Associations Between Marital Conflict and Adolescent Conflict Appraisals, Stress Physiology, and Mental Health

Abstract

The goal of the current study was to examine conflict appraisals and diurnal cortisol production as mediators of the robust association between marital conflict and adolescent adjustment problems. Parents reported their marital conflict and were observed engaging in a marital conflict discussion; they also reported adolescent internalizing and externalizing behaviors. Adolescents (n = 105, 52% female, 10-17 years of age) appraised their parents' marital conflict and reported their internalizing and externalizing behaviors. After the laboratory visit, adolescents provided four saliva samples on each of 2 consecutive days to assess diurnal cortisol production. More-negative marital conflict predicted more self-blame for parental conflict, which in turn predicted less robust decreases in cortisol across the day. Further, this flattened cortisol production pattern mediated the relationship between greater self-blame for parental conflict and adolescents' elevated internalizing behaviors. Feeling responsible for parental conflict appears to be particularly damaging in terms of physiological regulation and adjustment, and may therefore be a particularly useful intervention target.

FIGURE 1

Conceptual model guiding the current study.

FIGURE 2

Results of the structural equation model testing the associations between marital conflict, conflict appraisals, and diurnal cortisol production. Note: Standardized estimates for paths are displayed; solid arrows reflect significant paths, whereas dashed arrows reflect nonsignificant paths. The following pathways were modeled but were not displayed in the figure for simplicity: paths from ethnicity (Est. = −.39, p <.01) and family income (Est. = −.18, p = .14) to marital conflict, from wake time to cortisol slope (Est. = .24, p = .10), from dairy consumption to the afternoon cortisol sample (Est. = .28, p <.01), and from pubertal status to cortisol slope (Est. = .27, p <.05), conflict property appraisals (Est. = .35, p <.001), and wake time (Est. = −.17, p <.05). In addition, correlations between each pairwise combination of self-blame, conflict properties, and threat were included (Estimates ranging from .30 to .42, p <.001). Pathways between pubertal status and all other variables were nonsignificant and therefore constrained to 0. Pathways from marital conflict to cortisol level and slope were also non-significant and therefore constrained to 0. RMSEA = root mean square error of approximation; CFI = comparative fit index; SRMR = standardized root mean square residual. *p <.05. **p <.01. ***p <.001.

FIGURE 3

Self-blame (median split used to determine high vs. low groups) and diurnal cortisol production. Note: Error bars represent standard errors.

FIGURE 4

Results of the structural equation model testing the associations between conflict appraisals, diurnal cortisol production, and internalizing behaviors. Note: Standardized estimates for paths are displayed; solid arrows reflect significant paths, whereas dashed arrows reflect nonsignificant paths. The following pathways were modeled but were not displayed in the figure for simplicity: paths from wake time to the cortisol slope (Est. = .30, p <.05) and dairy consumption to the afternoon cortisol sample (Est. = .32, p <.001); correlations between each pairwise combination of self-blame, conflict properties, and threat (Estimates ranging from .41 to .51, p <.001); paths between income and conflict properties (Est. = −.18, p <.01), threat appraisals (Est. = −.13, p <.05), and cortisol slope (Est. = −.36, p <.05); paths between income and other variables were nonsignificant and therefore constrained to 0. RMSEA = root mean square error of approximation; CFI = comparative fit index; SRMR = standardized root mean square residual. *p <.05. **p <.01.

FIGURE 5

Results of the structural equation model testing the associations between conflict appraisals, diurnal cortisol production, and externalizing behaviors. Note: Standardized estimates are displayed; solid arrows reflect significant paths whereas dashed arrows reflect nonsignificant paths. The following pathways were modeled but were not displayed for simplicity: paths from wake time to cortisol slope (Est. = .30, p <.05) and from dairy consumption to the afternoon cortisol sample (Est. = .31, p <.001); correlations between each pairwise combination of self-blame, conflict properties, and threat (Estimates ranging from .41 to .52, p <.001); paths between income and conflict properties (Est. = −.18, p <.01), threat appraisals (Est. = −.13, p <.05), and cortisol slope (Est. = −.36, p <.05); paths between income and other variables were nonsignificant and therefore constrained to 0; correlations between pubertal status and youth ratings of externalizing (Est. = .26, p <.05) and conflict property appraisals (Est. = .25, p <.001); paths between pubertal status and other variables were nonsignificant and therefore constrained to 0. RMSEA = root mean square error of approximation; CFI = comparative fit index; SRMR = standardized root mean square residual. *p <.05. **p <.01.