Investigation of Change in Adolescent Perceptions of Mothers' and Fathers' Contributions to Interparental Discord From 7th to 9th Grades
Abstract
The purpose of the present study was to examine whether adolescent perceptions of mothers' and fathers' contributions to interparental discord changed from early to middle adolescence and if the changes were related to adolescent negative mood. Data were drawn from the National Longitudinal Survey of Youth 1997. Adolescents who were in 7th grade at the Wave 1 assessment and lived with both biological parents during Waves 1–4 were included in this research (n=812; 55% boys; 69% White). Findings indicated that adolescents' perceptions of their mothers' and their fathers' contributions to interparental discord increased at similar rates from early to middle adolescence. The largest increases in adolescent perceptions of mothers' and fathers' contributions to interparental discord from 7th to 9th grades were associated with the largest increases in adolescent negative mood from 7th to 10th grades. Girls' perceptions of their fathers' contributions to interparental relationship problems increased at a steeper rate compared with boys. Findings are discussed in the context of the sensitization hypothesis.
In recent years, there has been a notable increase in the number of longitudinal studies examining the relation between interparental discord and youth adjustment. These investigations have made important contributions to current understanding of processes underlying the complex associations between interparental difficulties and youth outcomes (Cummings, Schermerhorn, Davies, Goeke-Morey, & Cummings, 2006; Davies, Sturge-Apple, & Cummings, 2004; Gerard, Krishnakumar, & Buehler, 2006; Jenkins, Simpson, Dunn, Rasbash, & O'Connor, 2005). One aspect of longitudinal studies designed to elucidate the association between interparental discord and youth outcomes that has received little empirical attention is change in interparental discord over time (Cui, Conger, & Lorenz, 2005). Consideration of change in interparental discord in relation to youth adjustment seems particularly important in light of research documenting the course of marital functioning. Specifically, longitudinal studies consistently indicate that on average marital problems increase over time (Hatch & Bulcroft, 2004; VanLaningham, Johnson, & Amato, 2001). Findings also suggest that negative marital experiences tend to increase at the steepest rates for couples with children in the home (Kurdek, 1999). It thus appears that some youth are likely to experience increased levels of interparental discord over the course of development.
The sensitization hypothesis provides a framework for explaining why increases in interparental discord over time may have a negative impact on youth adjustment (Davies, Sturge-Apple, Winter, Cummings, & Farrell, 2006). This perspective, which is shared by both major theories in the area (i.e., emotional security theory and the cognitive-contextual framework; Cummings & Davies, 1996; Grych & Fincham, 1990, respectively), posits that children and adolescents who experience repeated exposure to interparental discord will show increasing reactivity to interparental discord over time (Cummings, Davies, & Campbell, 2000). This sensitization, indicated by changes in behavior (e.g., attempts to mediate disputes), emotion (e.g., feelings of anger and sadness), and cognition (e.g., worry and self-blame), has been documented in both laboratory and self-report studies (for a review see Cummings et al., 2000). Importantly, this process of sensitization is hypothesized to place youth at risk for emotional and behavioral problems over time. Hyper-vigilance may be adaptive in the immediate context of the perceived threat of conflict, but it may be disruptive if present during developmental tasks for which hyper-vigilance is not required (Cummings et al., 2000).
When interpreting findings from this literature, however, it is important to note that repeated exposure to conflict is often defined simply by ongoing exposure to high levels of conflict, or by repeated exposure to conflict within a short time period (e.g., in the laboratory). Longer-term change in interparental discord generally has not been investigated, even in studies using longitudinal designs. Instead, interparental discord has been operationalized as a stable construct, which as discussed is inconsistent with developmental research on the course of marriage (VanLaningham et al., 2001). This is an important limitation because, according to the sensitization hypothesis, exposure to increased levels of problems between parents over time should amplify how responsive youth are to the conflict, thereby putting them at increased risk for adjustment problems. To advance current understanding of the association between interparental discord and youth outcomes, there is a need to examine whether exposure to increasing levels of interparental discord over time contributes to increasing levels of emotional and behavioral problems.
To date, only one study has been designed specifically to examine the relation between change in interparental relationship problems and youth adjustment. In this prospective, longitudinal investigation (four time points) of 451 adolescents, Cui et al. (2005) found a significant increase in parent-reported marital distress from early to late adolescence. Moreover, the largest increases in marital distress were associated with increases in a wide range of adjustment difficulties during adolescence, including symptoms of depression and anxiety. It is important to note that the relation between change in marital distress and adolescent adjustment was evident after statistically controlling for initial levels of marital distress. Thus, change in marital distress made an independent contribution to the prediction of adolescent outcomes.
There are several ways to build on results reported by Cui et al. (2005). One is to consider adolescent perceptions of interparental discord instead of parent reports. Adolescent perceptions of interparental difficulties explain unique variance in adolescent adjustment even after controlling for parents' and observers' ratings of discord (e.g., Wierson, Forehand, & McCombs, 1988). Specifically, adolescent ratings of the properties of interparental conflict including frequency, intensity, and resolution quality correlate significantly with both internalizing and externalizing problems (e.g., Cummings et al., 2006; Dadds, Atkinson, Turner, Blums, & Lendich, 1999; Gerard, Buehler, Franck, & Anderson, 2005). Although adolescent perceptions of interparental discord have been linked to a wide range of outcomes, several studies have noted especially strong links between these perceptions and adolescent emotional rather than behavioral problems (Dadds et al., 1999; Gerard et al., 2005; Harold, Osborne, & Conger, 1997).
Of particular interest in the current research are adolescents' perceptions concerning mothers' compared with fathers' contributions to interparental discord. Most research assesses adolescent perceptions of interparental conflict at the dyadic level. When adolescents do rate mothers' and fathers' behavior separately, researchers commonly form a composite measure of interparental discord which does not distinguish between mothers' and fathers' contributions to conflict (e.g., Reese-Weber, 2000). In a notable exception, Mann and Gilliom (2004) found that adolescents' psychological distress was predicted by perceptions of fathers' (but not mothers') hostile control in the context of interparental conflict. Consistent with these findings, examination of how adolescent perceptions of their mothers' compared with their fathers' contributions to interparental discord change during adolescence may prove useful for better understanding associations between interparental discord and adolescent adjustment.
Another way to extend knowledge of the association between change in interparental discord and adolescent outcomes is to consider possible youth gender differences. Findings consistently indicate that boys and girls are exposed to, and typically report, similar levels of interparental discord in cross-sectional studies at various points in childhood (e.g., Grych & Fincham, 1990). What remains unclear is whether there are gender differences in the rate at which youth perceptions of interparental conflict change with development. The issue is more complex when considering potential differences in youth perceptions of mothers' contributions versus fathers' contributions to discord. Notable in this regard, children and adolescents may make attributions of blame based on their own interactions with each parent.
Research to date has been mixed regarding which parent–child relationships are most vulnerable, and by extension, how youth perceptions of mothers' and fathers' contributions to interparental conflict may be affected. Several authors have suggested that across childhood, fathers may be more vulnerable than mothers to marital discord “spillover” effects which, in turn, more adversely affect paternal parenting abilities (e.g., Brody, Arias, & Fincham, 1996; Jouriles & Farris, 1992). It is possible that boys and girls may be more likely to blame fathers for discord given perceptions of a more negative parent–child relationship. In one analogue study, children (and boys more than girls) attributed increased blame to fathers than to mothers for video-taped disagreements (Grych, 1998). Yet, other studies suggest that it is only the father–daughter relationship that is at risk. For example, Crockenberg and Forgays (1996) found that, even when objective reporters failed to observe differences, young girls (age 6) still perceived more paternal anger in marital conflicts than boys. There is also evidence that marital conflict may result in more parent–child negativity in cross-gender relationships (i.e., mother–son, father–daughter) than same-gender relationships (Osborne & Fincham, 1996; see also Snyder, 1998). By this rationale, girls may be more likely to blame their fathers for marital discord, whereas boys may be more likely to blame their mothers. Given that findings indicate possible gender differences related to appraisals of interparental relationship problems (e.g., Crockenberg & Forgays, 1996), it also seems important to consider whether boys' and girls' perceptions of their mothers' and fathers' contributions to interparental discord change at different rates.
The developmental period from early to middle adolescence may be an important time for examining the relation between change in youth perceptions of interparental discord and adjustment. Although adolescence is no longer viewed as a period of storm and stress, the physical, cognitive, emotional, and social changes that occur during this time present numerous adaptive demands (for discussion, see Arnett, 1999). Increased levels of interparental relationship discord may be a salient risk condition from early to middle adolescence because the problems add to the more normative challenges adolescents must manage. Literature on developmental differences concerning the association between interparental conflict and youth adjustment provide some albeit indirect support for this idea. Evidence suggests that children compared with adolescents are likely more affected by elevated levels of interparental discord (see Richmond & Stocker, 2007, for an overview). This age effect, however, may be moderated by the level of interparental conflict youth experience. Perhaps most notable in this regard, Richmond and Stocker (2007) recently found that although, on average, youth perceived less threat related to interparental conflict from childhood through adolescence, those exposed to increasing levels of marital conflict over time also exhibited increasing levels of perceived threat over time. Given that perceived threat has been associated with a wide range of emotional and behavioral problems (e.g., Grych, Harold, & Miles, 2003), it seems likely that youth exposed to increased levels of interparental discord during adolescence are at increased risk for adjustment difficulties.
Adolescence is characterized by heightened levels of negative emotional experiences (Arnett, 1999). Identification of factors that contribute to negative mood during this developmental period thus seems particularly important. The present research was designed to examine whether adolescent perceptions of mothers' and fathers' contributions to interparental discord were related to increases in adolescent negative mood. This approach is consistent with a developmental psychopathology perspective, which aims to understand individual differences not only in terms of symptoms and syndromes of psychopathology, but also more normative factors and processes (Cicchetti & Rogosh, 2002).
PRESENT STUDY
The present study had three primary aims designed to build on knowledge of the longitudinal relation between interparental discord and adolescent adjustment. A large, nationally representative, multiwave data set that targeted adolescents (National Longitudinal Survey of Youth 1997 [NLSY97]) was used. The first aim was to examine whether adolescents' perceptions of their mothers' and their fathers' contributions to interparental discord changed from early to middle adolescence. Informed by prior work on change in husbands' and wives' reports of marital problems (Cui et al., 2005), it was hypothesized that adolescent perceptions of mothers' and fathers' contributions to interparental discord would both increase over time. It should be noted that absolute changes in adolescent perceptions of each parent's contribution to the interparental relationship problems were considered rather than fluctuations in the percentage of discord attributed to each parent.
The second aim of the study was to examine whether change in adolescent perceptions of mothers' and fathers' contributions to interparental discord was related to change in adolescent negative mood. It was hypothesized that the largest increases in adolescent perceptions of interparental discord would be associated with increased levels of adolescent negative mood even after accounting for initial levels of interparental discord. The final aim of this study was to investigate if boys' and girls' perceptions of their mothers' compared with their fathers' contributions to interparental discord changed at different rates. Given that limited research has been conducted in this regard, we considered the third aim to be exploratory and thus made no specific hypotheses concerning possible patterns of sex differences.
METHOD
Participants
Data were drawn from the NLSY97, a nationally representative sample of approximately 9,000 youth initially assessed at ages 12–16. The NLSY97 is conducted by the U.S. Bureau of Labor to investigate youth's transition from adolescence to adulthood. Collected annually from 1997 to 2003, the NLSY97 currently includes seven assessment periods. The present study used data from adolescent-report batteries administered at Waves 1–4 (7th–10th grades). To account for possible developmental differences, only data from participants who were in 7th grade at Wave 1 were included in the present study (n=1,827; 20.3% of total sample). In addition, only adolescents who lived continually with both biological parents during Waves 1–4 (7th–10th grades) were included (n=812; 54.6% boys). The sample was limited by family composition in this manner because aims of the study were to better understand change in adolescent perceptions concerning interparental discord. Given that changes in family composition could potentially impact adolescents' perceptions of their parents' relationship and thus introduce an additional source of variability, a decision was made to use a relatively homogenous sample regarding family composition. The final sample was nationally representative in terms of racial/ethnic background of participants (14.7% Black, 69.2% White, 11.5% Hispanic, 3.2% Asian, and 1.4% other racial/ethnic background), as well as socioeconomic background of participants (mean gross household income=$53,745; 29.1% below $25,000 per year, 29.1% between $25,000 and $50,000 per year, 33.0% between $50,000 and $100,000, and 8.8% above 100,000 per year).
Procedure
The NLSY97 data set includes a stratified, nationally representative probability sample and a supplementary sample, which oversampled Black and Hispanic respondents. Sample segments from a total of 147 nonoverlapping primary sampling units were chosen, from which housing units were selected. Families of selected housing units were then asked to complete an in-home screener questionnaire. If a household included one or more individuals in the eligible age range (12–16 years), the family was invited to participate. At each wave, adolescents completed a self-report battery in their homes, which included measures pertaining to employment, education, family background, and social history. The present study used data from adolescent-report batteries administered at Waves 1–4 (7th–10th grades).
Measures
Interparental discord. A six-item measure assessing interparental discord was adapted from the widely used IOWA Youth and Family Project, a large scale project of families in the rural United States (Conger & Elder, 1994). In the present study, these six items were used to assess adolescent perceptions of mothers' and fathers' contributions to interparental discord. Items related to this scale were asked of adolescents at the first three assessment periods (7th–9th grades). Adolescents were asked to rate on a 5-point Likert scale ranging from 0=Never to 4=Always how often each parent engaged in six behaviors toward their husband/wife (i.e., screaming, insulting or criticizing, blaming, showing love, compromising, and encouraging). Separate scales were created for adolescent perceptions of their mothers' and fathers' contributions to interparental discord by summing the six items, with higher scores indicated greater levels of perceived interparental discord for each parent (the showing love, compromising, and encouraging items were reversed coded). For adolescent perceptions of mothers' contributions, internal consistencies (Cronbach's αs) from 7th to 9th grades were .71, .79, and .83, respectively. Similarly for adolescent perceptions of fathers' contributions, internal consistencies from 7th to 9th grades were .80, .83, and .85, respectively.
Adolescent negative mood. Wave 1 (7th grade) and Wave 4 (10th grade) measures of adolescent negative mood were included in the present research. A limitation of the NLSY97 data set for purposes of this investigation is that the same items related to adolescent negative mood were not included at the two assessment periods. Given that it was important for aims of the study to control for initial levels of adolescent negative mood in primary analyses, we decided to include all items related to adolescent negative mood that were available at each assessment period. At Wave 1 (7th grade), a one-item measure of adolescent negative mood was used. Adolescents were asked to rate on a 3-point Likert scale (0=Not True, 1=Somewhat/Sometimes True, 2=Often True) how often in the past month they felt unhappy, sad, or depressed. Adolescents reported low to moderate levels of negative mood at Wave 1 (mean=.42; standard deviation=.57; actual range of the measure was 0–2). At Wave 4 (10th grade), a five-item measure was used to assess adolescent negative mood. For each item, adolescents were asked to rate on a 4-point Likert scale (1=None of the Time, 2=Some of the Time, 3=Most of the Time, 4=All of the Time) how often in the past month they felt the following emotional experiences: nervous, calm and peaceful, downhearted or blue, happy, and down in the dumps/inconsolable. The measure of adolescent negative mood at Wave 4 was created by summing the five items (with calm and peaceful and happy reverse scored). Internal consistency for this measure was .76. Adolescents reported low to moderate levels of negative mood at Wave 4 (mean=12.56; standard deviation=2.28; actual range of the measure was 8–20). To put the Wave 1 and Wave 4 measures of adolescent mood on the same metric, we standardized both variables (mean=0; standard deviation=1). Higher scores on both measures indicated higher levels of negative mood.
RESULTS
Latent Trajectory Models (LTMs)
The design of the NLYS97 study has introduced numerous complexities for statistical analyses. Recommended procedures that correct for design effects thus yielding unbiased parameter estimates were followed for all analyses conducted in the present study. Descriptive statistics and zero-order correlations are presented in Table 1. As can be seen, on average, adolescents tended to perceive their mothers and fathers as engaging in low to moderate levels of interparental discord (mean scores ranged from 5.25 to 6.51 on scales that ranged from 1 to 24). Importantly, means for the measures of adolescent perceptions of mothers' and fathers' contributions to interparental discord increased from 7th to 9th grades. To investigate study aims, LTMs were estimated (Karney & Bradbury, 1995). LTMs were fit using the statistical package Mplus Version 3.1 (Muthén & Muthén, 1998–2004). Full information maximum likelihood with robust standard errors was used, which assumes that data are missing at random. For Waves 1 through 4, 761, 735, 746, and 720 adolescents, respectively, had complete data at each assessment point.
| Measures | 1a | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
|---|---|---|---|---|---|---|---|---|---|
| Background characteristics | |||||||||
| 1. Adolescent biological sex | |||||||||
| Interparental discord: Mothers' contribution | |||||||||
| 2. 7th grade | −.02 | ||||||||
| 3. 8th grade | .01 | .63*** | |||||||
| 4. 9th grade | .03 | .61*** | .76*** | ||||||
| Interparental discord: Fathers' contribution | |||||||||
| 5. 7th grade | .02 | .63*** | .43*** | .43*** | |||||
| 6. 8th grade | .09* | .47*** | .65*** | .53*** | .61*** | ||||
| 7. 9th grade | .11** | .43*** | .55*** | .70*** | .59*** | .73*** | |||
| Adolescent negative mood | |||||||||
| 8. Wave 1 (7th grade) | .03 | .26*** | .21*** | .15*** | .25*** | .18*** | .18*** | ||
| 9. Wave 4 (10th grade) | .21*** | .18*** | .30*** | .31*** | .18*** | .29*** | .32*** | .26*** | |
| Mean | 0.46b | 5.71 | 6.33 | 6.75 | 5.25 | 6.21 | 6.51 | −0.02 | −0.02 |
| SD | — | 3.63 | 3.84 | 4.25 | 3.99 | 4.22 | 4.52 | 0.98 | 0.98 |
| Skew | — | 1.11 | 0.99 | 0.87 | 1.27 | 1.02 | 1.02 | 0.96 | 0.50 |
| Kurtosis | — | 2.02 | 1.25 | 0.37 | 2.02 | 0.94 | 1.07 | −0.07 | 0.19 |
| Range | 0–1 | 1–24 | 1–22 | 1–24 | 1–24 | 1–24 | 1–24 | −0.75– 2.67 | −1.99– 3.19 |
- Note. SD=standard deviation. Ns ranged from 698 to 812. Boys=0 and Girls=1. a Point–biserial correlations. b % Girls. * p<.05; ** p<.01; *** p<.001.
Change in Adolescent Perceptions of Interparental Discord
The first aim of the study was to investigate change in adolescent perceptions of interparental discord from early to middle adolescence. Given that it is possible for adolescent perceptions of mothers' and fathers' contributions to interparental discord to follow different longitudinal trajectories, separate LTMs were estimated for each parent.1 For each LTM, two latent factors were defined: one representing initial level of adolescent perceptions of interparental discord (i.e., intercept) and the other representing linear change in adolescent perceptions of interparental discord over time (i.e., slope). Factor loadings for the intercept for the three observed measures of adolescent perceptions of interparental discord (7th–9th grades) were each fixed to 1; factor loadings for the slope were set to 0, 1, and 2 to reflect intervals between assessment periods in the study (see Figure 1). Based on this coding scheme, the intercept latent factor represented the level of adolescent perceptions of interparental discord during 7th grade (initial level) and the slope latent factor represented linear change in adolescent perceptions of interparental discord from 7th to 9th grades. A covariance was estimated in each LTM between the intercept and slope latent factors. Covariances were not estimated among error terms associated with the measures of adolescent perceptions of interparental discord from 7th to 9th grades.
Latent trajectory models estimated separately for adolescent perceptions of mothers' and fathers' contributions to interparental discord (ns=807, respectively). Standardized/unstandardized parameters. †p<.10; *p<.05; **p<.01; ***p<.001.
Adolescent perceptions of mothers' contributions. Findings indicated that the LTM based on adolescent perceptions of mothers' contributions to interparental discord provided a good fit to the data (χ2[1, n=807]=1.99, p=.16; TLI=.995; CFI=.998; RMSEA=.04; SRMR=.01). With regard to specific parameters (see Figure 1), there was a statistically significant increase in adolescent perceptions of mothers' contributions to interparental discord from 7th to 9th grades. Results also revealed a significant degree of variability associated with this estimate. These findings indicate that although on average there was growth in the extent to which adolescents perceived that their mothers contributed to interparental discord, individual differences related to this increase were also evident. A nonsignificant, but notable in terms of effect size, positive correlation was found between the intercept and slope latent factors. Thus, adolescents with higher initial levels of perceived interparental discord tended to increase at steeper rates (i.e., have more positive slopes) compared with those with lower initial levels.
Adolescent perceptions of fathers' contributions. As can be seen in Figure 1, results of the LTM using adolescent perceptions of fathers' contributions to interparental discord were similar to those reported for mothers' contributions. Specifically, the LTM for fathers provided a reasonable fit to the data (χ2 [1, n=807]=7.97, p<.01; TLI=.94; CFI=.98; RMSEA=.09; SRMR=.02). Similar to the pattern of change for mothers, a statistically significant increase in adolescent perceptions of fathers' contributions to interparental discord from 7th to 9th grades was observed (see Figure 1). There also was significant variability associated with the slope. A nonsignificant correlation was found between the intercept and slope latent factors.
Follow-up analyses were conducted to examine associations between intercept and slope estimates obtained from latent trajectory analyses using adolescent perceptions of mothers' compared with fathers' contributions to interparental discord. To examine these relations, we exported the factor scores for latent variables representing initial levels and slopes to a general statistics software package and then conducted the correlation analyses. Findings indicated that there was a significant positive association between the intercepts corresponding to adolescent perceptions of mothers' and fathers' contributions to interparental discord (r=.69, p<.001). Similarly, a large positive correlation was found between the slope latent factors for mother- and father-related data (r=.63, p<.001).
Adolescent Perceptions of Interparental Discord and Adolescent Negative Mood
To examine the second aim of the study (i.e., whether change in adolescent perceptions of interparental discord was related to change in adolescent negative mood) and the third aim (i.e., whether boys' and girls' perceptions of the contributions of their mothers compared with their fathers changed at different rates) the LTM estimated in previous analyses was extended (see 2, 3). Specifically, a directional path was added from the slope latent variable (representing change in adolescent perceptions of interparental discord from 7th to 9th grades) to the measure of adolescent negative mood in 10th grade (Aim 2). To control for initial levels of adolescent negative mood, a path from 7th grade negative mood to 10th grade negative mood was estimated. A path also was included from the intercept latent factor to the 10th grade measure of adolescent negative mood to control for the influence of initial level of adolescent perceptions of interparental discord on mood. As noted, there was a nonzero (although not statistically significant) relation between initial level of interparental discord (intercept) and change in interparental discord (slope) for models related to both mothers and fathers. To account for this association, a directional path between the intercept latent factor and slope latent factor was added. These additional paths provided a conservative test of whether change in adolescent perceptions of interparental discord predicted adolescent negative mood by statistically controlling for the influence of initial level of adolescent perceptions of interparental discord on both the rate of change and negative mood.
Latent trajectory model based on adolescent perceptions of mothers' contributions to interparental discord (n=812). Boys=0; Girls=1. Standardized/unstandardized parameters. **p<.01; ***p<.001.
Latent trajectory model based on adolescent perceptions of fathers' contributions to interparental discord (n=812). Boys=0; Girls=1. Standardized/unstandardized parameters. **p<.01; ***p<.001.
A path was included between adolescent biological sex and the slope latent variable to examine whether the rate of change in adolescent perceptions of interparental discord differed between boys and girls (Aim 3). Given that research indicates there are sex differences regarding adolescent mood (e.g., Weinstein, Mermelstein, Hankin, Hedeker, & Flay, 2007), a path from biological sex to 10th grade adolescent negative mood was estimated. A covariance also was estimated between adolescent biological sex and the adolescent perceptions of interparental discord intercept. Finally, covariances were estimated between 7th grade adolescent negative mood and both adolescent biological sex and the intercept latent factor. As in previous analyses, separate models were estimated for adolescent perceptions of mothers' and fathers' contributions to interparental discord.
Model using adolescent perceptions of mothers. Findings indicated that the LTM designed to examine the relation between change in adolescent perceptions of mothers' contributions to interparental discord from 7th to 9th grades and change in adolescent negative mood from 7th grade to 10th grade provided a good fit to the data (χ2 [5, n=812]=11.36, p<.05; TLI=.98; CFI=.99; RMSEA=.04; SRMR=.02). Several specific parameter estimates should be highlighted (see Figure 2). Most notable in terms of study aims, the adolescent perceptions of interparental discord slope was a unique predictor of 10th grade adolescent negative mood (Aim 2) after controlling for 7th grade adolescent negative mood. Thus, the largest increases in adolescent perceptions of mothers' contributions to interparental discord from 7th to 9th grades were associated with the largest increases in adolescent negative mood from 7th grade to 10th grade. It should be recalled that this was a conservative test given that initial level of adolescent perceptions of interparental discord and biological sex were statistically controlled. With regard to other findings, a nonsignificant relation was found between adolescent biological sex and the adolescent perceptions of interparental discord slope factor (Aim 3). In addition, a nonsignificant association was found between initial levels of adolescent perceptions of interparental discord and change in adolescent perceptions of interparental discord. Adolescent biological sex was not related to 7th grade adolescent negative mood.
Model using adolescent perceptions of fathers. Results revealed that the LTM designed to investigate the relation between change in adolescent perceptions of fathers' contributions to interparental discord and change in adolescent negative mood provided a reasonable fit to the data (χ2 [5, n=812]=13.50, p<.05; TLI=.97; CFI=.99; RMSEA=.05; SRMR=.02). Findings based on father-related data were similar to those in which mother-related data were used (see Figure 3). Specifically, the adolescent perceptions of interparental discord slope was a unique predictor of adolescent negative mood reported in 10th grade (Aim 2) even after controlling for adolescent negative mood reported in 7th grade. Initial levels of adolescent perceptions of interparental discord and adolescent biological sex were additional positive predictors of adolescent negative mood reported in 10th grade. Results also revealed that adolescent biological sex was related to the adolescent perceptions of interparental discord slope (Aim 3). This suggests that girls tended to increase at a steeper rate than boys in terms of their perceptions of their fathers' contributions to interparental discord. Similar to the model related to adolescent perceptions of mothers, adolescent biological sex was not related to 7th grade adolescent negative mood.
DISCUSSION
The present research was designed to further understanding of the complex relation between interparental discord and adolescent adjustment. One potentially important question that has received little research attention in this regard is the extent to which change in interparental discord over time contributes to adolescent adjustment difficulties (Cui et al., 2005). The purpose of the current investigation was to address this gap by examining whether change in adolescent perceptions of interparental discord from early to middle adolescence was related to change in adolescent negative mood. Our study extends earlier work by (a) focusing on change in adolescent perceptions of interparental discord, (b) considering change in adolescents' perceptions of their mothers' and fathers' contributions to interparental discord separately, and (c) using a large, nationally representative sample of adolescents.
Consistent with previous longitudinal research with couples (e.g., Kurdek, 1999; Umberson, Williams, Powers, Liu, & Needham, 2005), findings indicated that on average adolescent perceptions of interparental difficulties increased from early to middle adolescence (7th–9th grades). These results build on prior research by showing that adolescent perceptions of interparental discord follow a similar increasing trajectory compared with parent reports of interparental discord (Cui et al., 2005). As discussed later, changes in adolescents' perceptions of interparental discord may have important implications for their emotional adjustment.
This study also examined whether adolescents' perceptions of their mothers' and fathers' contributions to interparental discord changed at different rates. Results revealed that perceptions of mothers and fathers followed similar trajectories (i.e., increased at similar rates) from early to middle adolescence. Results further indicated that there was a strong positive association between the estimated patterns of change corresponding to adolescent perceptions of mothers' and fathers' contributions to interparental discord (i.e., correlation between slope latent factors). Together, these findings add to the growing literature supporting family-systems views of interparental discord which highlight the reciprocal, bidirectional nature of couples' interaction patterns (Gottman, 1994; see also, Ehrensaft, Moffitt, & Caspi, 2004).
A measurement issue should be considered when interpreting similarities found between patterns of change concerning adolescent perceptions of mothers' and fathers' contributions to interparental discord. In the current study, as in a large proportion of research investigating change in marital relationship functioning, we used a broad measure of interparental discord that may have masked important differences between adolescent perceptions of mothers' and fathers' conflict strategies. It is possible that differences between mothers and fathers would emerge for adolescent perceptions as well as for individual parent reports of relationship problems if process-oriented components of relationship problems were examined. For example, research indicates that among American couples women are more likely to engage in demand-related behaviors such as emotional demands, criticism, and complaints, whereas men are more likely to engage in withdrawal-related behaviors such as defensiveness and passive inaction (e.g., Christensen & Shenk, 1991; Gottman, 1994; Rehman & Holtzworth-Munroe, 2006). Differences between mothers and fathers in terms of change in more process-level interparental relationship factors may have important implications for adolescent emotional functioning. Adolescents may find one style more aversive than another, and these perceptions may directly create a negative mood state for the adolescent or contribute indirectly to negative mood by creating distance or conflict in the parent–child relationship.
The second aim of the present study was to investigate whether change in adolescent perceptions of interparental difficulties was related to change in adolescent negative mood. Findings indicated that the largest increases in adolescent perceptions of interparental discord from 7th to 9th grades were associated with the largest increases in adolescent negative mood from 7th grade to 10th grade. This pattern was found for adolescent perceptions of both mothers' and fathers' contributions to interparental discord. It is important to note that these relations were evident even after statistically controlling for initial level of adolescent perceptions of interparental discord and adolescent sex. Thus, the present analyses provided a conservative test of the association between change in adolescent perceptions of interparental discord and change in adolescent negative mood. Results indicating that change in adolescent perceptions of interparental discord was related to adolescent emotional adjustment are consistent with findings reported by Cui et al. (2005). Together, these studies highlight the importance of considering how change in interparental discord over the course of development can influence adolescent emotional adjustment.
According to the sensitization hypothesis (Davies et al., 2006), youth are expected to become more responsive to interparental discord when exposed to more problems between parents over time. Because studies have generally only examined interparental discord at one point in time, it has not been established that increases in exposure to interparental discord place youth at increased risk for emotional difficulties. Findings from this study thus provide unique support for the sensitization hypothesis.
The sensitization hypothesis provides an explanation for why youth who experience increases in interparental conflict also may experience increases in negative mood over time. Earlier research has documented that youth who are exposed to high levels of interparental discord develop heightened sensitivity to discord, seen in their behavioral, emotional, and cognitive responses to interadult conflict (Cummings et al., 2000). This sensitivity can be problematic in its own right (e.g., creating worry and physiological distress) and it may also interfere with the youth's ability to engage in adaptive, age-appropriate behaviors (such as engaging in a hobby rather than attending to the parents' discord) that would promote adjustment. Although results of the current study are consistent with this hypothesis, it will be important for future work to consider whether the association between increases in interparental discord and increases in adolescent negative mood is in fact mediated by behaviors that reflect sensitization (e.g., behavioral involvement in interparental conflict; youth's appraisal of self-blame; youth's worry about their parents' well-being).
A third aim of the study was to examine whether there were sex differences associated with patterns of change in adolescent perceptions of interparental discord. Findings indicated that girls' perceptions of their fathers' contributions to interparental discord tended to increase at a steeper rate than boys' perceptions. Adolescent sex differences were not found for perceptions of mothers' contributions to discord. This pattern of results can be interpreted in the context of prior research indicating that in families experiencing high levels of interparental discord, fathers may be more negative toward daughters compared with sons, whereas mothers may treat boys and girls similarly (for discussion, see Snyder, 1998). The steeper increase in perceptions of fathers' contributions to interparental discord for girls could be a function of daughters' negative views of the father–daughter relationship (for discussion, see Cummings et al., 2000). Our findings highlight the potential importance of considering the match between the sex of the parent and child when investigating the relation between change in interparental discord and youth adjustment.
Several limitations of the present study along with additional directions for future research should be noted. First, parent-report data are only available at Wave 1 of the NLSY97 data set. Given the value of multi-informant data for the comprehensive assessment of youth functioning (Achenbach, 1992), it will be important for future research investigating change in interparental discord to include information from parents as well as from adolescents. Second, the same questions assessing adolescent emotional functioning were not asked at each assessment period, and only one item was available that assessed adolescent negative mood at Wave 1. To enhance understanding of the relation between change in interparental discord and change in adolescent adjustment, future research should investigate change in adolescent negative mood using the same established measure. Finally, our sample was limited to couples who remained together throughout the course of the study. It is possible that families in which an adolescent perceived either their mother or father as contributing more to the interparental problems at Wave 1 were more likely to divorce or separate by later assessment points. Additional studies are needed to examine this possibility as well as patterns of change in the interparental relationship for other types of family compositions (e.g., divorced couples).
An advantage of latent trajectory analysis is that in addition to estimating the average level of change, variability (individual differences) in change patterns also can be identified (Karney & Bradbury, 1995). Our results indicated that there was significant variability in the patterns of change in adolescents' perceptions of both their mothers' and fathers' contributions to interparental discord. This means that although perceptions of interparental difficulties increased on average, some adolescents' perceptions of parents' conflict behaviors remained relatively stable and others' attributions decreased. To enhance understanding of change in adolescent perceptions of interparental discord, it will be important for future research to identify factors that contribute to individual differences in the course of interparental relationship functioning.
Our results suggest that there are some adolescents who are at elevated risk for experiencing increased levels of mood-related difficulties in part because of increased problems in their parents' relationship. These findings argue for the potential benefits of directly targeting interparental relationship problems in interventions designed to reduce adolescent emotional difficulties (e.g., Henggeler, Schoenwald, Rowland, & Cunningham, 2002; Kelley & Fals-Stewart, 2002). It is likely that interventions designed to change the increasing course of interparental discord will be of critical importance. Promising in this regard is that Kelley and Fals-Stewart (2002) found that a treatment designed to improve marital functioning led to reductions in youth emotional and behavioral problems. To build on these encouraging results, additional work is needed to examine whether interventions that directly target interparental relationship problems lead to short- and long-term reductions in youth emotional problems.
