covariates described above were included in each regression.

Lastly, this study examined if the above associations varied by adolescent sex. The same nonlinear regressions were run using a multiple group framework (Grouping: male adolescents and female adolescents). Each regression coefficient estimate in Eq. 1 was tested for the male and female adolescent samples were significantly different using a threshold of p < 0.05. Analyses were conducted using Mplus 8 (Muthén & Muthén, 1998–2017) using full information maximum likelihood (FIML) to handle missing data and maximum likelihood estimation with robust standard errors stand errors to handle non-normal variables. Of the variables used in the analyses, there was missing data percentages of 0–8% across all variables.

Results

Concordance between Father and Adolescent Reports

Table 1 presents the correlations between father report and adolescent report of the family process variables. Father and adolescent report of conflict was correlated for the full sample (r = 0.34), father–daughter dyads (r = 0.34), and for father–son dyads (r = 0.34). Concordance between reporters for harsh parenting were small (full sample r = 0.24, father–daughter r = 0.26, father–son r = 0.19). The Fisher’s z-tests show that the correlations for the father–daughter dyads of harsh parenting (z = −0.64, p = 0.26) and conflict (z = −0.04, p = 0.49) were not significantly different from the correlations for the father–son dyads.

Table 1.

Correlations between father and adolescent report of family process and parenting

	Father report of parent-adolescent conflict	Father report of harsh parenting
Full sample of adolescent-father dyads, N = 299
Adolescent report of parent-adolescent conflict	0.34**	0.22**
Adolescent report of harsh parenting	0.21**	0.24**
Female adolescent-father dyads, N = 143
Adolescent report of parent-adolescent conflict	0.34**	0.30**
Adolescent report of harsh parenting	0.18*	0.26**
Male adolescent-father dyads, N = 156
Adolescent report of parent-adolescent conflict	0.34**	0.13
Adolescent report of harsh parenting	0.20*	0.19*

^*p < 0.05,

^**p < 0.01,

^***p < 0.001

Harsh Parenting and Symptomatology

Internalizing symptoms

Table 2 presents the results of mental health symptoms regressed on father-reported harsh parenting, adolescent-reported harsh parenting, and their interaction. The linear estimate of adolescent-reported harsh parenting was significantly associated with adolescent-reported internalizing symptoms (B = 5.63, SE_B = 1.45, p < 0.001; β = 0.32); the interaction was not significant.

Table 2.

Adolescent and father reports of harsh parenting and their associations with mental health symptoms

	B	SE	95% CIa	p	B	SE	95% CIa	p
	Adolescent report of internalizing symptomsb				Father report of internalizing symptomsc
Harsh parenting
Adolescent report	5.63***	1.45	2.78, 8.48	<0.001	−0.57	1.21	−2.94, 1.81	0.64
Father report	−0.21	1.40	−2.95, 2.54	0.88	3.48**	1.19	1.15, 5.81	0.003
Adolescent report squared	−1.14	0.82	−2.74, 0.45	0.16	0.43	0.75	−1.03, 1.89	0.56
Father report squared	−1.66	1.44	−4.49, 1.16	0.25	−3.60*	1.63	−6.79, −0.41	0.03
Adolescent X Father report	1.15	1.67	−2.12, 4.42	0.49	0.79	1.47	−2.10, 3.67	0.59
	Adolescent report of externalizing symptomsb				Father report of externalizing symptomsc
Harsh parenting
Adolescent report	7.70***	1.18	5.40, 10.01	<0.001	1.22	1.14	−1.01, 3.45	0.28
Father report	0.49	1.34	−2.13, 3.11	0.71	6.09***	1.28	3.58, 8.60	<0.001
Adolescent report squared	−1.73**	0.61	−2.92, −0.55	0.004	−0.63	0.69	−1.98, 0.72	0.36
Father report squared	−0.55	1.51	−3.50, 2.41	0.72	−3.79*	1.51	−6.74, −0.83	0.01
Adolescent X Father report	1.34	1.46	−1.52, 4.20	0.36	4.16**	1.36	1.49, 6.83	0.002

^*p < 0.05,

^**p < 0.01,

^***p < 0 .0.001

^a95% of confidence interval for regression coefficients (b)

^bAdolescent report of symptoms via YSR

^cFather report of symptoms via CBCL

Linear father-reported harsh parenting and its quadratic term were significantly associated with father-reported internalizing problems (B = 3.48, SE_B = 1.19, p = 0.003; β = 0.18; B = −3.60, SE_B = 1.63, p = 0.03; β = 0.02). The significant quadratic effect created a non-monotonic line, meaning that the negative quadratic term made the linear effect become weaker when the level of harsh parenting increased. All subsequent significant quadratic effects mentioned below shared this type of relationship. The interaction was not significant.

Externalizing symptoms

Linear adolescent-reported harsh parenting (B = 7.70, SE_B = 1.18, p < 0.001; β = 0.47) and its quadratic term (B = −1.73, SE_B = 0.61, p = 0.004; β = −0.15) were significantly associated with adolescent-reported externalizing symptoms. The interaction between father- and adolescent-reported harsh parenting was not significant.

Comparatively, linear and quadratic father-reported harsh parenting (B = 6.09, SE_B = 1.28, p < 0.001; β = 0.31; B = −3.79, SE_B = 1.51, p = 0.01; β = −0.15) was significantly associated with father-reported externalizing problems. The interaction between father- and adolescent-reported harsh parenting was significant (B = 4.16, SE_B = 1.36, p = 0.002; β = 0.13). In other words, the association between father report of harsh parenting and father-reported externalizing problems was moderated by adolescent report of harsh parenting (Fig. 1). Post-hoc probing of the interaction revealed that among fathers whose adolescents reported higher-than-average levels of harsh parenting (i.e., +1 SD from mean adolescent-reported harsh parenting), father-reported harsh parenting was significantly associated with father-reported adolescent externalizing symptoms (linear: B = 8.56, SE_B = 1.55, p < 0.001; β = 0.43; quadratic: B = −3.79, SE_B = 1.51, p < 0.01; β = −0.15). For fathers whose adolescents reported lower-than-average levels of harsh parenting (i.e., −1 SD from mean adolescent-reported harsh parenting), father-reported harsh parenting was also significantly associated with externalizing symptoms (linear: B = 3.63, SE_B = 1.48, p < 0.01, β = 0.18; quadratic: B = −3.79, SE_B = 1.51, p < 0.01, β = −0.15). Both levels of harsh parenting had a significant curvilinear relationship, however, the linear effects for high harsh parenting was stronger.

Fig. 1.

Father report of adolescent externalizing problems as a function of father report of harsh parenting by high and low levels of adolescent report of harsh parenting

Father–Adolescent Conflict and Symptomatology

Internalizing symptoms

Table 3 presents the results of mental health symptoms regressed on father–adolescent conflict and their interaction. For adolescent report of internalizing symptoms, linear adolescent-reported conflict was the sole significant independent variable (B = 6.16, SE_B = 1.20, p < 0.001, β = 0.36). For father-reported symptoms, linear adolescent-reported conflict (B = 2.26, SE_B = 1.09, p = 0.04, β = 0.14), linear father-reported conflict (B = 4.32, SE_B = 1.15, p < 0.001; β = 0.27), and their interaction (B = −2.58, SE_B = 0.97, p = 0.008; β = −0.13) were significantly associated with father-reported adolescent internalizing problems. The coefficients of the quadratic terms were not significant. Post-hoc analyses revealed that for fathers whose adolescents reported lower-than-average levels of father–adolescent conflict (i.e., −1SD from the mean) and fathers whose adolescents reported higher-than-average levels of father–adolescent conflict (i.e., +1 SD from the mean), father-reported father–adolescent conflict was both significantly positively associated with father-reported adolescent internalizing symptoms (at −1SD: linear: B = 5.89, SE_B = 1.28, p < 0.001, β = 0.37; at +1 SD: linear: B = 2.75, SE_B = 1.30, p = 0.04, β = 0.17); however, as shown in Fig. 2, the linear effect was stronger for fathers whose adolescents reported lower-than-average levels of father–adolescent conflict.

Table 3.

Adolescent and father reports of parent-adolescent conflict and their associations with mental health symptoms

	B	SE	95% CIa	p	B	SE	95% CIa	p
	Adolescent report of internalizing symptomsb				Father report of internalizing symptomsc
Parent-adolescent conflict
Adolescent report	6.16***	1.20	3.80, 8.52	<0.001	2.26*	1.09	0.13, 4.39	0.04
Father report	−2.02	1.25	−4.47, 0.44	0.11	4.32***	1.15	2.07, 6.57	<0.001
Adolescent report squared	−0.57	1.12	−2.77, 1.64	0.62	0.82	0.88	−0.90, 2.54	0.35
Father report squared	−1.55	1.22	−3.94, 0.85	0.21	−1.22	0.92	−3.02, 0.57	0.18
Adolescent × Father report	1.64	1.47	−1.25, 4.53	0.27	−2.58**	0.97	−4.48, −0.68	0.008
	Adolescent report of externalizing symptomsb				Father report of externalizing symptomsc
Parent-adolescent conflict
Adolescent report	5.54***	1.11	3.37, 7.70	<0.001	2.53*	1.07	0.55, 4.50	0.01
Father report	1.17	1.28	−1.34, 3.67	0.36	7.96***	1.12	5.78, 10.15	<0.001
Adolescent report squared	−0.48	1.03	−2.50, 1.54	0.64	0.33	0.92	−1.49, 2.14	0.73
Father report squared	−1.97*	0.94	−3.81, −0.13	0.04	−2.07*	0.81	− 3.65, − 0.49	0.01
Adolescent × Father report	0.85	1.28	−1.67, 3.37	0.51	−1.16	1.02	−3.16, 0.84	0.26

^*p < 0.05,

^**p < 0.01,

^***p < 0.001

^a95% of confidence interval for regression coefficients (b)

^bAdolescent report of symptoms via YSR

^cFather report of symptoms via CBCL

Fig. 2.

Father report of adolescent internalizing problems as a function of father report of parent–adolescent conflict by high and low levels of adolescent report of parent–adolescent conflict

Externalizing symptoms

Linear adolescent-reported conflict (B = 5.54, SE_B = 1.11, p < 0.001, β = 0.35) and the quadratic of father-reported conflict (B = −1.97, SE_B = 0.94, p = 0.04, β = −0.14) were significantly associated with adolescent-reported externalizing symptoms. For father-reported externalizing problems, linear adolescent-reported conflict (B = 2.53, SE_B = 1.07, p = 0.01; β = 0.15), linear father-reported conflict (B = 7.96, SE_B = 1.12, p < 0.001; β = 0.50), and the quadratic of father-reported conflict (B = −2.07, SE_B = 0.81, p = .006; β = −0.14) were significant. The interaction term was not significant.¹

Moderation by Adolescent Sex

A series of four multi-group nonlinear regression models revealed that associations between adolescent report, father report, and their interactions and adolescent behavioral problems was not moderated by adolescent sex. This was true for both adolescent- and father-reported outcomes. All chi-square tests comparing the father–adolescent interaction estimates for female adolescents versus male adolescents revealed that coefficients did not vary by adolescent sex (Range p = 0.08–0.85; See Supplemental Tables 1–4). Thus, associations were examined within the single group framework described above.

Discussion

Previous discrepancy studies that examined adolescent outcomes have focused on the mother–adolescent dyad and therefore, there is a lack of research on the father–adolescent dyad. More research is required in this area to better comprehend how differing father and adolescent perceptions can predict adolescent outcomes. In order to bridge this gap, the current study examined (1) if discrepancies between father and adolescent perceptions of harsh parenting and family conflict were associated with adolescent mental health symptomatology and (2) if these associations varied for father–son dyads versus father–daughter dyads. Discrepancies were operationalized as interaction terms between father and adolescent reports of family phenomena. Of the four regression models examining associations between harsh parenting and adolescent symptomatology, only one showed a significant interaction between father and adolescent report of harsh parenting. The association between father-reported harsh parenting and father-reported externalizing symptoms was moderated by child-reported harsh parenting. Similarly, of the four regression models examining associations between family conflict and adolescent symptomatology, there was one significant interaction. The association between father-reported conflict and father-reported internalizing symptoms was moderated by child-reported conflict. In line with seminal studies on father’s harsh parenting, this study found that father-reported and adolescent-reported harsh parenting were only modestly correlated (Simons et al., 1991).

Overall, the current study affirmed previously found links between harsh parenting and adolescent mental health (Pinquart, 2017; Caples & Barrera, 2006). With some variations based on father versus adolescent report of symptoms, this study found evidence for main effects of father and adolescent report of harsh parenting on adolescent symptoms than their interaction. Only one model of adolescent symptomatology regressed on father- and adolescent-reported harsh parenting revealed a signification interaction between the two informant reports, suggesting that moderation by another informant’s report is limited to specific circumstances.

Interestingly, adolescent report of harsh parenting was significantly associated with adolescent-reported internalizing and externalizing symptoms whereas father report of harsh parenting (both linear and quadratic effects) was significantly associated with father-reported symptoms. Although the links between harsh parenting and symptoms are presumed to be robust, these links appear restricted within one informant’s report both in this study and others (e.g., Gonzales et al., 2017; Pinquart, 2017; Afifi et al., 2019). For example, cross-sessional associations between father’s harsh parenting and child outcomes featured a single informant on harsh parenting (child emotion dysregulation: Chang et al., 2003; child internalizing and externalizing problems: McKee et al., 2007). Even in exceptions when multiple informants reported on harsh parenting, the two informants’ perspectives were not included in the same model (e.g., mothers and daughters in Deardorff et al., 2013) or the two informants were both parents (Kelley et al., 2015; Li et al., 2021). More research is needed measuring both father and adolescent perceptions of harsh parenting and ameliorating these different perspectives. Though the current study focused on how differences in father–adolescent perceptions may produce greater risk for mental health symptoms, father–adolescent convergence on harsh parenting and other family processes may also provide meaningful insights into points for intervention.

Both father and adolescent report of father–adolescent conflict were significantly associated with adolescent symptoms in models regardless of who reported on adolescent symptoms. For example, when adolescent-reported externalizing symptoms was regressed on father–adolescent conflict, the linear effect of adolescent-reported conflict and quadratic effect of father-reported conflict were both significant predictors. Likewise, both father- and adolescent-reported conflict were significantly positively related to father-reported adolescent internalizing and externalizing symptoms. Unlike with associations between harsh parenting and symptoms, here, the links between father–adolescent conflict and symptoms transcend one informant’s perceptions. Similar to harsh parenting, studies measuring parent–adolescent conflict still largely favor using one informant over dual or multiple informants (e.g., Keizer et al., 2019; McCormick et al., 2016). However, measures of parent–adolescent conflict vary greatly from measures of parenting behaviors as they require parents and adolescents to reflect on both actors’ behaviors and interactions. Interestingly, this did not result in a greater correlation between father and adolescent report of conflict compared with harsh parenting. Instead, both father and adolescent reports of father–adolescent conflict may have unique, additive contributions to adolescent symptomatology.

The current study found only two instances where father-reported harsh parenting and father–adolescent conflict’s associations with adolescent symptoms were moderated by child-report of the same phenomena. In addition to the limited evidence of these informant discrepancies, interpretation is complex. For one, as father report of father–adolescent conflict increased, father report of adolescent internalizing symptoms increased at a greater rate for higher-than-average levels of adolescent-reported father–adolescent conflict, providing some support to the idea that discrepancies (i.e., high father-reported conflict in the face of low child-reported conflict) are a risk factor. However, as father report of harsh parenting increased, father report of adolescent externalizing symptoms also increased at a greater rate for higher-than-average levels of adolescent-reported harsh parenting. That is, as both fathers and adolescents reported higher-than-average levels of harsh parenting, fathers tended to report higher externalizing symptoms, suggesting that in this instance agreement on harsh parenting may be more a risk factor than disagreement. Recall that harsh parenting is not merely the absence of positive parenting (e.g., praise) but rather, the presence of negative and/or punitive verbal and physical behaviors such as fathers yelling at their children, calling them names, or punishing their children in front of others as a way to reprimand or discipline children (Gonzales et al., 2011; White et al., 2013). Studies have shown that harsh parenting emerges with well-intentioned parents as a response to neighborhood disadvantage and danger (White et al., 2019). Nonetheless, the detrimental effects of harsh parenting as undeniable (Germán et al., 2017). Indeed, the father–adolescent agreement plot illustrates that both informants’ agreement that father engagement in higher-than-average harsh parenting is associated with the highest levels of adolescent externalizing problems. Despite the study’s original aim to examine the role of discrepancies, findings point to the importance of also considering father–adolescent agreement on such highly negatively valenced constructs such as harsh parenting. Not only are discrepancies important, but depending on the parenting behavior under consideration, high levels of multi-informant agreement may also be meaningfully associated with adolescent outcomes.

Despite expectations, adolescent sex did not moderate the relation between discrepancies in harsh parenting and father–adolescent conflict and adolescent symptoms in any of the eight models examined. Gender intensification theory (Hill & Lynch, 1983; Crouter et al., 1995, 2007) and social cognitive theories of gender development (Bussey & Bandura, 1999), discuss differential socialization goals and parenting behaviors between mothers and fathers. Indeed, early and middle adolescence are periods of strong gender identification and socialization as a result of family contexts, ecological cues, biological potentialities, and time allocated to activities (Crouter et al. 2007). Early adolescence is also a time of increasing father involvement with sons’ activities (Dyer et al., 2014; Negraia et al., 2018). Thus, it was expected that the nature of father–adolescent discrepancies and their impact would vary for male versus female adolescents. However, adolescent sex did not moderate the relation between father–adolescent discrepancies and adolescent symptomatology. The current study may have been underpowered to detect such differences. It is also plausible that the processes of harsh parenting and father–adolescent conflict may not be sensitive to gender differences as much as other family phenomena. Gendered parenting is expressed more commonly in specific and implicit behaviors (Mesman & Groeneveld, 2018). Thus, more discrete examinations of parenting such as language differences (Adams et al., 1995; Leaper et al., 1998), specific disciplinary behaviors (Lytton & Romney, 1991), parental control (Endendijk et al., 2016), and specific play patterns (Morawska, 2020) may be better candidates for examining sex differences in father–adolescent discrepancies and their impact on adolescent mental health. Important to note, the current study did not feature any adolescents who identified as transgender or non-binary. Future studies of fathering’s distinct influence on the development of trans, non-binary, or gender fluid adolescents is needed. Further, it is yet to be understood how gender intensification can, if at all, be applied to this population.

Several limitations restrict study implications. First, the current study was cross-sectional design, therefore directionality of relations between harsh parenting, conflict, and informant discrepancies with symptomatology is not determined. In general, relations between harsh parenting and conflict with adolescent mental health are bidirectional. However, the current study examined harsh parenting and conflict as antecedents of mental health based off the chosen research question. Future studies that use a longitudinal design open a plethora of related and important research questions. In addition to examining questions of causality, longitudinal measurement of parenting, family process, and adolescent mental health variables will allow for the investigation of discrepancies over time. For example, are discrepancies/concordance between fathers and adolescents always stable over time? Is it stable for everyone or only for certain groups? Are there trajectories of families who become more discrepant over time and other trajectories who become more concordant over time? Understanding not just how discrepancies are related to adolescent mental health, but taking a step further back to understand the nature of discrepant versus concordant perceptions may prove to be extremely valuable in understanding family processes and associations with later outcomes. Second, the current study sample was underpowered to examine ethnoracial differences. Given the role of culture in socialization experiences, future studies should examine the potential moderating role of culture or ethnoracial status. Third, the current study is a secondary data analysis of data which commenced collection in 2015–2017. Fathers were asked about their adolescent’s sex and adolescents confirmed this information. As we have learned much about different sexual identities, future studies would be advised to ask adolescents about their own sex/gender identification. Regarding the analytic models, it must be noted that the significant interactions were only found in father-reported adolescent symptoms. Some may argue that for this early adolescent age group, adolescents may be the best reporters of their own behavioral health symptoms, particularly internalizing ones (e.g., Smith, 2007). More research is needed to clarify the mixed findings based on reporter of adolescent symptoms. Additionally, this study focused on the association between father–adolescent discrepancies and adolescent symptoms. Future studies may benefit from also examining the mother–adolescent dyad for comparison and contextualization. Finally, discrepancy analyses can be challenging to explain, future studies may benefit from trying methods such as the trifactor model (Bauer et al., 2013) which includes multiple report data in a covariate-informed factor model.

Conclusion

Research into fathering and its impact on adolescent development is a comparatively small but growing literature. This is the first study to examine discrepancies between father- and adolescent-reported family phenomena as discrepancies may be differentially related to adolescent mental health symptoms than single informant perceptions. Overall, the current study found significant main effects of (1) single informant perceptions of father–adolescent conflict on adolescent internalizing and externalizing symptoms and (2) single informant perceptions of fathers’ harsh parenting on adolescent internalizing and externalizing symptoms. There was small support of our hypothesis that discrepancies between father and adolescent perceptions of phenomena would be differentially associated with adolescent mental health symptoms. There was no evidence of these relations being moderated by adolescent sex. The mixed pattern of results emphasizes the need for more research investigating discrepancies using polynomial regressions as well as other methods. Findings from this study add to the growing literature on the importance of fathering and father–adolescent interactions during adolescent development.

Biographies

Sarah Hidalgo is a doctoral student in Developmental Psychology at Arizona State University. Her major research interests include prevention and intervention science, parenting, and mental health symptomology.

Joanna J. Kim is an Assistant Professor at Arizona State University. Her major research interests include parent engagement in prevention and treatment services as a way to combat behavioral health disparities.

Jenn-Yun Tein is a Research Professor at Arizona State University. Her major research interests include program evaluation, statistical analysis of prevention studies and their effects, specifically with minority populations.

Nancy A. Gonzales is the Executive Vice President and University Provost and a Foundation Professor at Arizona State University. Her major research interests focus on cultural and contextual influences on adolescent mental health.

Data Sharing and Declaration

The data that support the findings of this study are available from the corresponding author upon reasonable request.