DISENTANGLING THE WEIGHT OF SCHOOL DROPOUT PREDICTORS - A TEST ON 2 LONGITUDINAL SAMPLES

The aims of this study are to identify the most powerful predictors of school dropout and to determine how stable they are over time. Two ge nerations of White French-speaking boys and girls from 12 to 16 years old (n = 791 in 1974, n = 791 in 1985) completed a self-administered q uestionnaire on their psychosocial adjustment at least one year before leaving school. As expected the analyses showed that school, family, behavioral, social, and personality variables could all predict droppi ng out of school in the two samples. Furthermore, these predictors wer e quite stable over time. However statistical improvement measures in logistic regression analyses indicated that school experience variable s (i.e., grade retention, school achievement, school commitment) were the best screening variables Sor potential dropouts. The contribution of other psychosocial variables, even though significant, did not impr ove very much the capacity to identify who will drop out of school. Th e discussion highlights the implications of the findings for secondary prevention and screening practices.