Before applying a cross-lagged model to the data, we examined the unidimensionality of the alcohol use and antisocial behavior items at each occasion using exploratory factor analysis (EFA). We also examined the mean and covariance structure of the alcohol use and antisocial behavior factors by fitting a confirmatory factor analysis (CFA) with unstructured factor means and covariance matrix. This model allows the estimation of the mean and covariance of factors of alcohol use and antisocial behavior with the minimal number of parameter constraints that is required to identify the model (Mehta et al., 2004; Millsap and Yun-Tein, 2004). We then fit cross-lagged models to examine the directional relationship between the 2 behaviors.
