All analyses were performed using SPSS 13.0 [34]. For each drug, we performed a MANCOVA with user status as the independent variable, personality factors and facets as the dependent variables, and age, sex, education, and ethnicity as covariates. Post-hoc comparisons among never, former, and current users groups were based on LSD estimates. The assumption of homogeneity of variance was tested using the Levene statistic, and no large violations were found. Effect sizes were estimated using partial η2. According to Cohen [35], η2 values of 0.0099, 0.0588 and 0.1379 correspond to small, medium and large effect sizes, respectively.
