When individuals without criterion endorsement were excluded, factor analyses (CFA) resulted in higher heterogeneity (more factors). The reduction of homogeneity in the RS was related to a reduction of inter-item correlations. The correlation matrix revealed that abuse criteria had more positive correlations among each other than with dependence criteria. It can be assumed that the inclusion of individuals with no endorsed criteria in a sample “deteriorates” the covariance matrix and, therefore, the factor structure. This may be the reason why previous studies using general population samples did not find abuse and dependence as separable factors