In the final stage of the analyses we explored whether there were separable symptom dimensions within the ODD symptoms. In an exploratory factor analysis of ODD symptoms in the full sample at Wave 1, a two factor solution provided an acceptable fit to the data (RMSEA=.021). Computation of a three factor model failed, showing a negative estimated residual variance for the spiteful symptom which may indicate that too many factors had been specified. Factor loadings from the two factor model are shown in Table 4, following Promax rotation. The two factors correlated at .55, and high loading items were similar to the irritable and headstrong dimensions proposed by Stringaris and Goodman (2009b); the symptom of spitefulness loaded clearly on the headstrong factor in this solution.
