Latent class analysis (LCA) was applied to test whether groups of individuals with similar response patterns exist. The optimal number of classes was determined based on the Bayesian information criterion (BIC) (Nylund, Asparouhov, & Muthén, 2007). Analyses were performed until a relative minimum of the BIC value was identified. For graphical presentation, the estimated criterion probabilities for the seven dependence criteria and the four abuse criteria were merged. Expectation values representing the expected sum score of abuse (EA) and dependence (ED) criteria in each class were based on the number of criteria (four for abuse; seven for dependence) multiplied by the mean estimated probability per class