Parenting Effects are in the Eye of the Beholder: Parent-Adolescent Differences in Perceptions Affects Adolescent Problem Behaviors

  • Laura M. Dimler
  • Misaki N. Natsuaki
  • Paul D. Hastings
  • Carolyn Zahn-Waxler
  • Bonnie Klimes-Dougan
Empirical Research

DOI: 10.1007/s10964-016-0612-1

Cite this article as:
Dimler, L.M., Natsuaki, M.N., Hastings, P.D. et al. J Youth Adolescence (2016). doi:10.1007/s10964-016-0612-1

Abstract

Although it is known that parents and adolescents hold different views regarding adolescent characteristics (e.g., inter-rater agreement on adolescent behaviors between parents and adolescents is low), we know little about parent-adolescent (dis)agreement in their perceptions of parenting. The current study used 220 parent-adolescent dyads (Mage = 13.3 years; 50.5 % female) to address this gap and examined how the discrepancy between parents’ and adolescents’ perceptions of the parent’s negative reactions toward an adolescent’s anger affects the adolescent’s problem behaviors. Results suggested the direction of the disagreement between the two parties is important: when adolescents viewed parenting more negatively than parents did, adolescents showed elevated levels of broadband externalizing behaviors and, specifically, aggressive behaviors. This finding suggests the importance of adolescents’ subjective views of how mothers and fathers react to them. The findings are discussed in terms of methodology in family studies and implications toward future research.

Keywords

Perceived parenting Discrepancy Anger Adolescent problem behaviors Inter-rater agreement 

Introduction

Prior research shows a robust association between negative parent-adolescent interactions and the exacerbation of adolescent problem behaviors (e.g., Hoeve et al. 2009; Hoskins 2014; Klahr et al. 2014). Specifically, parents’ mismanagement of and reaction to their adolescent’s challenging behaviors (e.g., anger) is strongly associated with the adolescent’s problem behaviors (Brand and Klimes-Dougan 2010; Crosswhite and Kerpelman 2009; Crum et al. 2015; Hoeve et al. 2009; Patterson et al. 1990). However, parents and adolescents perceive these types of interactions differently (Guion et al. 2009; Spilt et al. 2014), and little is known about whether the discrepancy between parents’ and adolescents’ perceptions of how parents manage adolescent anger is associated with problem behaviors. The present study aims to fill this gap in the literature.

Parent-Adolescent Perceptual Discrepancies in Negative Parent Reactions

While a previous meta-analysis (Hoeve et al. 2009) confirms that parents who react to their adolescent with negativity are more likely to have adolescents who engage in problem behaviors, the literature suggests that parents and adolescents often interpret parental behavior differently (Guion et al. 2009; Kendler et al. 1997; Tilton-Weaver et al. 2010) and that adolescents’ and parents’ perceptions of parenting uniquely predict adolescent problem behaviors (Mackenbach et al. 2014; White and Renk 2012). Typically, parents view themselves as more supportive, more involved, and less negative toward their adolescents compared to the adolescents’ reports of the same parental behaviors (De Los Reyes et al. 2013; Fleming et al. 2015; Hoeve et al. 2009; Korelitz and Garber 2016). While there does not seem to be a consensus as to whose report is a more accurate estimate of behavior (Korelitz and Garber 2016), recent empirical studies find that adolescent report but not parent report predicts adolescent problem behaviors, leading researchers to focus on how the discrepancies in the adolescents’ perception compared to the parents’ perceptions of the parents’ behaviors or parent-adolescent relationship quality affects the adolescent psychopathology (Fleming et al. 2015; Reidler and Swenson 2012).

Although discrepancies or disagreements between parent- and adolescent-report are often cast as problematic methodologically (De Los Reyes and Kazdin 2005), these discrepancies are ubiquitous (Feinberg et al. 2000) and allow researchers access to psychologically revealing aspects of the parent-adolescent relationship. For example, a dyad in concordance regarding low parental negativity may indicate that the parent and adolescent engage in good communication and problems within the dyadic relationship are handled appropriately (e.g., a parent calmly talks with the adolescent when the adolescent is angry). On the other hand, a discordant dyad, such that the parent perceives him/herself to respond more negatively to the adolescent’s anger than the child reports may also reflect the notion that the parent is a harsh self-critic (Holmbeck and O’Donnell 1991). In another type of discordant dyad, the adolescent may perceive the parent to respond more negatively to the adolescent’s anger than the parent reports. Possible reasons for this include parental underestimation of the adolescent’s anger leading him or her to dismiss it, for instance by telling the adolescent to “get over it”. However, the adolescent may interpret such a reaction as being treated as insignificant and overtly neglectful. Parents who disregard their adolescent’s vulnerable emotions (i.e., anger) may put their adolescent at risk for developing problematic behaviors (Klimes-Dougan et al. 2007). Taken together, a dyadic measure of perceptual discrepancy captures an aspect of the parent-adolescent relationship that is not reflected in an independent measure of adolescent- or parent-report of parenting.

Previous empirical work shows the importance of mismatching views in parent-adolescents dyads in predicting adolescent externalizing behaviors. For example, De Los Reyes et al. (2010) found that mother-adolescent discrepant reporting on parenting provides distinct information in predicting an adolescent’s problem behaviors, such that when the mother perceived herself as monitoring her adolescent more so than the adolescent perceived, the adolescent engaged in more externalizing behaviors over a period of two years. Additionally, Spilt et al. (2014) found that adolescents give unique information over and above both their mother and best friend in regards to reporting the perceived warmth and relationship quality within each dyad. When adolescents perceived their mothers to give more warmth than the mother perceived, the adolescents were more likely to have lower rates of depressive and aggressive symptoms two years later. Their results instead suggested that it is more important for adolescents to hold positive perceptions of those relationships, regardless of their agreement on the quality of the relationship.

The contribution of these studies notwithstanding, there are still a few missing pieces in the literature. First, how the discrepant views in parent-adolescent dyads on parental negative reactions toward adolescent anger is yet to be examined. Second, a relatively neglected area has been the discrepancy between father’s views on his parenting and his adolescent’s view of his parenting. It is important to examine fathers’ views on parenting because fathers may react to and socialize their child’s negative emotions differently from mothers (Brand and Klimes-Dougan 2010; Klimes-Dougan et al. 2007). Furthermore, a meta-analysis of 161 articles suggested that negativity and poor support from fathers yielded larger effect sizes on adolescent problem behaviors compared to negativity and poor support from mothers, yet fewer than 20 % of the included articles focused on or included paternal parenting behavior (Hoeve et al. 2009). In this article, we aimed to examine whether this discrepancy is particularly important in predicting problem behaviors, specifically focusing on both mother-adolescent and father-adolescent relationships. As De Los Reyes and Ohannessian (2016) call for a continued probe into the measurement of parents’ and adolescents’ views of family functioning and adolescent adjustment, we utilized various discrepancy indices to understand patters of discrepant perceptions between the parent and adolescent and how this affects adolescent problem behaviors. Furthermore, this research may inform future research and theories as to how to best assess discrepant reports.

Measuring Parent-Adolescent Discrepancy in Perceived Parenting: Methodological Considerations

Use of multiple sources of information (including parent- and adolescent-report) is a gold standard method in family studies and developmental science (Achenbach 2011). However, because parent- and adolescent-reports correlate only modestly (Achenbach et al. 1987; De Los Reyes and Kazdin 2005; Kraemer et al. 2003), it has been debated when and how researchers should integrate (or select) the information from multiple sources. For example, when researchers collect information through multiple sources (e.g., child-, parent-, and teacher-report), Kraemer et al. (2003) suggest researchers choose an informant based on conceptualizing the contexts and perspectives that influence expression of the characteristic of interest, and then identifying informants who represent those contexts and perspectives accurately. Most relevant to this study, De Los Reyes et al. (2015) propose that researchers and clinicians should actively use informant discrepancy scores (i.e., scores conveying the degree of mismatching views between informants) to gain rich information that may not be otherwise accessible through other forms of measurement.

Available evidence suggests that discrepancies between informants’ perceptions uniquely predict child outcomes above and beyond the main effects of each informant’s perception (De Los Reyes et al. 2015; De Los Reyes and Ohannessian 2016). For instance, a discrepancy in perceptions of child symptomology between mother and father (Treutler and Epkins 2003), parent and psychiatrist (Clarke-Stewart et al. 2003), parent and teacher (De Los Reyes et al. 2009), and adolescent and friend (Spilt et al. 2014) are known to hold important information for predicting child outcomes. The discrepancy in each of these dyadic relationships seems to represent a unique construct that cannot be captured by simply aggregating multiple informants into one latent construct or using one informant instead of another (i.e., using mother report instead of child report).

Although previous studies may have taken into account multiple informants as a method of data collection, few have taken advantage of the fact that discrepancies between informants (e.g., [mis]matching views between the parent and adolescent regarding parenting) may have substantive implications for understanding the development of adolescent psychopathology (De Los Reyes et al. 2011). One reason for the scarcity of this kind of study is the confusion in computation and interpretation of these multi-informant discrepancies. Recently, Spilt et al. (2014) provided a guideline for how to calculate the discrepancy between perceptions held by multiple informants, and we applied the guideline to this study.

The first form of a measure of discrepancy is the raw difference score between two informants (e.g., subtracting the child’s report on negative parenting toward child anger from that of parent’s [Parent’s report on parenting (P)—Adolescent’s report on parenting (A)]). Because it retains the directionality (positive or negative sign), it provides information regarding who perceives more or less favorably than whom. Prior empirical evidence suggests that this type of discrepancy index is helpful in predicting child problem behaviors (Reidler and Swenson 2012). Specifically, Reidler and Swenson (2012) and Spilt et al. (2014) indicate that when a child perceives the relationship with his or her mother as less positive than the mother reports, the child engages in more externalizing behaviors compared to peers.

The second, and simplest, form of a discrepancy index is the absolute difference between two informants (e.g., |P–A|). This absolute score does not convey the information regarding characteristics of each individual in the dyad, but is informative when indexing the dyadic characteristics. Discrepancies that are absolute, but not directional, may indicate inconsistent behaviors across contexts or miscommunication between the two informants. Evidence has shown that the absolute discrepancy score plays a unique role in predicting child problem behaviors. For example, Reidler and Swenson (2012) indicated that there was a positive correlation between this absolute discrepancy index (i.e., there is a disagreement between an adolescent and a parent) on the negative relationship quality and adolescent externalizing behaviors.

The third measure of discrepancy is the interaction term between the parent and child report of parenting (e.g., P × A), which allows Person A’s (e.g., the parent) perception of parenting to potentially moderate the association between Person B’s (e.g., the adolescent) perception of parenting and their problem behaviors. This discrepancy index may be able to indicate how each individual’s (i.e., the parent and the adolescent) subjective experience strengthens or weakens the adolescent’s problem behaviors. An advantage of this P × A measure is that it retains the original unit of scale. For instance, unlike the |P–A| or the P–A indices, it can distinguish dyads whose amount of disagreement between parents’ and adolescents’ views is the same but the nature of disagreement is different. For instance, in Dyad A, a parent and adolescent score 5 and 3 on a 10-point scale of the perceived negativity in parenting, respectively (therefore, |P–A| and P–A are 2), indicating that both members of the dyad perceive parenting to be less negative. In contrast, in Dyad B, a parent and child score 10 and 8 (|P–A| and P–A are also 2), indicating that both members of the dyad perceive parenting to be very negative. Despite the comparable amount of disagreement in perceptions, the adolescent in Dyad B would be expected to develop severe behavioral problems than the adolescent in Dyad A because the disagreement in Dyad A is on the lower end in the spectrum of perceived negativity in parenting. Therefore, a moderation analysis allows for a more nuanced understanding of which adolescents are at risk for (or protected against) problem behaviors depending upon their and their parents’ subjective experiences. In fact, De Los Reyes et al. (2013) argued that using an interaction as a type of discrepancy index may provide the most informative, intuitive, and valid information compared to the other types of traditional discrepancy indices because the interaction is able to explain the variance in the outcome variables beyond what main effects or traditional discrepancy indices are able to explain.

In this study, we examined all three indices of discrepancy scores and separately analyzed both the mother-adolescent and father-adolescent dyads. Therefore, an ancillary aim of this study was to assess the types of discrepancy indices and to understand how each predicts the adolescent problem behaviors.

The Present Study

Although there is a multitude of studies that assess the effect of parenting on adolescents’ problem behaviors that use multiple informants (e.g., teacher-report, self-report, and caregiver-report; De Los Reyes and Kazdin 2005; De Los Reyes et al. 2011; Guion et al. 2009; Lavigne et al. 2014), little is known about whether and how (dis)agreement between parents’ and adolescents’ views are associated with adolescent problem behaviors. Therefore, the purpose of this study was to assess how the discrepancy between a parent’s and adolescent’s perceptions of a parent’s negative reactions toward an adolescent’s anger affects the adolescent’s problem behaviors. In addition, our subsidiary aim was to investigate the different types of discrepancy indices that are informative when predicting problem behaviors. This study is among the first to examine the importance of discrepancies between the perceptions of parenting in both types of parent-adolescent dyads (i.e., mother-adolescent dyads and father-adolescent dyads) on adolescent problem behaviors.

Based on previous research, and specific suggestions to understand the mother-adolescent and father-adolescent relationships’ effect on adolescent problem behaviors, we formulated a series of hypotheses. Based on findings from Hoeve et al.’s (2009) meta-analysis indicated that both parents’ and adolescents’ views on the parents’ negative aspects of parenting holds unique predictability toward child development, we hypothesized that (1) parents’ reports of their reactions to adolescent anger would be positively correlated with adolescents’ problem behaviors, such that parents who perceive themselves to respond more negatively to their adolescent’s anger would have adolescents who engage in more problem behaviors (i.e., a main effect of parent-report of parenting); and (2) adolescents’ reports of parents’ reactions to adolescents’ anger would be positively correlated with adolescents’ problem behaviors, such that adolescents who perceive their parents to respond more negatively to their anger would engage in more problem behaviors (i.e., a main effect of adolescent-report of parenting).

The current study expanded upon prior parent-adolescent discrepancy research by analyzing three distinct discrepancy scores and by separating mother-adolescent and father-adolescent dyads. Based on prior research stating that adolescents and parents view the same scenario differently and that adolescents’ subjective views of these scenarios may increase problem behaviors (Reidler and Swenson 2012), we hypothesized that (3) the raw discrepancy score (P–A) would be negatively associated with adolescent problem behaviors such that dyads with negative scores (i.e., adolescent’s greater perception of negativity in parent’s reaction relative to the parent’s report) would have adolescents with higher levels of problem behaviors than and dyads with positive scores (i.e., parent’s greater perception of negativity in own reaction relative to the child’s report); and (4) the absolute discrepancy score (|P–A|) would be positively associated with the adolescent’s problem behaviors, such that children in dyads who disagree more in terms of how well the parent manage child anger would show more problem behaviors. Finally, following recent suggestions made by De Los Reyes and colleagues (2013, 2015) to assess interaction terms as a potentially more valid discrepancy index than other discrepancy indices, we hypothesized that (5) the association between the parent’s perception of their negative reactions toward their adolescent’s anger and their adolescent’s problem behaviors would be moderated by the adolescent’s perception. This moderation would indicate that the effect of parents’ perceived negativity in parenting would be positively, yet less strongly, correlated with problem behaviors in adolescents who perceive high negativity in parenting than among adolescents who do not perceive parenting to be highly negative (P × A).

Methods

Sample

The current study is based on the data from a sample of 220 adolescents (50.5 % male) and their mothers and fathers who participated in the Adolescent Emotion Study (AES). The AES is a multi-method, longitudinal investigation that investigated the role of emotion in psychopathology in adolescence (Klimes-Dougan et al. 2001). Age of participating adolescents ranged from 11.05 to 16.96 years (M = 13.3, SD = 1.57). The adolescent sample was 70.4 % Caucasian, 16.2 % African-American, 1.9 % Hispanic, 2.8 % Asian-American, and 8.8 % mixed race or other, and included families from a wide range of socioeconomic groups (16.4 % reported annual household incomes of $0–40,000, 35 % reported $41,000–80,000, and 45.5 % reported earning over $81,000). Mothers (birth mother, adopted mother, or stepmother) were required to be the consistent, primary caretaker of the child (since the child was 6 years of age) in order to be considered for this study. The majority of adolescents (76.4 %) lived in two-parent (biological, adopted, or step-parents) families, but other types of family structure (e.g., single-parent household run by either mother or father and grandparent households) were also included in AES (for more information regarding the participant criteria and study procedures, please see Klimes-Dougan et al. 2001).

Of the initial 220 families who participated in the study, 140 father-adolescent dyads and 198 mother-adolescent dyads had complete data for the predictor variables for the current investigation. The analytical sample of the present study is based on these pairs with complete data. Comparison between the 198 mother-adolescent dyads in the analytical sample and the 22 mother-adolescent dyads without complete data (e.g., mother responses were missing on one or more measures) revealed that there were no significant differences in terms of demographics (i.e., income, age of the adolescent, and sex of the adolescent) or outcome variables. Comparison between the 140 father-adolescent dyads with complete data and the 80 father-adolescent dyads without complete data (e.g., father responses were missing on one or more measures or fathers were absent in the family) revealed that there was no significant difference in terms of the age or sex of the adolescent or outcome variables, but there was a significant difference in income; fathers were more likely to have missing data or be absent in the family if their income was under $60,000.

Procedure

Participants were recruited through announcements (e.g., newspapers, flyers) in the Washington, DC metropolitan community. Eligibility for participation was based on the youth’s age (between the ages of 11 and 16 years at the time of the initial visit) and competency to complete procedures (e.g., lack of debilitating language, cognitive, physical, or psychotic impairment). Additionally, because the overarching goal of the investigation was to investigate adolescent psychopathology, the recruitment strategy was designed to cover a wide range of emotional and behavioral problems. At the initial telephone interview for the recruitment, abbreviated versions of the Youth Self-Report (Achenbach 1991b) and the Child Behavior Checklist (Achenbach 1991a) were administered to the youth and the mother for screening purposes. This procedure yielded a sample in which approximately equal numbers of male and female adolescents exhibiting a range of subclinical to clinical emotional and behavioral problems were included. For a more detailed description of the sample and recruitment procedure, see Klimes-Dougan and colleagues (Klimes-Dougan et al. 2001, 2007).

Measures

Parents’ response to adolescent anger

Parents’ response to adolescent anger was measured with the anger subscale of the Emotion Socialization Measure (Klimes-Dougan et al. 2001, 2007) by using adolescent and parent reports. This 15-item scale is designed to measure a parent’s response to their adolescent’s anger. For the child report, adolescents read the prompt “When you have been angry, what did your mother/father do?” Using a five-point scale (1 = not at all typical, 5 = very typical), adolescents rated 5 types of parental behaviors in reaction to adolescents’ anger (reward, punish, magnify, neglect, and override). For the purposes of this study, the three types of parental reactions toward an adolescent’s anger (punish, magnify, and neglect) were included to create a score of negative reactions toward child’s anger. Cronbach’s alphas for the adolescent report of maternal and paternal responses to anger were α = 0.801 and 0.825, respectively. The parents used identical items to rate their reactions to the adolescent’s anger. Cronbach’s alphas for mother-report and father-report of own behaviors were α = 0.748 and 0.795, respectively, signifying acceptable reliability for all informants. This measure has demonstrated acceptable levels of reliability and validity in previous literature (Garside and Klimes-Dougan 2002; O’Neal and Magai 2005). Higher scores on this measure indicated more parental negative reactions to an adolescent’s anger.

Externalizing behaviors

The Youth Self-Report (YSR) is a 112-item, well-validated self-report measure of adolescent psychopathology in ages 11–18 (Achenbach 1991b). This adolescent-report measure consists of broadband (i.e., overall internalizing and overall externalizing behaviors) and narrowband scales (i.e., delinquent behaviors), with higher scores indicating the endorsement of more problem behaviors. The current study used the broadband externalizing t-scores which assesses behaviors such as rule-breaking, aggression, and disobedience. Some sample items are, “I threaten to hurt people,” “I disobey my parents”, and “I break rules at home, school, or elsewhere”. The reliability in this sample was acceptable, with a Cronbach’s alpha of α = 0.730.

Aggressive behaviors

The Buss-Perry Aggression Questionnaire (AGQ) is a well-validated self-report measure of aggression (Buss and Perry 1992). The scale consists of 29 items in which adolescents rated how characteristic the items were of their aggressive behaviors using a 5-point Likert scale (1 = extremely uncharacteristic of me, 5 = extremely characteristic of me). The sample item included “Once in a while I can’t control the urge to strike another person.” Higher scores on this measure indicated more aggressive behaviors. Compared to the Youth Self-Report, this measure allows for more in depth analysis into aggression. The reliability in this sample was acceptable, with a Cronbach’s alpha of α = 0.860.

Calculation of Parent-Adolescent Discrepancy Scores on Parenting

Following the guideline developed by Spilt et al. (2014), we computed three ways to capture the perceptual discrepancy between the adolescent and the parent in parenting (mother and father separately).

First, we subtracted the adolescent’s score on the Perceived Response to Child’s Anger measure from each parent’s score (the Parent [P]—the Adolescent [A]). This discrepancy score is sensitive to the direction of the disagreement as well as the amount of disagreement at the mean level, such that negative scores indicated the adolescent rated his/her parent to react more negatively to their adolescent’s anger than what parents self-reported. Positive scores indicated the adolescent perceived his/her parent to react more positively to their anger than the parent perceived. Second, we created an absolute difference term (|P–A|) to assess any differences in the parent-adolescent dyads. The higher the score is, the more disagreement in the parent-adolescent dyads with respect to how negatively the parent reacts to adolescent anger. Finally, an interaction term was computed by multiplying the parent’s perception score and the adolescent’s perception score (P × A). This interactive index provides not only the amount of disagreement in views but also directionality. We expected that if each parent and adolescent perception of parental behavior has unique information and they jointly add unique variance to explain problem behaviors, the aforementioned three indices of discrepancy scores would show similar patterns of results.

Covariates

Three covariates were included in the subsequent models: age, sex, and family income. Adolescent age was considered as a covariate because problem behaviors tend to show an increase with age during adolescence (Moffitt 1993; Squeglia et al. 2014). Adolescent sex was considered as a covariate because sex differences in problem behaviors during adolescence are well documented, with males showing higher problem problems than females (Crick and Zahn-Waxler 2003; Ge et al. 2006; Negriff and Susman 2011; Weinstein and Dannon 2015). Family income was also considered as a covariate because previous literature suggests that adolescents from a low-income family exhibit more problem behaviors than adolescents in middle- or high-income families (Defoe et al. 2013; McLaughlin et al. 2012).

Results

Analytical Strategy

First, we computed descriptive statistics of the study variables. Second, we used a series of hierarchical linear regression analysis to test the aforementioned hypotheses. This type of analysis allows us to observe if the discrepancy indices predict problem behaviors over and above the main effects of parent and adolescent perceptions. These analyses were run separately for each discrepancy score (P–A, |P–A|, P × A) so as to avoid multicollinearity. We also ran separate models for maternal and paternal parenting. For all analyses, covariates and main effects of parent-report of parenting and adolescent-report of parenting were included. All analyses were performed using SPSS 22.0.

Descriptive statistics

The means and standard deviations for the discrepancy scores are presented in Table 1, and bivariate correlations in Table 2. Bivariate correlational analyses indicated mothers’ report on parenting and adolescent report of parenting were not significantly correlated. However, father and adolescent reports of parent reactions were significantly correlated (r = 0.172, p < 0.05). Child sex was not significantly correlated with any predictor or outcome variables. Household income was significantly associated only with father’s report of parenting behaviors (r = 0.226, p < 0.01). Age was significantly correlated with self-reported broadband externalizing behaviors (r = 0.154, p < 0.05).
Table 1

Means and standard deviations of parenting variables

 

Mother

Father

 

M

SD

M

SD

Parent: Negative parental reactions (P)

2.021

0.498

2.010

0.493

Adolescent: Negative parental reactions (A)

1.970

0.732

2.050

0.790

Raw discrepancy (P–A)

0.050

0.847

−0.038

0.858

Absolute discrepancy (|P–A|)

0.683

0.502

0.697

0.498

Note: P–A indicates the direction of disagreement discrepancy score between parent and adolescent. |P–A| indicates the absolute discrepancy score between parent and adolescent

Table 2

Bivariate correlations among the study variables

  

1

2

3

4

5

6

7

1

P

0.091

0.510+

0.204+

0.109

0.167*

0.128

2

A

0.172*

−0.810+

0.245+

−0.034

0.388+

0.460+

3

P–A

0.418+

−0.823+

−0.091

0.099

−0.238+

−0.320+

4

|P–A|

0.083

0.343+

−0.269+

−0.019

0.048

0.015

5

P × A

0.056

0.053

−0.027

−0.001

−0.001

−0.007

6

YSR

0.211*

0.277+

−0.132

0.038

−0.084

0.626+

7

AGQ

0.220+

0.423+

−0.275+

0.146

−0.184

0.053

Note: Variables relating to the mother are above the diagonal, while variables relating to the father are below the diagonal

P Parent report of parenting, A Adolescent report of parenting, PA Direction of disagreement discrepancy score, |PA| Absolute value of disagreement discrepancy score, P × A Interaction index, YSR Youth Self-Report, AGQ Buss-Perry Aggression Questionnaire

*p ≤ 0.05; +p ≤ 0.01

Mother’s perception of her response to her adolescent’s anger was significantly correlated with broadband externalizing behaviors (r = 0.167, p < 0.05), but not with aggressive behaviors (r = 0.128, ns). The child’s perception of the mother’s parenting was significantly correlated with broadband externalizing behaviors (r = 0.388, p < 0.01) and aggressive behaviors (r = 0.460, p < 0.01). The direction of disagreement discrepancy score (P–A) was significantly negatively correlated with both broadband externalizing behaviors (r = −0.238, p < 0.01) and aggressive behaviors (r = −0.320, p < 0.05), indicating that the adolescent will report higher levels of problem behaviors when the adolescent perceives more negativity in the parenting (with reference to his/her mothers’ perception).

Father’s perception of his response to his adolescent’s anger was significantly correlated with externalizing behaviors and aggressive behaviors (r = 0.211, p < 0.05 and 0.220, p < 0.01, respectively). The child’s perception of the father’s parenting was significantly correlated with broadband externalizing behaviors (r = 0.277, p < 0.01) and aggressive behaviors (r = 0.423, p < 0.01). The direction of disagreement discrepancy score (P–A) was negatively associated with aggressive behaviors (r = −0.275, p < 0.01), yet the absolute discrepancy score (|P–A|) was not (r = 0.146, ns), suggesting that the direction of the discrepant perceptions between a father and adolescent provides important information.

Hierarchical Regression Analyses

Next, we tested the associations between a set of discrepancy variables and externalizing behaviors through a series of hierarchical linear regression analyses controlling for the adolescent’s sex, socioeconomic status, and age. Each discrepancy score was entered in a separate regression models. In the following sections, results for maternal (Table 3) and paternal (Table 4) parenting are discussed, respectively.
Table 3

Mother-adolescent discrepancy in perceptions of parenting and problem behaviors

Variable

Model 1

Model 2

Model 3

Model 4

Model 5

 

b

SE

b

SE

b

SE

b

SE

b

SE

Broadband externalizing behaviors

 Constant

38.20

6.16

39.14

5.75

37.64

6.05

37.39

6.25

39.93

5.73

 Age

0.68

0.43

0.59

0.40

0.75

0.42

0.74

0.43

0.52

0.40

 Sex

1.87

1.33

2.30

1.24

1.72

1.30

1.59

1.35

2.54*

1.24

 Income

0.50

0.43

0.44

0.40

0.54

0.42

0.54

0.44

0.40

0.40

 1. P

2.97*

1.34

      

2.50

1.25

 2. A

  

5.08**

0.85

    

4.97**

0.85

 3. P–A

    

−2.70**

0.77

    

 4. |P–A|

      

0.57

1.34

  

 5. P × A

        

0.32

1.77

  R2

0.06

0.18

0.09

0.01

0.20

Aggressive behaviors

 Constant

2.71

0.39

2.81

0.35

2.71

0.37

2.69

0.392

2.84

0.35

 Age

0.00

0.03

−0.01

0.02

0.00

0.03

0.00

0.027

−0.01

0.02

 Sex

−0.14

0.08

−0.10

0.08

−0.14

0.08

−0.15

0.085

−0.08

0.08

 Income

0.00

0.03

−0.01

0.02

0.00

0.03

0.00

0.027

−0.01

0.02

 1. P

0.15

0.09

      

0.11

0.08

 2. A

  

0.38**

0.05

    

0.37**

0.05

 3. P–A

    

−0.23**

0.05

    

 4. |P–A|

      

0.03

0.08

  

 5. P × A

        

0.04

0.11

  R2

0.03

0.23

0.12

0.02

0.24

Note: Outcome measures were adolescent report. The variables for P × A were centered at their means

P Mother report of parenting, A Adolescent report of parenting, PA Direction of disagreement discrepancy score, |PA| Absolute value of disagreement discrepancy score, P × A Interaction index

*p ≤ 0.05; **p ≤ 0.01

Table 4

Father-adolescent discrepancy in perceptions of parenting and problem behaviors

Variable

Model 1

Model 2

Model 3

Model 4

Model 5

 

b

SE

b

SE

b

SE

b

SE

b

SE

Broadband externalizing behaviors

 Constant

38.74

6.54

37.63

6.29

38.66

6.60

38.66

6.60

38.66

6.60

 Age

0.77

0.46

0.79

0.44

0.77

0.46

0.77

0.46

0.77

0.46

 Sex

1.22

1.47

1.23

1.41

1.23

1.48

1.23

1.48

1.23

1.48

 Income

0.38

0.55

0.52

0.51

0.38

0.55

0.38

0.55

0.38

0.55

 1. P

3.48*

1.52

      

3.26*

1.50

 2. A

  

3.41**

0.89

    

3.22**

0.92

 3. P–A

    

−1.65

0.86

    

 4. |P–A|

      

0.62

1.51

  

 5. P × A

        

−2.81

1.51

  R2

0.06

0.11

0.05

0.03

0.15

Aggressive behaviors

 Constant

2.83

0.41

2.622

0.37

2.616

0.41

2.69

0.42

2.71

0.38

 Age

0.00

0.03

0.01

0.03

0.00

0.03

−0.01

0.03

0.00

0.03

 Sex

−0.18

0.09

−0.13

0.08

−0.16

0.09

−0.20*

0.09

−0.10

0.09

 Income

−0.01

0.04

−0.01

0.03

0.02

0.03

0.02

0.04

−0.01

0.03

 1. P

0.25*

0.10

      

0.22*

0.09

 2. A

  

0.31**

0.05

    

0.31**

0.06

 3. P–A

    

−0.18**

0.05

    

 4. |P–A|

      

0.03

0.08

  

 5. P × A

        

−0.22*

0.09

  R2

0.05

0.18

0.07

0.05

0.22

Note: Outcome measures were adolescent report. The variables for P × A were centered at their means

P Father report of parenting, A Adolescent report of parenting, PA Direction of disagreement discrepancy score, |PA| Absolute value of disagreement discrepancy score, P × A Interaction index

*p ≤ 0.01; **p ≤ 0.001

Mother

As expected, the main effect of the adolescent’s perception of mother’s responses was associated with broadband externalizing behaviors (Table 3, Model 2, b= 0.567, p < 0.001). The direction of disagreement discrepancy score (P–A; Model 3) was associated with an adolescent’s broadband externalizing behaviors (b = −2.698, p < 0.001) and aggressive behaviors (b = −0.226, p < 0.001), suggesting that more negativity perceived by the adolescent, compared to the perception of the parent (resulting in a negative discrepancy score), was related to concurrent problem behaviors. Interestingly, other indices of mother-adolescent perception discrepancy were not significantly associated with problem behaviors.

Father

Consistent with the mother model, the main effect of adolescent’s perception of father’s responses (Table 4; Model 2) associated adolescent externalizing behaviors (b = 0.425, p < 0.01) and aggressive behaviors (b = 0.249, p < 0.05). Most notably, the interaction term parent×child (P × A; Model 5) was associated with an adolescent’s aggressive behaviors (b = −0.216, p < 0.05). Figure 1 presents a graphical presentation of the interaction. The adolescent’s perception of the father’s reactions was used as a moderator of the association between the father’s perception of his parenting and the adolescent’s aggressive behaviors. To further probe this interaction, we calculated the simple slope for three dyads: a father-adolescent dyad whose child perceived parenting to be low (one standard deviation below the mean), average (the statistical mean), and high on negativity (one standard deviation above the mean). When evaluating the effect of the father’s perception of negative responses on the adolescent’s aggressive behaviors at different values of the adolescent’s perception of negative parental responses, the analyses indicated that as the adolescent’s perception of father negativity decreased, the relationship between the father’s perception and adolescent aggressive behaviors became more positive. Specifically, a simple slopes analysis indicated that when the adolescent perceived the father’s negativity to be low (b = 0.39, p < 0.001) and average (b = 0.22, p < 0.001), the adolescent engaged in significantly more aggressive behaviors.
Fig. 1

Adolescent aggressive behaviors from father and adolescent perception of negative responses to the adolescent’s anger (±1SD). AGQ Buss-Perry Aggression Questionnaire. *p<.001

Sensitivity Analysis

A sensitivity analysis was performed to ensure the robustness of the discrepancy indices. The sample for the sensitivity analysis was restricted to adolescents who had both mother- and father-report (n = 136). For both externalizing behaviors and aggressive behaviors, the sensitivity analysis replicated the same pattern of results in the full sample and restricted sample for both mothers and fathers in all models (main effects, direction of disagreement index, absolute value of disagreement index, and the interaction index).

Discussion

The use of multiple sources of information (e.g., self, parent, teacher, observation, official records) to increase the validity of psychological constructs has been recommended in developmental science (e.g., De Los Reyes and Kazdin 2005). For example, adolescents tend to report parents as more negative and less warm than the parents’ self-reports claim (De Los Reyes et al. 2013; Fleming et al. 2015), therefore, obtaining reports from multiple sources allows researchers to assess the reasons for and the consequences of the discrepancies. In most cases, researchers either use one informant or combine multiple informants’ responses into one aggregate (either observed or latent) variable. However, another fruitful avenue for research is to assess the discrepancy between informants, which is known to have a unique predictive effect on child psychopathology (Clarke-Stewart et al. 2003; De Los Reyes et al. 2009; Spilt et al. 2014). Moreover, using multiple sources of information allows researchers to understand which report has the most validity or if there is a specific discrepancy index that provides the most validity and clarity into adolescent developmental psychopathology. In this study, we sought to assess how the discrepancy between a parent’s and adolescent’s perceptions of parenting affects the adolescent’s problem behaviors, while specifically examining the mother-adolescent and father-adolescent dyads through a multitude of discrepancy indices.

Informant Discrepancies

As expected, both child and parent report of negative parental reactions to the adolescent’s anger was associated with externalizing behaviors and aggressive behaviors in the adolescent (with one exception of a nonsignificant effect of maternal report of parenting on aggressive behaviors). Not only does this give more credence to the body of literature stating parents’ mismanagement of child anger is associated with a host of psychopathological symptoms in adolescents (e.g., Padilla-Walker 2008; Snyder et al. 2003), but it also highlights how each individual’s perspective within the family unit relates to adolescent problem behaviors.

More pertinent to the aim of this study, the results from the analyses testing the discrepancy indices demonstrated that the discrepancy in how adolescents and parents view parental management of adolescent anger matters for problem behaviors, especially when adolescents viewed parenting more negatively than parents did. In particular, the direction of disagreement (P–A) score was associated with adolescent externalizing behaviors, but the absolute disagreement (|P–A|) score was not. These results indicate that it is not the mere presence of discrepancy within dyads which matters, but rather it is both directionality of the parent-child discrepancy and amount of the parent-child discrepancy that are important in predicting externalizing behaviors. As expected, it was the adolescent perception (in comparison to parent perception) that carried the weight; findings on the direction of disagreement index (P–A) indicated that when adolescents viewed parenting more negatively than the parents did, adolescents showed elevated levels of broadband externalizing behaviors and aggressive behaviors. In other words, this directional index finding suggests the importance of adolescents’ subjective views of how mothers and fathers react to them.

An important question remains in understanding what the directional discrepancy score (P–A) means theoretically. Given that our finding supports the importance of adolescents’ subjectivity, one possible interpretation is that the directional discrepancy index, when the score is negative, may be indirectly capturing the adolescents’ feelings about the gap in parent-child views on parenting. For example, the adolescent may be feeling rejected, neglected, or simply unheard when they perceive that parents treat their anger lightly (Garside and Klimes-Dougan 2002). Such sentiments would not be captured by the absolute discrepancy index. Future effort is needed to unravel the meaning of the directional discrepancy index.

The results from the interaction index (P × A) partially confirm the hypothesis that the adolescent’s perception of their parent’s negative reactions toward their anger moderated the relationship between the parent’s perceptions and the adolescent’s problem behaviors. Interestingly, this moderation was found in the father model only. Aggressive behaviors were elevated when the father perceived his parenting to be low in negativity and the adolescent perceived the father to be low and average in negativity. This finding is consistent with the aforementioned results from the directional discrepancy index, in that adolescents’ perception of parenting may be overriding fathers’ perception in some instances. In other instances, it seems that the fathers’ perceptions may be overriding the adolescent’s perception of the father’s negative responses (i.e., a high-low father-adolescent dyad), which seems to be leading the adolescent to self-report more aggressive behaviors. This finding extends the existing literature that parental mismanagement of adolescent anger can result in adolescent behavior problems (e.g., Garside and Klimes-Dougan 2002; Hoeve et al. 2009; Hoskins 2014), in that the relation between the adolescent’s and father’s perceptions of parenting must be considered. Specifically, this result may be a proxy for problems within the father-adolescent relationship. Moreover, this finding lends support to De Los Reyes et al.’s (2013) supposition that using an interaction term in lieu of a difference score may yield the most valid information.

It is important to consider why the interaction term (P × A) yielded significant results in father-adolescent dyads but not in mother-adolescent dyads. Although it was out of this study’s design and scope to compare maternal and paternal parenting, existing evidence has demonstrated that fathers tend to favor confrontational resolution strategies than mothers when engaging in a conflict-laden discussion with adolescents (Marceau et al. 2015). That finding suggests that fathers may generate more disagreements in the conversation with the adolescent than mothers as opposed to agreeing on a mutual resolution, which could potentially explain the father-adolescent moderation effect. Future replication studies that examine differences in maternal and paternal reports of parenting are needed.

Strengths and Limitations

The present study has several strengths that are worth mentioning. To our knowledge, this study was the first to examine discrepancies between how parents and adolescents view parental management of adolescent anger and its associations with adolescent problem behaviors. Additionally, the maternal and paternal effects were examined separately, which adds to the growing body of empirical literature regarding the father-adolescent relationship (Bronte-Tinkew et al. 2006; Fanti et al. 2008). Furthermore, our approach examined various types of discrepancy scores and took into account the main effects of perception on the outcomes. This study also purposefully examined distinct reactions from parents as opposed to general parenting behaviors (e.g., general harsh or negative parenting).

However, there are also several limitations that need to be noted. First, we did not examine an objective measure of the quality of these parent-adolescent relationships (e.g., observational measures) that could have served as an anchor that is free of subjective biases. Interpretation of discrepancy scores, especially the directional (P–A) and absolute (|P–A|) discrepancy indices, may include measurement errors. Second, our outcome measures for externalizing behaviors and aggressive behaviors were reported by adolescents. Given that our independent variables (i.e., discrepancy indices) were computed partially from adolescents’ report, the association between the dependent and independent variables may reflect shared methods variance. Thus, we recognize that obtaining and utilizing a third-party report of problem behaviors (e.g., teachers’ report) would have been preferable in order to eliminate the risk of shared method variance. Third, this current study is a cross-sectional design, thus we were not able to delineate the sequence of events (e.g., discrepancy precedes problem behaviors, or adolescent problem behaviors precede discrepancy in parent-adolescent views on parenting). Further, although we assumed that parents and adolescents reported on parental negative responses to an adolescent’s anger independently, it is likely that the views of the two individuals influence each other bidirectionally. This important interdependence in views within a dyad was not considered in this study design. Longitudinal studies are needed to clarify the sequence and transactions of views held by the two parties. Fourth, the sample was predominantly Caucasian, therefore generalizability may be hindered to a sample that is also primarily Caucasian.

These limitations notwithstanding, this study found support for the notion that discrepant perceptions of parenting constitute a risk for problem behaviors (i.e., aggression) in adolescence. The results suggest that the perception of both parents and adolescents needs to be considered when attempting to predict and understand externalizing behaviors and aggressive behaviors.

Future Directions

Overall, these results suggest that parenting effects are in the eye of the beholder. Rather than future investigators considering adolescents’ self-reports an insufficient substitute for a measure that is possibly more objective (e.g., observation), researchers can conceptualize the self-report measures as an important source of information whose subjectivity cannot be simply captured by other sources. Also, the direction and magnitude of discrepancy between the parent’s perception and the adolescent’s perception of parenting may allow researchers to begin to understand the relationship quality between the mother-adolescent and father-adolescent dyads. (De Los Reyes et al. 2013; Guion et al. 2009; Mackenbach et al. 2014; Spilt et al. 2014). Future investigators should continue include father-adolescent dyads in the investigation, as this relationship is understudied in the current literature (Hoeve et al. 2009; Lamb and Tamis-Lemonada 2004; O’Connoll 1993). Specifically, although this study was not designed to test mother and father differences, it would be imperative for future studies to incorporate a design to statistically compare results between mothers and fathers.

In the current study, the various difference indices produced different results, indicating a potential quantitative and qualitative difference in the meaning of each index. Each discrepancy index seems to be capturing a different aspect of the perceptions of the individuals within the family unit. The next step is to further unravel the mechanism that is underlying these family dynamics and the true meaning of these indices.

Conclusion

The results of the present investigation underscore the importance of understanding that not only each informant but also the dyadic synergy between the two parties contribute unique information in predicting adolescent problem behaviors. Through the use of multiple types of discrepancy indices, this study affirms De Los Reyes et al.’s (2013) results indicating that using an index of moderation to assess discrepant reports yields additional information about discrepant reports than do typical discrepancy scores, such as the direction of discrepancy. The results of the moderation index in this study also provides further support that the father-adolescent relationship is unique and has the potential to exacerbate or hinder an adolescent’s problem behaviors, specifically aggressive behaviors. This study not only provides empirical support to previous studies for the inclusion of father-adolescent dyads in adolescent research (Korelitz and Garber 2016; Shulman and Seiffge-Krenke 2016), but also advances the field by understanding how the interaction between the father-adolescent dyad affects the adolescent’s problem behaviors, specifically, aggressive behaviors.

Acknowledgments

This study is an archival study that was funded by the Section on Developmental Psychopathology (directed by Carolyn Zahn-Waxler) in the Child Psychiatry Branch of Intramural Research Program of the National Institute of Mental Health. The referenced protocol is 1 Z01 MH002775 Adolescence: Anxiety, Mood and Behavior Disorders (referred to elsewhere as The Role of Emotion in the Development of Psychopathology in Adolescence or the Adolescent Emotion Study—Protocol #97-M-0116; PIs Bonnie Klimes-Dougan and Carolyn Zahn-Waxler). We thank the youths and their parents who willingly participated in this investigation. We are indebted to the significant contributions of Barbara Usher and Kimberly Kendziora, as well as the diligent efforts of numerous fellows, interns, research assistants and volunteers who have contributed so much to the implementation of this project. We would also like to extend our gratefulness to Drs. Tuppett Yates, Cecilia Cheung, and Minako Deno for their insightful and valuable feedback on this manuscript. Finally, we have great appreciation for Danielle Samuels’ and Sofia Stepanyan’s continual comments on all versions of the article.

Author’s Contributions

LMD conceived of the current study, performed the statistical analyses, interpreted the data, and drafted the manuscript. MNN participated in the study’s design, interpreted data, and helped to draft and critically revise the manuscript. PH participated in the design and coordination of the larger study. CZW conceived of the larger study and managed and interpreted the data. BKD conceived of the larger study and participated in the coordination of participants. PH, CZW, and BKD provided critical feedback on the manuscript. All authors read and approved the final manuscript.

Compliance with Ethical Standards

Conflict of Interest

The authors declare that they have no competing interests.

Ethical Approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed Consent

Informed consent and assent was obtained from all individual participants included in the study.

Copyright information

© Springer Science+Business Media New York 2016

Authors and Affiliations

  • Laura M. Dimler
    • 1
  • Misaki N. Natsuaki
    • 1
  • Paul D. Hastings
    • 2
  • Carolyn Zahn-Waxler
    • 3
  • Bonnie Klimes-Dougan
    • 4
  1. 1.University of CaliforniaRiversideUSA
  2. 2.University of CaliforniaDavisUSA
  3. 3.University of WisconsinMadisonUSA
  4. 4.University of MinnesotaTwin CitiesUSA