Developmental Changes in Discrepancies Between Adolescents’ and Their Mothers’ Views of Family Communication

Abstract

Prior work indicates that adolescents perceive the family more negatively than do their parents. These discrepant views comprise some of the most robust observations in psychological science, and are observed on survey reports collected in vastly different cultures worldwide. Yet, whether developmental changes occur with these discrepant views remains unclear. In a sample of 141 adolescents and their mothers, we examined 1-year developmental changes in discrepancies between parents’ and adolescents’ views of family functioning. We focused on discrepant views about a relatively covert domain of family functioning (i.e., internal views of open communication) and a relatively overt domain of such functioning (i.e., views about observable communication problems). We observed significant developmental changes in discrepant views for open communication, but not for communication problems. These findings have important implications for research examining links between discrepant views of family functioning and whether these discrepancies serve as risk or protective factors for adolescent psychosocial functioning.

Introduction

Adolescents and parents interact with each other in complex ways. Although most adolescents experience positive relationships with their family (Smetana et al. 2006), adolescents tend to report less family closeness and support, less communication with their parents, and increased family conflict relative to children (Farrell and White 1998; Laursen and Collins 2009; Mooney et al. 2006). These relatively negative views of the family are believed to be in part due to developmental changes that take place within the adolescent, namely cognitive changes and the adolescent’s push for increased autonomy (Montemayor 1983; Smetana et al. 2006; Spear 2000). Adolescents’ perceived views of the family often exhibit relatively little correspondence with parental views of the family (De Los Reyes et al. 2012; Ohannessian et al. 2000). These discrepant views in perceived family functioning can be defined as the distance or separation between adolescents’ perceptions and parents’ perceptions of the same domain of family functioning. For example, a mother may believe that she has an open communicative relationship with her adolescent (e.g., parent feels that it is “easy” to talk with her adolescent), whereas the adolescent may believe that very little open communication exists between herself and her mother (De Los Reyes et al. 2013a, d).

Discrepant views between adolescents and their parents occur across many domains of adolescent and family functioning (e.g., family conflict and relationship quality; parental monitoring of adolescents’ whereabouts and activities; adolescent mental health; for a review, see De Los Reyes 2013; De Los Reyes and Kazdin 2005, 2006). In fact, a recent meta-analysis of 341 studies of multiple informants’ reports of child and adolescent psychosocial functioning indicates that these discrepancies are some of the most robust observations in psychological research (De Los Reyes et al. 2015). Further, these discrepant views occur not only in research conducted in the United States, but also in research based on adolescent and parent reports of psychosocial functioning across vastly different cultures worldwide (Rescorla et al. 2013). Thus, if we increase our understanding of discrepancies between adolescents’ and parents’ views of the family, we may better understand how these discrepancies manifest not only within assessments of the various domains of psychosocial functioning relevant to adolescent development, but also within assessments carried out in psychological research internationally.

Importantly, discrepancies between parents’ and adolescents’ views of the family can be empirically distinguished from other domains of family functioning. For instance, indices of adolescent–parent discrepant views and adolescent–parent conflict share between 2 and 36 % of overlapping variance depending on the measurement method and informant (De Los Reyes et al. 2012). Further, whereas indices of adolescent–parent discrepant views uniquely predict scores on performance-based measures of interpersonal perception (i.e., emotion recognition), indices of adolescent–parent conflict do not (De Los Reyes et al. 2013b). These findings indicate that discrepant views and conflict, although related, provide distinct information about family functioning and interpersonal perception.

Although recent work has begun to examine the developmental implications of adolescent–parent discrepant views, inconsistencies across studies hinder our understanding of the role these discrepant views play in adolescent development. That is, as mentioned previously, relative to younger children, adolescents evidence increased negative views of the family in comparison to their parents (Fung and Lau 2010; Gaylord et al. 2003; Ohannessian et al. 2000; Ohannessian and De Los Reyes 2014). Further, adolescents tend to report lower levels of family cohesion, family satisfaction, parental monitoring and communication than their parents (De Los Reyes et al. 2013a; Laird and De Los Reyes 2013; Ohannessian et al. 2000; Ohannessian and De Los Reyes 2014; Reynolds et al. 2011; Yu et al. 2006). Yet, unlike other domains of family functioning, we know relatively little about discrepant adolescent–parent perceptions of the family and whether they are linked to normal or atypical developmental processes during adolescence. Consequently, key issues remain unresolved in the literature. For instance, some studies find that increased discrepancies between adolescents’ and parents’ views of the family relate to increased adaptive family and adolescent functioning (Butner et al. 2009; Carlson et al. 1991; Holmbeck and O’Donnell 1991). Results from this first set of studies are in line with the idea that discrepancies in adolescent–parent perceptions may be essential for the realignment of family relationships and the successful mastery of developmental tasks during adolescence (e.g., the development of autonomy and identity). In contrast, more recent work suggests that increased discrepancies in adolescent–parent views of family functioning relate to increased maladaptive adolescent outcomes including anxiety and disturbed mood, conduct problems, and substance use (De Los Reyes 2011; De Los Reyes et al. 2010; Juang et al. 2007; Ohannessian 2012). Results from this second set of studies reflect the idea that underlying these discrepant views is a lack of understanding in the family relationship, which may be associated with deficits in family functioning, predisposing adolescents to poor outcomes (Goodman et al. 2010). Given the mixed literature, it is not clear whether discrepancies between adolescents’ and parents’ views of the family signal risk for atypical adolescent development.

Two key factors might account for these inconsistencies across studies. First, prior work has elucidated the developmental course of specific domains of family functioning in the context of adolescent development. For instance, during the adolescent period, the frequency of conflict between adolescents and parents remains stable across early-, mid-, and late-adolescence, and what changes across these periods is the intensity of such conflict (i.e., steady increases across developmental periods; see Laursen et al. 1998). Yet, we have a poor understanding of developmental changes in adolescent–parent discrepant views of the family. That is, when do discrepant views change over the course of adolescent development and when do they not? Second, historically, assessments of discrepant views have relied on metrics of questionable reliability and validity, namely the calculation of difference scores between parent and adolescent reports (e.g., De Los Reyes and Kazdin 2004; De Los Reyes et al. 2010). Recent work indicates that these difference scores are statistically redundant with the individual reports from which they are calculated, and as such they cannot reveal meaningful information about discrepant views beyond the effects accounted for by the individual reports (De Los Reyes et al. 2013c). Recently, researchers have moved to using methods that rely on calculating and examining statistical interactions between informants’ reports (i.e., within a multiple regression framework; Laird and Weems 2011). These methods allow for the direct examination of whether differences between reports contribute to prediction, over-and-above the main effects of individual reports (Laird and De Los Reyes 2013). Further, these methods can be modified for use in examining discrepant views as either predictors, outcomes, or both (Laird and LaFleur 2015). Therefore, these methods open new avenues for testing issues surrounding developmental changes in discrepant views. In sum, by addressing gaps in the literature and leveraging new methods to do so, we may advance our understanding of the family domains for which developmental changes in adolescent–parent discrepant views do and do not occur. This work may inform future research on the family domains for which discrepant views are most likely to predict improvements or deteriorations in adolescent psychosocial functioning.

As a first step toward addressing these gaps in the literature we examined developmental changes in discrepancies between parents’ and adolescents’ views of family functioning. In selecting domains of family functioning for which understanding discrepant views may be of interest, we drew from meta-analytic work on multi-informant assessments of child and adolescent psychosocial functioning. Specifically, informants tend to exhibit less correspondence in their reports of internal or relatively covert behaviors (e.g., anxiety and mood), relative to reports of relatively observable behaviors (e.g., aggressive and oppositional behavior) (Achenbach et al. 1987; De Los Reyes et al. 2015). Similarly, we expected to observe changes over time in adolescent–parent discrepant views in the extent to which they display an open communication with each other (e.g., adolescents and/or parents feeling that it is “easy” to talk to each other). Indeed, behaviors indicative of open communication may be covertly expressed and facilitated, in part, by adolescent behaviors for which adolescents and parents tend to exhibit low correspondence in reports (i.e., adolescent’s disclosure to parent of their whereabouts; De Los Reyes et al. 2008, 2010). The covert nature of these behaviors may result in adolescents and parents encountering difficulty in observing manifestations of these behaviors at any one point, and thus changes in these behaviors across time points. Conversely, we expected to observe little change over time in discrepant adolescent–parent views on relatively overt behaviors displayed in the adolescent–parent relationship, namely the extent to which adolescents and parents perceive problems in their communication (e.g., derogatory statements made to each other). In sum, the following research questions were addressed (1) Do adolescents and their parents have similar views of their openness in communication and communication problems? and (2) Do discrepant adolescent–parent views of communication (openness and problems) change over time?

Method

Participants

The sample included 141 adolescents (57 % girls) and their mothers. At Time 1, the mean age of the adolescents was 15.99 years (SD = .70, range = 15–18). Seventy-five percent of the adolescents were European American, 12 % were African-American, 7 % were Latin American, and 2 % were Asian American (the remainder described themselves as “other”). These percentages reflect the area from which the sample was drawn (71 % European American, 23 % African American, 4 % Asian American, 7 % Latin American; U.S. Census Bureau 2008). Most of the adolescents (72 %) lived with both of their biological parents (96 % of the adolescents lived with their biological mother, 73 % lived with their biological father). The majority of mothers (96 %) and fathers (99 %) had graduated from high school. Some of the adolescents’ parents also had completed 2 years of college (19 % of mothers and 16 % of fathers) or 4 years of college (35 % of mothers and 27 % of fathers) and a minority (13 % of mothers and 15 % of fathers) had attended graduate or medical school.

Procedure

The study protocol was approved by the University of Delaware’s Institutional Review Board. Public high schools in Delaware, Pennsylvania, and Maryland (within a 60 mile radius of the University of Delaware) were invited to participate in the study. The administration from seven public high schools agreed to have their school participate. In the spring of 2007, 10th and 11th grade students from participating schools, who provided assent and had parental consent, completed a self-report survey in school by trained research personnel (all of whom were certified with human subjects training). Seventy-one percent of the students attending the study schools participated. Most of the students who did not participate were absent on the day of data collection. Only three percent of the students present on the day of data collection declined to participate.

All of the adolescents were told that participation was voluntary, that they could withdraw from the study at any time, and that the data collected were confidential. They also were informed that an active Certificate of Confidentiality from the U.S. government would further protect their privacy. The adolescent survey included measures relating to the family, coping, extracurricular activities, technology use, and their own and their parents’ substance use and psychological problems. The survey took approximately 40 min to complete. Adolescents who turned in a completed survey received a movie pass for their participation. The adolescents were invited to participate again in the spring of 2008 (Time 2). The same protocol (approved by the University of Delaware’s Institutional Review Board) was used at that time.

At both times of measurement, the parents of participating adolescents were mailed a packet with an invitation to participate in the study. The parent packet included a cover letter, a consent form, a parent survey, and a prepaid envelope for the return of the survey materials. The parent survey included measures relating to the family, and their own and their adolescent’s substance use and psychological problems. Participating parents were mailed a $20 gift card upon receipt of their completed survey. Both mothers and fathers were invited to participate. However, the response rate from fathers was relatively low (n = 67, 46 % at Time 1). Therefore, only adolescents and their mothers (n = 141 adolescent–mother dyads) were included in this study. It is important to note, however, that a focus on discrepancies between mother and adolescent reports is consistent with previous work in the informant discrepancies literature (see De Los Reyes et al. 2010, 2013a; Laird and De Los Reyes 2013).

We compared adolescents and parents who participated at both times of measurement to those who only participated on one occasion to examine whether they systematically differed from one another. We observed no significant differences for any of the demographic variables (age, gender, race/ethnicity) or the communication variables between the longitudinal and non-longitudinal subsamples.

Measures

All of the participants completed a demographic questionnaire. This questionnaire included items relating to age, gender, race/ethnicity, and education. In addition, adolescents and their mothers completed the measure of communication discussed below.

Adolescent–Mother Communication

The Parent–Adolescent Communication Scale (PACS; Barnes and Olson 2003) includes two 10-item subscales—Open Family Communication and Problems in Family Communication. The Open Family Communication sub-scale focuses on the free exchange of emotional and factual information. In contrast, the Problems in Family Communication subscale focuses on negative aspects of communication. A sample item from the Open Family Communication subscale is “My mother/child is always a good listener.” A representative item from the Problems in Family Communication subscale is “My mother/child insults me when she/he is angry with me.” Adolescents and parents respond to the same items in reference to each other. The PACS response scale ranges from 1 = strongly disagree to 5 = strongly agree. Prior research has supported the validity and reliability of the PACS (Barnes and Olson 2003; Hartos and Power 2000). In our sample, the Cronbach’s alpha coefficients ranged from .81 to .93.

Data-Analytic Plan

Analyses first estimated cross-informant correspondence at the two time points using paired t tests to assess mean-level correspondence and correlations to assess congruence in reports. Longitudinal tests of developmental change in discrepant views utilized multi-level modeling procedures to test longitudinal changes in informant correspondence and to separate between-family variance (i.e., some adolescent–parent dyads are more discrepant than others) from within-family variance over time (i.e., degree to which fluctuations in communication levels are time-lined within dyads). Parent reports were modeled as the outcome variable with adolescent reports and time (i.e., dummy coded 0 = Time 1, 1 = Time 2) serving as predictors. The mean level of adolescent reports across the two time-points was included in the model as a between-person predictor at level 1. Person-centered adolescent reports at the two time points were included in the model as within-person predictors at level 2 along with time and the time × adolescent report interaction as follows (where i refers to time-point and j to adolescent–parent dyad):

$$\text{Level}1:Y_{ij}={\beta }_{oj}+{\beta }_{1j}\times \text{Mean}\text{Adolescent}{\text{Report}}_{ij}+e_{ij}.$$ (1)

$$\text{Level}2:{\beta }_{oj}={\gamma }_{00}+{\gamma }_{01}\times \text{Adolescent}{\text{Report}}_{1j}+{\gamma }_{02}\times {\text{Time}}_{2j}+{\gamma }_{03}\times \text{Time}\times \text{Adolescent}{\text{Report}}_{3j}+u_{0j}.$$ (2)

The coefficient for the level 1 adolescent reports (β_1j) indexes cross-informant correspondence at the between-person level. The coefficient for the level 2 adolescent reports (γ₀₁) indexes time-linked cross-informant correspondence at Time 1 and the coefficient for the time × adolescent repot interaction (γ₀₃) indexes change in cross-informant correspondence from Time 1 to Time 2.

Results

Correspondence in Reports of Adolescent–Parent Communication

Table 1 presents the descriptive statistics and correlations for both communication variables for both informants at both time points. Mean comparison via paired-sample t tests show that parents reported significantly more open family communication than did adolescents at both Time 1, t(117) = 4.46, p < .001, and Time 2, t(89) = 2.69, p = .009. In contrast, adolescents reported significantly more communication problems than did parents at both Time 1, t(122) = −7.16, p <.001, and Time 2, t(117) = −4.98, p < .001. There were no significant changes from Time 1 to Time 2 in parent or adolescent reports of open communication or communication problems (all ps > .33).

Table 1.

Descriptive statistics and bivariate correlations for family communication

Variable	N	M (SD)	1	2	3	4	5	6	7
Open family communication
Parent T1	118	40.28 (5.59)
Adolescent T1	118	37.01 (8.71)	.45***
Parent T2	90	40.24 (6.43)	.74***	.25*
Adolescent T2	90	37.34 (9.03)	.25*	.75***	.15
Problems in family communication
Parent T1	123	21.85 (6.45)	−.63***	−.31***	−.63***	−.18
Adolescent T1	123	27.67 (7.60)	−.28**	−.63***	−.12	−.52***	.19*
Parent T2	90	22.04 (7.14)	−.52***	−.29*	−.58***	−.13	.61***	.14
Adolescent T2	90	27.13 (8.60)	−.04	−.49***	.03	−.64***	.15	.58***	.21*

T1 Time 1, T2 Time 2

^*p < .05

^**p < .01

^***p < .001

Correlations show that correspondence levels between parent and adolescent reports of open communication appeared to be moderate-to-large at Time 1 and low at Time 2 (rs = .45 and .15, respectively), based on effect size conventions from Cohen (1988). In contrast, correlations between parent and adolescent reports of communication problems appeared to be low in magnitude at both Time 1 and Time 2 (rs = .19 and .21, respectively). For both indices of family communication, correlations across time and informants (i.e., rs = .15 to .45) ranged from low-to-moderate/large in magnitude, whereas correlations across time and within informants on the same domain (i.e., rs = .58 to .75) were consistently large in magnitude.

Longitudinal Changes in Discrepant Views of Adolescent–Parent Communication

Table 2 presents the fixed effects from the multi-level models. For open communication, mean adolescent reports at level 1 were a significant predictor of parent reports showing that there was significant congruence between adolescent and parent reports across families. At level 2, both the adolescent report of open communication and the interaction between time and adolescent reports of open communication were significant predictors of parent reports. These level 2 effects showed that the time-specific link between adolescent and parent reports of open family communication was significantly greater at Time 1 than at Time 2.

Table 2.

Multi-level modeling of developmental changes in discrepant views of family communication

	Open family communication		Problems in family communication
	β/γ (SE)	p	β/γ (SE)	p
Level 1
Mean adolescent report	.25 (.06)	.000	.18 (.07)	.019
Level 2
Adolescent report	.46 (.17)	.008	.11 (.17)	.53
Time	−.01 (.48)	.99	.07 (.66)	.92
Time × adolescent report	−.79 (.31)	.011	−.02 (.29)	.95

For communication problems, only mean adolescent reports at level 1 were a significant predictor of parent reports indicating that there was modest congruence between adolescent and parent reports across families. There was no evidence that the time-specific link between adolescent and parent reports of communication problems differed from Time 1 to Time 2.

Discussion

The purpose of this study was to extend the literature on the longitudinal assessment of adolescents’ and parents’ views of family functioning. Our study yielded three primary findings. First, consistent with prior work (e.g., De Los Reyes et al. 2015; De Los Reyes et al. 2013d), we observed low-to-moderate magnitudes of correspondence between parent- and adolescent-reports of the two key family domains assessed in the study, open communication and communication problems. Second, adolescent- and parent-reports differed based on the type of family domain assessed. Specifically, adolescents reported significantly higher levels of communication problems relative to parents, whereas parents reported significantly higher levels of open communication relative to adolescents. These differences manifested when comparing adolescent and parent reports at both time points of assessment in our study. Further, these findings are consistent with a large body of work indicating that adolescents tend to report lower levels of family functioning relative to reports provided by parents (De Los Reyes et al. 2013a; Laird and De Los Reyes 2013; Ohannessian et al. 2000; Ohannessian and De Los Reyes 2014; Reynolds et al. 2011; Yu et al. 2006). Our study extends this work by demonstrating that this effect appears to persist over time (i.e., within-dyad changes over time). Third, we observed significant changes in discrepant views over time for adolescents’ and parents’ reports of open communication, but not for reports of communication problems. This finding was consistent with our hypothesis that we would observe significant changes in discrepant views on a family domain for which its constituent behaviors might manifest covertly (i.e., open communication) relative to a family domain for which behaviors reflective of the domain would be more likely to be expressed overtly (i.e., communication problems). Thus, our findings suggest that when comparing adolescent and parent reports of family functioning and longitudinal changes in discrepant views on these domains, observing significant changes in such discrepant views may depend on the family domain assessed.

Limitations and Directions for Future Research

Although the present study contributes to the literature by using methodologically sound techniques to examine discrepant views between adolescents and parents, some caveats should be noted. First, consistent with the majority of studies within this area, parental data relied on mothers’ reports. Fathers were invited to participate in the larger project; however, the response rate from fathers was low. As such, we only included adolescent- and mother-report data to address aims for this study. Because fathers were not included, we can only speculate about whether these effects generalize to understanding discrepant views between adolescents and their fathers. According to prior meta-analytic work, mother and father reports yield some of the largest levels of correspondence (i.e., rs in .50 s to 60 s) relative to other informant pairs (e.g., parent and teacher; teacher and child; Achenbach et al. 1987; De Los Reyes et al. 2015). Thus, although mother and father reports are not redundant with each other, they correspond to a sufficient extent that our findings regarding adolescent and mother reports may generalize to understanding discrepant views between adolescents and fathers. Nevertheless, future research including fathers needs to be conducted to test these expectations. Of note, all of the families resided in the Mid-Atlantic United States. Therefore, the results may not generalize to adolescents living outside of this area.

Second, measuring discrepant views via multi-informant statistical interactions carries with it a number of shortcomings. These limitations include low reliability of scores taken from interaction computations, low statistical power to detect interaction effects, and the greater difficulty inherent in correctly interpreting interaction effects (for a review, see Cohen et al. 2003). Importantly, readers should not assume that our findings indicate that multi-informant statistical interactions should be deemed primary measures of discrepant views. That is, recent work indicates that such discrepant views can be reliably and validly assessed using direct structured interview assessments of adolescents’ and parents’ subjective impressions of discrepant views of the family (De Los Reyes et al. 2012, 2013b, c). Thus, we encourage future research to take a multi-method approach to assessing discrepant views, including assessments that use indirect (e.g., statistical interactions) and direct (e.g., structured interviews) methods to assess such views.

Despite these limitations, this study extends the literature in many respects. Much of the extant research is characterized by design problems, including the use of non-representative samples. The sample for the present study was drawn from the community. Moreover, this study employed a longitudinal design, enabling the examination of change in discrepant adolescent–parent views of the family over time.

Findings from this study also point to interesting directions for future research. Specifically, research has long supported the importance of taking both adolescents’ and their parents’ perceptions into consideration given that family members may contribute different, but equally important, information relating to their family (De Los Reyes and Kazdin 2005). Our study contributes to the growing body of support for multi-informant approaches to psychosocial assessment in that it points to candidate psychosocial domains amenable to studying the risk and/or protective benefits of discrepant views of family functioning. Specifically, we postulate that a key reason for inconsistencies across studies in terms of whether discrepant views predict adaptive versus maladaptive outcomes for adolescents is that we have a poor understanding for which psychosocial domains that adolescents and their parents exhibit changes in their discrepant views over time. In line with these gaps in knowledge, we found that developmental changes in discrepant views appear to most likely occur for psychosocial domains for which its constituent behaviors are covertly expressed (e.g., open communication), rather than overtly expressed (e.g., communication problems). Although speculative, we surmise that discrepant views about covertly expressed psychosocial phenomena may be those discrepant views that hold the most promise for predicting adolescent outcomes over time. This is because the covert nature of such views may escape the attention of both parent and adolescent as a cause for concern. At minimum, the overt/covert distinction between discrepant views of family psychosocial domains may be a useful heuristic for future work seeking to identify domains for which adolescents and parents exhibit developmental changes in discrepant views.