Predicting drop out of incarcerated men from a long-term aggression programme

This study aims to use pre-treatment assessment scores to predict the drop out of 103 incarcerated male violent perpetrators undertaking a long-term aggression programme, namely the Cognitive Self Change Programme (CSCP), in six English prisons. A hierarchy of best predictors of attrition in this sample is developed. Results found eight out of the 46 assessment variables analysed had a significant association with treatment drop out. Further to this, discriminant function analysis predicted group membership with 80% accuracy, successfully distinguishing perpetrators who dropped out of the programme from those who completed it. The findings support the use of identifying risk factors pre-treatment to predict drop out and offer a practical way to identify group members who are likely to drop out of the CSCP in addition to identifying markers for programme improvement. The need for further research to increase our understanding of the underlying causal explanations that link specific assessment items to treatment dropout is discussed.