Background: The present study examined dimensionality of oppositional defiant disorder (ODD) using 10 alternative items using network analysis and confirmatory factor analysis. Methods: The sample constituted 551 Norwegian children aged 2-12 and their parents. We used network analysis to investigate the connections between different symptoms. Next, we analysed data using traditional confirmatory factor methods, including the more recently proposed bifactor approach. Results: The bifactor model, with a strong general factor and three specific factors, provided the best model fit. Omega values did, however, reveal that only the general and the specific hurtful factor had satisfactory reliability. Network analysis showed that symptoms in general were positively connected within the ODD network. Strong connections between several symptoms within the irritable and hurtful cluster emerged, whereas some symptoms of the headstrong cluster seem to function as bridge nodes between the irritable and hurtful symptom clusters. Conclusions: The findings support a bifactor model of ODD indicators, but omegas only gave support to the use of a general latent factor, and one specific factor. Network analysis did, however, provide some additional and interesting findings, revealing clusters of strongly connected symptoms and central bridge node symptoms. Implications of the results are discussed.
Key Practitioner Message• Oppositional Defiant Disorder (ODD) is a leading cause of referral for youth mental health problems. • Findings organize ODD symptoms into three dimensions reflecting irritable, headstrong, and hurtful behavior.• The present study evaluated ten alternative items reflecting ODD, using explorative network analysis and confirmatory factor analysis.• Network analysis represents an alternative to the latent reflective approach, as it assumes that ODD constitutes a network of symptoms that associate and interact with each other.• Factor analysis showed that the bifactor model provided best fit to the data, but omega reliabilities only supported the use of a general ODD factor, not the distinct factors.• Network analysis revealed clusters of symptoms reflecting irritability and hurtful, but to a lesser extent headstrong.