Dimensions of Oppositionality: A Latent Class Analysis

Introduction: There has been little agreement on how to classify the documented dimensions of oppositional defiant disorder (ODD), with some defining oppositional behavior, negative affect and antagonistic behavior (Burke et al., 2010), and others identifying irritable, headstrong and hurtful facets (Stringaris & Goodman, 2009). Affective dimensions are associated with internalizing disorders, while behavioral dimensions are linked to externalizing behaviors (Burke et al.; Stringaris & Goodman). The goal of the current study was to determine which classification of oppositionality was supported in a diverse, community sample of adolescents. We also examined if the affective dimensions predicted later depression and if the behavioral dimensions were related to concurrent callous-unemotional traits.

Methods: Data were collected from 521 adolescents (52.2% male; 44.5% White, 27.4% Black, 16.9% Asian, 10.2% Hispanic) as part of a longitudinal epidemiological study. Diagnostic information regarding ODD and depression was collected annually from the Diagnostic Interview Schedule for Children (Shaffer, 2000) in 6^th through 9^th grades. Callous-unemotional traits were measured with the Antisocial Process Screening Device (Frick, 2001) in 6^thgrade.

Results: Confirmatory factor analyses (CFAs) and latent class analyses (LCAs) were conducted in Mplus version 7 (Muthén & Muthén, 2012). The 12-month prevalence of ODD in the sample was 15.1% in 6^th grade. CFAs at 6^th grade provided support for Stringaris and Goodman’s 3-factor model, χ² (51) = 76.12, p < .05; CFI = .99, TLI = .99, RMSEA = .03. Based on LCAs at 6^th grade, a three-group solution provided the most parsimonious fit. Class 1 (n = 120) was high on all items (oppositional), Class 2 (n = 150) was low on all items (well-adjusted), and Class 3 (n = 239) was high on affective but low on behavioral items (moody). Logit coefficients indicated that the oppositional group (.44, p < .001), but not the moody group (.22, p = .06), demonstrated more callous-unemotional traits than the well-adjusted group. Based on multinomial logistic regressions, membership in the moody group, but not the oppositional or well-adjusted groups, predicted later depression diagnoses (b_{7th grade} = 17.43, p < .001; b_{8th grade} = 17.50, p < .001; b_{9th grade} = 17.82, p< .001).

Conclusions: Given the high prevalence of ODD as well as reconceptualizations by the DSM-5 workgroups (i.e., 3 categories of ODD), it is important to understand the structure and associated comorbidities of this disorder. Findings suggest that different interventions might be considered depending on the primary dimension exhibited by the adolescent (e.g., treatments targeting depression might be effective for adolescents who fit into the moody dimension of ODD).