Introduction

A large body of research has demonstrated significant discrepancies between parent and youths’ perceptions of youth adjustment, family functioning, and parenting quality (Achenbach et al. 1987; Bell et al. 2001; Feinberg et al. 2001). Parent–child discrepancies for these constructs (e.g., parenting skills) may reflect disrupted communication patterns (Gaylord et al. 2003; Guion et al. 2009) with deleterious effects for youth adjustment (Barnes and Olson 1985; Crouter et al. 2005). Much of the literature on parent–child discrepancies focuses on how family, child, parent, and relationship problems predict discrepancies (Achenbach et al. 1987; De Los Reyes and Kazdin 2005). However, a growing body of literature has begun to examine the opposite question – the impact of informant discrepancies themselves on youth adjustment (e.g., Guion et al. 2009; Mounts 2007).

Are Discrepancies Good or Bad

Historically, informant discrepancies often were perceived as “methodological nuisance” that obscures the objective reality of children’s and adolescents’ experience (Bell et al. 2001). Recent theoretical work suggests that discrepancies provide important information about the parent–child relationship and can directly affect youth adjustment (Feinberg et al. 2000; Guion et al. 2009; Mounts 2007). Feinberg et al. (2000) summarized two contrasting perspectives on parent–child discrepancies: one that views them as indicators of normative/healthy individuation and another that views them as evidence of problematic individual and family functioning.

From the normative perspective, parent–child discrepancies are thought to result from developmentally appropriate separation and individuation and are indicative of “healthy, autonomous [family] relationships” (see Feinberg et al. 2000, p. 533). For example, parents have been found to be more encouraging of adolescent independence and autonomy when they viewed their children as less competent than did the children themselves (Butner et al. 2009). Similarly, Holmbeck and O’Donnell (1991) found that discrepancies related to parents’ granting of autonomy predicted stronger mother–child attachment. Taken together, these studies provide some support for the normative perspective (i.e., that parent–child discrepancies, especially those related to youth autonomy, promote youth adjustment).

On the other hand, family members’ discrepant perceptions are viewed widely as indicating a lack of family cohesion and immature family structure (Feinberg et al. 2000; Guion et al. 2009) as well as problematic parent–child communication and attachment (Barker et al. 2007; Ehrlich et al. 2011; Gaylord et al. 2003; Guion et al. 2009). This view is supported by studies finding that parent–child discrepancies on domains central to the parent–child relationship and family functioning predict problematic youth adjustment. For example, discrepancies regarding parenting discipline have been found to elevate youths’ risk for poor social competence (Carlson et al. 1991; Guion et al. 2009) and internalizing problems (e.g., Gaylord et al. 2003; Guion et al. 2009). Ohannessian et al. (2000) found that greater discrepancy regarding family functioning predicted greater youth anxiety and lower youth self-competence and perceived attractiveness (see also Phinney and Ong 2002). Similarly, parent–child disagreement regarding parental warmth and parenting quality is associated with youth internalizing (Guion et al. 2009) and externalizing problems (Feinberg et al. 2000; Pelton et al. 2001). Furthermore, De Los Reyes et al. (2010) found that mother–child discrepancies related to parental monitoring and youth self-disclosure were related prospectively to greater child-delinquency. These studies demonstrate strong evidence for the negative impact of discrepancies related to parenting quality and family function, as well as to parental monitoring and youth self-disclosure, for multiple domains of child adjustment.

In sum, the literature suggests that discrepancies often predict youth adjustment problems, (Carlson et al. 1991; Feinberg et al. 2000; Gaylord et al. 2003; Holmbeck and O’Donnell 1991; Ohannessian et al. 2000; Pelton et al. 2001) but sometimes, particularly in domains relevant to youth autonomy, may predict positive youth adaptation (e.g., Butner et al. 2009; Feinberg et al. 2000; Holmbeck and O’Donnell 1991). The current study examines discrepancies in the domains of parent–child relationship quality and child self-disclosure. The aim of the study is to understand how discrepancies in these domains are related to both parent- and child-reports of youth internalizing and externalizing symptoms. In light of the theory that family members’ discrepant perceptions might indicate a lack of family member cohesion and communication, we expected that discrepancies would be associated with problematic youth internalizing and externalizing adjustment. However, given the diversity of findings in this area, and consistent with the notion that parent–child discrepancy may signify healthy, autonomous relationships, it is also possible that discrepancies will be associated with positive adjustment.

Issues in Discrepancy Research

Over the past 20 years, researchers have examined varying aspects of discrepancy (e.g., magnitude, direction, level) using several different analytical methods (e.g., discrepancy scores, interaction effects) and considering a variety of outcome reporters (i.e., parent, child, observer, criminal records). The variety of approaches used makes it challenging to draw definite conclusions about the meaning of discrepancy and to compare the findings from different studies. Many questions still remain regarding the ideal way to measure discrepancy and to understand the nature of its relationship with youth outcomes.

At the most basic level, the question of discrepancy can be viewed as one of magnitude (i.e., does more discrepancy predict more positive/negative youth outcomes?). Many studies have documented relationships between the absolute value of discrepancy scores and youth dysfunction (Barker et al. 2007; Carlson et al. 1991; Feinberg et al. 2000; Ohannessian et al. 2000; Paikoff et al. 1993; Papini and Micka 1991; Pelton et al. 2001) or increased adolescent autonomy (Carlson et al. 1991; Feinberg et al. 2000). Overall, these findings suggest that the magnitude of the parent–child discrepancy has meaningful implications for youth development.

While the magnitude of the parent–child discrepancy has been demonstrated to be important for youth adjustment, there is a relative dearth of studies examining the direction of the discrepancy (i.e., which member of the parent–child dyad reports more problems/dysfunction; Barker et al. 2007). This is unfortunate, as the implications for youth adjustment may vary depending on which member of the parent–child dyad reports greater impairment (e.g., Butner et al. 2009). The few studies that have examined the direction of the discrepancy have found meaningful effects on youth outcomes (De Los Reyes et al. 2010; Ferdinand et al. 2004; Gaylord et al. 2003; Guion et al. 2009). For example, Gaylord et al. (2003) found that teachers identified greater internalizing problems among youth who saw their parents as less disciplinary than parents saw themselves. Ferdinand et al. (2004) found that elevated parent-reported youth attention problems, compared to child-report, were related to increased externalizing behaviors (i.e., expulsion from school), whereas elevated youth-reported attention problems were associated with more internalizing problems (as demonstrated by referral to mental health services). Guion et al. (2009) found that children who reported more negative reports of parenting than their parents also reported more internalizing symptoms; when children reported more positive parenting reports than did parents, these children reported higher social competence. In addition, De Los Reyes et al. (2010) found that discrepancy was most related to youth-reported delinquent behavior in dyads where mothers reported greater parental monitoring and self-disclosure than did their children. Collectively, these findings indicate that important information is present in the direction of the discrepancy in addition to the magnitude of the discrepancy.

Methods for Examining Parent–Child Discrepancy

In the past, researchers have utilized a number of different methods for examining the discrepancy’s direction and/or magnitude. Although a full review of these various approaches is beyond the scope of the current article, these methods generally fall into two categories: interaction methods and discrepancy score methods. In the discrepancy score method, discrepancy scores are created by subtracting mother-reported scores from youth scores (or vice versa) and using these scores as predictors of youth adjustment (e.g., Feinberg et al. 2000; Guion et al. 2009; Pelton et al. 2001). In the interaction method, the interaction between youth and mother reports is examined to determine whether the relationship between youth predictor variable reports and outcome differs depending on the level of the mother’s predictor variable reports (or vice versa; e.g., Bell et al. 2001; Holmbeck and O’Donnell 1991; Miller and Drotar 2003). Both approaches have strengths and weaknesses.

The discrepancy method is superior to the interaction method in some ways, but also possesses several drawbacks. The major strength of the discrepancy method is that it can directly measure the relationship between discrepancy and outcome while also examining the impact of the magnitude of the discrepancy. The discrepancy method, however, suffers from two significant limitations. First, discrepancy scores alone fail to account for the levels of the original variables (Holmbeck and O’Donnell 1991). For example, if youth-reported positive parenting quality is strongly correlated with youth-reported adjustment, and mother-reported positive parenting quality is weakly related to youth-reported adjustment, parent–child discrepancy may predict adjustment simply due to the main effect of youth-reported parenting quality on youth-reported outcome. In other words, the strong main effect, rather than the discrepancy itself, may be responsible for the significant finding. It is possible that some of the significant findings described previously simply may be artifacts of the main effects of the primary variables. One suggestion for dealing with this limitation is to model main effects in addition to the discrepancy scores. However, including main effects in regression equations changes the interpretation of discrepancy scores so that they no longer accurately measure discrepancy (see Edwards 2001).

Holmbeck and O’Donnell (1991)asserted that, in contrast to the discrepancy method, the interaction method allows for a more thorough examination of main effects. However, the interaction effect method does not allow for direct comparison between adolescent and mothers’ scores and, therefore, does not directly measure the magnitude of the discrepancy (Edwards 2001). For example, a high mother-predictor score conveys that the mother’s score is high compared to other mothers’ scores; it does not convey information relative to child-reports. It is possible that, using this method, a dyad with seemingly congruent perceptions (e.g., both high) actually may consist of a mother with scores at the top of the range of possible scores and an adolescent with scores at the bottom of the range. This is especially likely considering that adolescents often report more problems than parents in multiple domains (Barker et al. 2007; De Los Reyes and Kazdin 2005; Guion et al. 2009; Ohannessian et al. 2000).

Holmbeck and O’Donnell (1991) also explain that the interaction method, unlike the discrepancy score method, is not affected by a curvilinear distribution. This is particularly important given that discrepancy scores fall on a curvilinear distribution (i.e., high mother/low youth and high youth/low mother scores at the extremes and agreement at the center), which can make results difficult to interpret (Holmbeck and O’Donnell 1991). However, this can be addressed in analyses involving discrepancy scores, as opposed to interaction terms, by directly examining the curvilinear relationship between mother–child discrepancy scores with youth outcomes (e.g., De Los Reyes et al. 2010; see Edwards 2001).

The Importance of Examining Multiple Outcome Informants

Another challenge in understanding the possible impact of parent–child discrepancy on youth adjustment is that effects may differ depending upon whose perception of youth outcomes is being considered. Though most of the previously mentioned studies considered discrepancy direction, none of them (except Holmbeck and O’Donnell 1991) measured youth outcomes as reported by multiple informants independently. Instead, either a single perspective is considered (e.g., Guion et al. 2009) or an aggregated composite variable is constructed from multiple informants (e.g., Feinberg et al. 2000). The field of discrepancy research is built on studies demonstrating that there exist significant differences between youths’ and parents’ perceptions of multiple constructs (see Achenbach et al. 1987; De Los Reyes and Kazdin 2005). Therefore, measuring both youth and parents’ reports of outcome variables as well as youth and parents’ reports of predictor variables is essential for a complete understanding of the relationship between discrepancy and youth adjustment.

The research reviewed thus far suggests that studies examining the impact of parent–child discrepancies should consider both discrepancies in the magnitude and direction as well as include both youth and parents’ reports of youth adjustment. To the knowledge of these authors, there are only four studies that do so (Bell et al. 2001; Holmbeck and O’Donnell 1991; Miller and Drotar 2003; Mounts 2007). Each of these studies found that, on some variables, the relationship between discrepancy and youth outcome differed depending on the direction of discrepancy and who was reporting outcome. For example, Miller and Drotar (2003) found that discrepancies related to adolescent decision-making predicted greater mother-reported diabetes-related conflict when adolescents felt that they were more in charge of decisions than did mothers, but not when mothers thought they were more in charge of decision-making than adolescents. Along similar lines, Holmbeck and O’Donnell (1991) found that, when adolescents reported being more in charge of family decision making than did mothers, mothers reported considerably more parent–child conflict compared to adolescents. Both of these studies indicate that the direction of discrepancy and who is reporting the outcome has an impact on the relationship between discrepancy and adjustment.

Furthermore, some of the findings presented by Bell et al. (2001) and Mounts (2007) suggest an interesting pattern. Specifically, when parents reported more problems than children, parents reported worse child outcomes; when children reported more problems than parents, children reported worse child outcomes. For example, Bell et al. (2001) found that the lowest adolescent-reported social competence scores occurred when adolescents rated father–child relationship quality as low and fathers rated it as high. Mounts’ (2007) study represents a particularly robust example of this pattern. She found that when adolescents reported more conflict with mothers over peer relationships than did mothers, adolescents also reported more delinquent behavior; when adolescents reported less conflict than their mothers did, then adolescents reported less delinquent behavior. A comparable pattern emerged for mother reports of adolescent delinquent behavior and drug use. This study also found that when mothers reported higher levels of involvement in daughters’ peer relationships than their daughters did, mothers reported lower levels of adolescent drug use, and vice versa. Taken together, these findings suggest that when mothers saw their relationships with their children more positively or negatively than youth did, mothers reported more positive or negative youth outcomes, respectively. The same pattern seemed to hold for youth.

It is important to note that some of the studies cited above (i.e. Bell et al. 2001; Holmbeck and O’Donnell 1991) found results that did not fit this pattern (i.e., the pattern in which the member of the dyad who saw more problems in the relationship [than the other member of the dyad] perceives more problematic youth outcomes). For example, Bell et al. (2001) found the highest parent reported adolescent social comfort scores occurred when daughters reported their relationship with fathers as more close and fathers reported the relationship as less close. Furthermore, Holmbeck and O’Donnell (1991) found that adolescents reported more conflict in the presence of greater discrepancy regarding adolescent decision making autonomy regardless of which member of the dyad reported more autonomy than the other. In light of the diversity of patterns in these studies, Bell et al. (2001) suggested that more research is needed in this area to clarify the relationship between discrepant perceptions and youth- and mother-reported youth adjustment.

Indices of Family Functioning

Studies show that conflicts within the parent–child relationship are predictive of problems in children and adolescents (Formoso et al. 2000). Conversely, positive relationships with parents serve as an important source of strength and resilience for adolescents, and have been shown to favorably impact adolescent adjustment (Formoso et al. 2000; Grotevant 1998; Masten and Shaffer 2006). Guion et al. (2009) point out that, since parenting variables strongly impact youth adjustment, discrepancies between youth and parents related to these variables are likely to affect adjustment as well. Furthermore, these authors suggest that discrepancies in this area may reflect problems in parent–child communication. Therefore, it is not surprising that much of the research examining the impact of discrepancy on youth adjustment has focused on parent–child relationship quality.

Additionally, Bell et al. (2001) cite the importance of including family process variables in discrepancy work. Children’s voluntary self-disclosure to parents is an index of family process that has received increased attention; studies show that parents’ knowledge of their children’s experiences predicts adolescent adjustment and that parents obtain most of this information through youth self-disclosure (e.g., Crouter et al. 2005; Kerr and Stattin 2000; Kerr et al. 2010). Importantly, De Los Reyes et al. (2010) found that mother–child discrepancies related to parental monitoring were related prospectively to greater child-delinquency. Of note, their index of parental monitoring included child disclosure in its construction. However, as the analyses did not examine child disclosure specifically, the impact of parent–child discrepancies on child self-disclosure remains understudied.

Gender and Developmental Considerations

Potential developmental differences in these relations largely have been ignored by prior research in this area. This is surprising, as younger children spend more time with, and self-disclose more to, parents than do older youth (Buhrmester and Furman 1987; Furman and Buhrmester 1992; Hartup and Stevens 1997). Two studies found that discrepancies related to parenting behavior are actually greater in younger children than in adolescents (Lanz et al. 2001). Similarly, Feinberg et al. (2000) found that, in younger children, greater discrepancy related to parenting behavior consistently predicted fewer problems, whereas in older adolescents very high and very low levels of discrepancy predicted problems. Although promising, the limited evidence available indicates that further examination of potential age-related differences in the impact of parent–child discrepancies is needed.

Similar to the literature on the effects of age, findings regarding gender differences in the area of parent–child discrepancies are inconclusive (De Los Reyes and Kazdin 2005). Some studies have found that the negative impact of discrepancies and the positive imact of congruence is stronger in boys than in girls (Feinberg et al. 2000; Papini and Micka 1991). Other studies, however, have found the opposite effect (Offer et al. 1982; Ohannessian et al. 2000; Feiring and Lewis 1998). As with age-related differences, additional attention to the impact of gender on discrepancy effects is warranted.

The Present Study

The goal of the current research is to add to the growing body of literature examining the effects of parent–child discrepancies on youth adjustment. In particular, this study examines the impact of mother–child discrepancies related to perceptions of parent–child relationship quality and youth self-disclosure on mother- and child-reported internalizing and externalizing adjustment. We are examining mother- and child-reported symptoms separately to ascertain whether linkages between discrepancy and child functioning differ depending upon whose perspective is being considered.

This research will employ a comprehensive analytic strategy to examine the effects of discrepancy. First, the main effects of mothers’ and children’s perceptions of relationship qualities and self-disclosure will be examined through correlational analyses and multiple regression analyses that also will include the relevant mother–child interaction terms. Next, analyses will examine whether the magnitude and direction of mother–child discrepancies impacts children’s and/or mothers’ perceptions of youth internalizing and externalizing adjustment in a wider range of variables on youth- and mother-reported youth adjustment. In addition, possible curvilinear effects will be explored. These questions will be addressed in a community sample of children in grades 5, 8, and 11, thus allowing for additional analyses testing for potential gender- and developmental differences.

Method

Sample

A total of 743 students in grades 5, 8, and 11 from eight schools in three Midwestern school districts were invited to participate. Parental consent forms were mailed directly to the parents of these students, and 623 students (84 % of the eligible youth) received parental consent. Thirteen students did not participate because of special needs, relocation out of the district, or repeated absence. The 610 youth who did participate included 176 fifth-grade (90 boys, 86 girls), 226 eighth-grade (105 boys, 121 girls), and 208 11th-grade (95 boys, 113 girls) students. This sample was 86 % European American, 5 % African American, 4 % Native American, 1 % Asian American, 1 % Hispanic or Latino, and 3 % other (i.e., biracial).

Recruitment of maternal caregivers (hereafter referred to as mothers) commenced once the youth sample was obtained. Of the 610 youth participants, 232 maternal caregivers (mothers) also consented to participate in this research. This final sample of 232 mother–child dyads included 65 fifth-grade students (29 boys, 36 girls), 80 eighth-grade students (36 boys, 44 girls), and 87 11th-grade students (40 boys, 47 girls).

Procedures

Data were collected as part of a larger study of social relationships and youth adjustment. Youth participants signed assent forms and completed questionnaires during two 45-min group-administered classroom sessions conducted approximately 2 weeks apart. All questionnaires were read aloud by trained research assistants; additional assistants were present to assist with questions. Researchers returned to the schools at least once to collect data from participants who were absent during initial administrations. Youth participants received a small token (i.e., a pencil) as compensation for their participation.

The mother-specific consent form and maternal-report questionnaires were mailed directly to mothers or, at two schools, were sent home with the students in sealed envelopes. Mothers were instructed to complete the questionnaires apart from their children. Mothers were provided with self-addressed stamped envelopes to return the completed questionnaires. Up to three packets of consent forms and questionnaires were distributed to each mother in an effort to maximize maternal participation. Mothers were not compensated for their participation.

Measures

Youth and Parent Demographics

Youth participants completed items assessing gender, age, racial/ethnic background, and the identity of their primary female caregiver. They were instructed to think about this person when answering subsequent questions about “Mom.” Mothers completed a similar questionnaire assessing age, racial/ethnic background, maternal education, yearly household income, and whether they were living with a spouse/partner.

Youth Self-Report and Child Behavior Checklist

Self- and mother-reported internalizing and externalizing adjustment was assessed with the cross-informant items common to both the Youth Self-Report (YSR) and the Child Behavior Checklist (CBCL; Achenbach and Rescorla 2001). Of the 57 items that appear on the cross-informant internalizing and externalizing indices, 53 were used in the present study (29 internalizing and 24 externalizing). At the schools’ request, four items addressing suicidality, substance abuse, and sexual behaviors were removed. For each item, youth and mothers rated how well the item describes the youth’s adjustment over the past 6 months on a three-point scale ranging from 0 (“Not True”) to 2 (“Very True/Often True”). The average sum of the 29 youth-reported internalizing items was 11.76 (SD = 7.94; possible range = 0–58); for mother-reported internalizing adjustment, M = 5.73 (SD = 5.36). The average sum of 24 youth- and mother-reported externalizing items were M = 10.42 and M = 5.53, respectively (SDs = 6.45 and 6.41). Similar to other community samples (e.g., Achenbach and Rescorla 2001), the internalizing and externalizing total scores were indicative of mild-to-moderate distress. For the current research, participants were assigned separate scores for self- and mother-reported internalizing and externalizing adjustment that were the mean of the relevant items (self-report internalizing [M = .42, SD = .28] and externalizing adjustment [M = .45, SD = .28] αs = .87; mother-report internalizing [M = .20, SD = .19] and externalizing adjustment [M = .24, SD = .28] αs = .80 and .86).

Network of Relationships Inventory Social Provisions Version

Youth- and mother-reported relationship quality was assessed with parallel versions of the Network of Relationships Inventory (NRI) Social Provisions Version (Furman 1996, 2003; see also Furman and Buhrmester 1985). Youth and their mothers were asked to report on 13 features of the mother–child relationship. The 13 features include eight social provisions (affection, admiration, companionship, instrumental aid, intimacy, nurturance, reliable alliance, and support), four negative interactions (antagonism, conflict, criticism, and dominance), and overall relationship satisfaction. Three items are used to assess each of the features. Items are rated on a five-point Likert scale ranging from 1 (“Little or None”) to 5 (“The Most”).

The scores for positive relationship quality (i.e., the eight social provisions and satisfaction) and negative relationship quality (i.e., the four negative interactions) are of interest for the present research. Therefore, youth and mothers were each assigned separate scores that were the means of the relevant items for positive relationship quality (Mchild = 3.73 [SD = .79), Mmother = 3.34 [SD = .50)] and for negative relationship quality (Mchild = 2.28 [SD = .77), Mmother = 1.70 [SD = .56)]; self-report positive and negative relationship qualities αs = .94 and .87; mother-report positive and negative relationship qualities αs = .89 and .90).

Self-Disclosure: Youth and Parent Versions

Youth’s perceptions of self-disclosure to mothers and mothers’ perceptions of the youth’s self-disclosure were assessed with a revised version of the Self-Disclosure Questionnaire (Rose 2002; adapted from Parker and Asher 1993). For the current study, the measure was revised to assess self-disclosure specifically to a mother figure (for the youth participants) and from the child (for the mothers). For example, the original item “We tell each other about our problems” was revised as two distinct items: (a) “I tell Mom private things a lot,” on the youth report and (b) “My child tells me private things a lot.” This resulted in parallel versions of a 5 item measure assessing perceptions of youth self-disclosure to mothers. Items were rated on a 5-point Likert scale ranging from 1 (“not at all true”) to 5 (“really true”). These scores demonstrated adequate reliability in the current study (i.e., child-report α = .91, M = 3.02, SD = 1.12; mother report a = .88, M = 3.46, SD = .85).

Management of Missing Data

For some participants, data were not available for each measure either due to absence or due to participants electing not to complete certain measures/items. If participants skipped more than 33 % of the items on any scale of any measure used in this study, then their data for that particular scale were excluded from analyses. If they skipped fewer than 33 % of the items on a scale, their mean score across the other scale items was used in place of the missing item(s).

Results

Preliminary Analyses

Participants with complete self- and mother-reported adjustment data did not differ from participants excluded from analyses due to missing data or lack of mother-report data on demographics, relationship quality/process variables, or externalizing symptoms (all ps > .07). Participants with complete data did differ slightly from excluded participants for child-reported internalizing symptoms (t[522] = 2.05 p < .05) such that youth with complete mother data reported slightly more internalizing symptoms. However, the difference between the means of these two groups was negligible (M = .42 included; M = .47 excluded).

Main Effects of Mother–Child Perceptions

Before examining mother–child discrepancy scores on youth internalizing and externalizing adjustment, main effects were examined in two ways. First, bivariate correlations were examined (see Table 1). As expected, there was small-to-moderate agreement between child- and mother-reported internalizing adjustment (r = .34), externalizing adjustment (r = .32), positive relationship quality (r = .23), negative relationship quality (r = .17), and youth self-disclosure (r = .38; all ps < .05).

Table 1 Correlations between self- and mother-reported internalizing adjustment, externalizing adjustment, positive relationship quality, negative relationship quality, balanced cohesion, balanced flexibility, and self-disclosure
Full size table

Of particular interest are the bivariate relations between mother–child relationship quality and youth self-disclosure with youth internalizing and externalizing adjustment. Child-reported internalizing adjustment was related significantly only to child-reported negative relationship quality (r = .26). Mother-reported internalizing adjustment was related significantly to mother-reported positive relationship quality (r = −.14), mother-reported negative relationship quality (r = .38) and mother-reported youth self-disclosure (r = −.17). For externalizing adjustment, child-reported externalizing adjustment was related significantly to child-reported positive relationship quality (r = −.22), child- and mother-reported negative relationship quality (rs = .38 and .17 respectively) and to child- and mother-reported youth self-disclosure (rs = −.24 and −.18). Mother-reported externalizing adjustment was related significantly to mother-reported positive relationship quality (r = −.22), child- and mother-reported negative relationship quality (rs = .15 and .53), and to mother-reported youth self-disclosure (r = −.21).

Next, separate hierarchical multiple regressions were conducted to examine the main effects of child- and mother-reported positive relationship quality, negative relationship quality, and self-disclosure for child- and mother-reported internalizing and externalizing adjustment (see Tables 2, 3, 4). For each analysis, gender and grade were entered in Step 1 as control variables, the main effects of child- and mother-report predictor (e.g., positive relationship quality) were entered in Step 2, and the interaction between child- by mother-report were entered in Step 3.

Table 2 Summary of the hierarchical regression analyses examining relations between self- and mother-reported positive relationship quality with self- and mother-reported internalizing and externalizing adjustment
Full size table
Table 3 Summary of the hierarchical regression analyses examining relations between self- and mother-reported negative relationship quality with self- and mother-reported internalizing and externalizing adjustment
Full size table
Table 4 Summary of the hierarchical regression analyses examining relations between self- and mother-reported youth self-disclosure with self- and mother-reported internalizing and externalizing adjustment
Full size table

Child-reported internalizing adjustment was predicted significantly by child-reported positive relationship quality (Table 2) and by child-reported negative relationship quality (Table 3). Mother-reported internalizing adjustment, on the other hand, was predicted significantly by child- and mother-reported positive relationship quality (Table 2), mother-reported negative relationship quality (Table 3), and by mother-reported youth self-disclosure (Table 4). These findings were largely consistent with the correlational analyses, except that child-reported positive relationship quality was not correlated with child- or mother-reported internalizing adjustment but was related significantly to both outcomes in the regression analyses.

Child-reported externalizing adjustment was predicted significantly by child-reported positive relationship quality (Table 2), child- and mother-reported negative relationship quality (Table 3), and by child-reported youth self-disclosure (Table 4). Mother-reported externalizing adjustment was predicted significantly by mother-reported positive relationship quality (Table 2), child- and mother-reported negative relationship quality (Table 3), and by mother-reported youth self-disclosure (Table 4). Again, these findings were largely consistent with the correlational analyses with one exception: mother-reported youth self-disclosure was correlated significantly with child-reported externalizing adjustment but was not a significant predictor in the regression analysis. Notably, none of the 12 child- by mother-reported interaction terms were significant (see Tables 2, 3, 4).

Table 5 Hierarchical regression analyses examining relations between self- and mother-reported positive relationship quality discrepancy scores with self- and mother-reported internalizing and externalizing adjustment
Full size table

In sum, mothers’ perspectives on positive relationship quality, negative relationship quality, and youth self-disclosure were related significantly to mother-reported internalizing and externalizing adjustment. Similarly, youths’ perspectives on positive relationship quality, negative relationship quality, and youth self-disclosure were related significantly to child-reported externalizing adjustment. Importantly, child-reported negative relationship quality was related consistently to both child- and mother-reported externalizing adjustment. The findings for child-reported internalizing adjustment differed somewhat across the analyses but this variable was related consistently to child-reported negative relationship quality.

Discrepancy Scores Predicting Youth- and Mother-Reported Outcomes

The next set of analyses examined whether mother–child discrepancies for positive relationship quality, negative relationship quality, and youth self-disclosure were related significantly to child- and mother-reported internalizing and externalizing adjustment. Discrepancy scores were constructed by subtracting mother-reported scores on the variables of interest (i.e., positive relationship quality, negative relationship quality, and self-disclosure) from the respective child-reported scores. As a result, positive values indicate that youth reported more positive and negative relationship quality and self-disclosure than did mothers. Negative values indicate that mothers reported more positive and negative relationship quality and self-disclosure than youth. On average, children reported slightly more positive and negative relationship quality and less youth self-disclosure than did mothers (Positive Quality Discrepancy M = .42 [SD = .84], Negative Quality Discrepancy M = .58 [SD = .87], Youth Self-Disclosure Discrepancy M = − .45 [SD = 1.11]; see Table 1 for correlations between discrepancy scores and all study variables). The discrepancy score variables met the assumptions of normality (i.e., all skewness and kurtosis values fell within the normal range between 1 and −1).

Separate hierarchical multiple regressions were conducted to examine whether mother–child discrepancy scores for positive relationship quality, negative relationship quality, and self-disclosure significantly predicted child- and mother-reported internalizing and externalizing adjustment (see Tables 5, 6, 7). Gender and grade were entered in Step 1 as control variables and the specific discrepancy score of interest (e.g., positive relationship quality) was entered in Step 2. Curvilinear effects were tested by entering the squared discrepancy score in Step 3.

Table 6 Hierarchical regression analyses examining relations between self- and mother-reported negative relationship quality discrepancy scores with self- and mother-reported Internalizing and externalizing adjustment
Full size table
Table 7 Hierarchical regression analyses examining relations between self- and mother-reported negative relationship quality discrepancy scores with self- and mother-reported Internalizing and externalizing adjustment
Full size table

Table 5 presents the results for mother–child positive relationship quality discrepancies. For internalizing adjustment, positive relationship quality discrepancy was significantly associated negatively with child-reported, but not mother-reported, internalizing symptoms (ß = −.15 p < .05). In other words, as discrepancy scores increased (indicating that youth increasingly perceived their relationships with their mothers as more positive than did mothers), child-reported internalizing problems decreased. For externalizing adjustment, a significant curvilinear relationship was found between positive relationship discrepancy scores and child-reported externalizing adjustment (ß = .15, p < .05; see Table 5), such that at extreme ends of the distribution (i.e., when mothers reported more positive relationship quality than youth and when youth reported more positive relationship quality than mothers) youth reported less externalizing symptoms. No relationships were evident between mother–child positive relationship quality discrepancy and mother-reported internalizing or externalizing adjustment.

Table 6 presents the results for mother–child negative relationship quality discrepancies. For internalizing adjustment, negative relationship quality discrepancy significantly was associated positively with child-reported, but not mother-reported, internalizing symptoms (ß = .13 p < .05). In other words, as discrepancy scores increased (indicating that youth increasingly perceived their relationships with their mothers as more negative than did mothers), child-reported internalizing problems increased. A similar pattern was evident for negative relationship quality discrepancy predicting child-reported externalizing adjustment, (ß = .20 p < .05). For mother-reported externalizing adjustment, negative relationship quality discrepancy was a significant predictor (ß = −.13 p < .05), but this was qualified by the significant curvilinear effect of negative relationship quality discrepancy (ß = .24 p < .05). At extreme ends of the distribution (i.e., when mothers reported more negative relationship quality than youth and vice versa) mothers reported less youth externalizing symptoms.

Table 7 presents the results for mother–child self-disclosure discrepancies. No linear or curvilinear effects were evident for child- or mother-reported internalizing adjustment. Similarly, no linear effects were evident for child- or mother-reported externalizing adjustment. However, the curvilinear effect of self-disclosure discrepancy significantly predicted both child- and mother-reported externalizing adjustment (ßs = .21 and .27, respectively; ps < .01). When mothers reported more self-disclosure than youth and when youth reported more self-disclosure than mothers, both youth and their mothers reported less youth externalizing problems.

Interactions Between Discrepancy Scores with Grade and Gender

To explore possible gender and grade effects, subsequent regressions tested all two- and three-way interactions between gender and mother–child discrepancy (linear and curvilinear) and between grade and mother–child discrepancy (linear and curvilinear) separately for positive relationship quality, negative relationship quality, and self-disclosure. Only two of the 48 interaction terms tested reached significance. Given that this was less than would be expected by chance (i.e., 4 % of the interaction terms were significant), these were not interpreted further.

In summary, these analyses revealed that when youth reported more positive relationship quality than their mothers (i.e., higher positive relationship quality discrepancy scores), youth reported less internalizing symptoms. Similarly, when youth reported more negative relationship quality than their mothers, youth reported more internalizing and externalizing symptoms. However, the curvilinear analyses demonstrated that greater discrepancies in the domain of self-disclosure were associated with less child- and mother-reported externalizing problems. Furthermore, greater discrepancies in the domain of negative relationship quality were associated with less mother- (but not child-) reported youth externalizing problems.

Absolute-Value Discrepancy Scores Predicting Youth- and Mother-Reported Outcomes

In the preceding analyses, the curvilinear effects of discrepancy scores on child- and mother-reported internalizing and externalizing adjustment were examined. These analyses were conducted to examine whether the amount of discrepancy, irrespective of the source of the discrepancy, predicted adjustment. Another approach to this question, one widely used in the literature examining parent–child discrepancy, is to examine the absolute value of the discrepancy scores (e.g., Barker et al. 2007; Feinberg et al. 2000; Pelton et al. 2001). Thus, the final set of analyses examined whether the absolute-value positive relationship quality discrepancy score, negative relationship quality discrepancy score, and self-disclosure discrepancy score predicted child- or mother-reported internalizing and externalizing adjustment. For these hierarchical regressions, gender and grade were entered in Step 1 as control variables, and the specific absolute value discrepancy score (e.g., positive relationship quality) was entered in Step 2. Analyses were conducted separately for child- and mother-reported internalizing and externalizing adjustment for positive relationship quality, negative relationship quality, and self-disclosure; the results of these 12 regressions are presented in Table 8.

Table 8 Hierarchical regression analyses examining relations between the absolute value discrepancy scores for positive relationship quality discrepancy, negative relationship quality discrepancy, and youth self-disclosure with self- and mother-reported internalizing and externalizing adjustment
Full size table

None of the absolute value discrepancy scores predicted child- or mother-reported internalizing adjustment. This is consistent with the nonsignificant curvilinear effects presented in Tables 5, 6, 7. For externalizing adjustment, the absolute value of mother–child self-disclosure discrepancy significantly predicted both child- and mother-reported externalizing adjustment (ßs = .20 and .16, respectively; ps < .05). This is also consistent with the results of the curvilinear analyses (see Table 7); as the discrepancy between child- and mother-reported youth self-disclosure increased, both youth and their mothers reported less youth externalizing problems.

The absolute-value negative relationship quality discrepancy score significantly predicted child-reported externalizing adjustment (ß = .15, p < .05) but not mother-reported externalizing adjustment (ß = −.02). This is opposite to the pattern evident in the curvilinear analyses (see Table 6). It is not clear why the curvilinear effect of negative relationship quality discrepancy was significant for mother-reported externalizing adjustment while the absolute-value negative relationship quality discrepancy score was significant for child-reported externalizing adjustment. It should be noted that the curvilinear analyses—by necessity—also included the main effect of mother–child negative relationship quality in the full model (see Table 6).

Discussion

Discrepancies between children’s and their parents’ perceptions of youth adjustment, family functioning, and parenting quality are the rule rather than the exception (Cantwell et al. 1997; Crouter et al. 2005; De Los Reyes and Kazdin 2005; Guion et al. 2009). Recent research has demonstrated that discrepancies related to relationship quality and parental monitoring provide important information about the parent–child relationship and can directly affect youth adjustment (e.g., Bell et al. 2001; De Los Reyes et al. 2010; Feinberg et al. 2000; Guion et al. 2009; Holmbeck and O’Donnell 1991; Mounts 2007). Yet, questions remain regarding the utility of various approaches to evaluating the meaning of discrepancies. The present research extends this literature by employing multiple approaches to examining mothers’ and their children’s discrepant perceptions: specifically, our analytic strategy evaluated main effects, whether findings differed depending upon the source of the discrepancy, and also the magnitude of the mother–child discrepancy in predicting child outcomes.

In addition, the effects of mother–child discrepancies for positive relationship quality, negative relationship quality, and self-disclosure were examined separately for child-reported internalizing adjustment, child-reported externalizing adjustment, mother-reported internalizing adjustment, and mother-reported externalizing adjustment. This purposeful design aspect was employed to determine whether mother–child discrepancies were differentially related to distinct domains of psychological functioning and whether this depended upon the source of information.

For youth internalizing adjustment, surprisingly few significant effects were evident. Furthermore, the findings consistently indicated that mother–child discrepancies on relationship quality and self-disclosure were unrelated to youth- or mother-reported internalizing symptoms. Instead, informant-specific perceptions of mother–child relationship functioning were most relevant. Specifically, the main effects of the children and adolescents’ perceptions of positive and negative relationship qualities were most relevant for child-reported internalizing symptoms; as youth viewed their relationship more positively (i.e., higher in positive quality and lower in negative quality), youth reported less internalized distress. The one discrepancy score related to child-reported internalizing symptoms (i.e., when children reported more negative relationship quality than mothers, this discrepancy predicted internalizing symptoms) also was consistent with this interpretation. Similarly, mother-reported internalizing symptoms were predicted by mothers’, but not youths’, perceptions of positive relationship quality, negative relationship quality, and youth self-disclosure. Again, mothers who viewed their relationship with their children positively reported less youth internalizing distress.

These findings were unexpected and contradict prior research that has found relationships between parenting quality and youth internalizing problems (e.g., Gaylord et al. 2003; Guion et al. 2009; Ohannessian et al. 2000). This could be due to the fact that the current study examined relationship qualities (e.g., intimacy) rather than parenting behaviors directly. However, the current results are consistent with other literature demontrating linkages between positive relationship quality and internalizing behavior (Sentse and Laird 2010).

On the other-hand, mother–child discrepancies pertaining to negative mother–child relationship quality and to youth self-disclosure were found to be relevant for externalizing symptoms. Importantly, this was evident when considering both youth and mothers’ perceptions of youth externalizing behavior. For each source of information, the main-effects analyses indicated that perceptions of problematic relationship functioning (e. g., high conflict, low child self-disclosure) were related to worse behavioral functioning. However, the curvilinear and absolute value discrepancy analyses highlighted the need to consider disagreement between mothers’ and their children’s perceptions of relationship functioning. As mothers and their children demonstrated greater disagreement for negative relationship quality and for self-disclosure, irrespective of the source of disagreement, both mothers and the youth themselves reported lower levels of problematic externalizing behavior.

This somewhat counterintuitive pattern of findings is consistent with the normative/healthy perception of parent–child discrepancies, which postulates that disagreement results from developmentally appropriate separation and individuation (see Feinberg et al. 2000). An alternative interpretation is that greater disagreement in domains of negative relationship quality and self-disclosure is indicative of a lack of parent–child communication (e.g., Barber et al. 2001; Kerr et al. 2010; Pelton et al. 2001). As a result of poor mother–child communication, mothers may be unaware of their youths’ problems. Consequently, the positive links between discrepancy and adjustment may reflect mothers’ over-estimation of youth adjustment rather than the actual relationship between discrepancy and outcome.

Similarly, greater self-disclosure discrepancy represents a lack of attention, by youth and mothers, to the frequency and content of youth self-disclosure. Perhaps when youth are experiencing fewer behavioral adjustment problems, the content of youth self-disclosure is less salient and, therefore, neither member of the dyad attends as closely to these exchanges. These findings somewhat contradict the recent work of De Los Reyes et al. (2010), which found concurrent and prospective relationships between mother–child discrepancies on parental monitoring (which included child disclosure) and child delinquent behavior. Given the increased attention placed upon the importance of children’s spontaneous disclosure to their parents on subsequent behavioral adjustment (Crouter et al. 2005; Kerr and Stattin 2000; Kerr et al. 2010), further research is needed to understand the impact of disagreement between mothers’ and youths’ perceptions of youth self-disclosure on children’s and adolescents’ externalized distress.

Limitations

Limitations of the present study should be noted. First, significant data reduction occurred as mother-reports of youth adjustment were not available for many participants (i.e., 232 participants were retained from the larger sample of 610 youth). While this participation rate is comparable to rates found in prior school-based studies in which parental participation was sought by mail (e.g., Pomerantz 2001), employing more intensive recruitment strategies, (e.g., financial incentives) particularly for maternal participants, may have prevented some data reduction. Importantly, the analyses examining the representativeness of the youth participants retained for analyses suggested few differences between included and excluded participants.

Second, participants were drawn from a predominantly white sample, which could limit the generalizability of the findings. There is evidence demonstrating that patterns of self-disclosure (Yau et al. 2009), family function (Baer and Schmitz 2007), and parent–child relationships (Jackson-Newsom et al. 2008) differ cross-culturally. Therefore, future research would benefit from exploring the current hypotheses in a more diverse sample. Third, this study employed a cross-sectional design, which limited our ability to draw conclusions about causal or prospective relationships between mother–child discrepancy and youth adjustment. Longitudinal designs would allow researchers to better understand the relationships between these variables. Recent work by De Los Reyes et al. (2010) is an excellent example of such a design, with findings that supported the utility of using discrepancies to prospectively predict child delinquent behaviors.

Another limitation of the current study was its reliance on pen-and-pencil assessment measures. Although this approach was relevant to the study aims, implementing alternative methods, such as qualitative interviews and behavioral observations, and/or more objective indices of adjustment (e.g., academic performance, treatment utilization, arrest records) would allow future researchers to determine the effects of shared-method variance on findings. However, it should be noted that the field lacks a “gold standard” for assessing child psychopathology (Achenbach et al. 1987; De Los Reyes and Kazdin 2005).

Implications and Conclusions

The current findings have important implications for research and clinical contexts. When considering youths’ affective and emotional distress, the findings support decades of research highlighting the need to consider the source of information (e.g., Achenbach et al. 1987; Cantwell et al. 1997; De Los Reyes and Kazdin 2005; Youngstrom et al. 2000). In other words, when working to understand how individual and family functioning across domains is related to youth internalizing adjustment, these and prior findings suggest that it is important to understand the particular informant’s perceptions of functioning as these may be most relevant to understanding the informant’s unique perspective on the child’s internalizing adjustment.

When considering externalizing pathology, our findings provide important evidence further supporting the need to collect multiple informants’ reports of behavioral problems and family functioning. It has been long recognized that multiple informants can provide information on how behavioral distress manifests in different contexts and address problems associated with self-report data (e.g., Achenbach et al. 1987; Campbell and Fiske 1959). The current findings, in conjunction with recent findings from related research (e.g., De Los Reyes et al. 2010; Feinberg et al. 2000; Guion et al. 2009), indicate that the mother–child discrepancies themselves provide unique, valuable information that can aid in understanding children’s and adolescents’ behavioral pathology. Building upon this body of research, future research is needed to fully understand how perceptual discrepancies in negative or maladaptive (e.g., low disclosure) aspects of mother–child relationships are formed, are understood by the members of the dyad, and contribute to the onset, maintenance, and treatment of problematic behavioral outcomes.