A Multilevel Mixed Effects Model to Evaluate Effectiveness of Treatment for Problem Gambling

Abstract

Problem gambling treatment is a challenging present-day topic. This paper proposes an empirical model for evaluating the effectiveness of treatment for problem gambling using a sample of problem gamblers treated by a set of Spanish associations dedicated to gambling addiction issues. Data consists of multiple levels of nested groups (individuals and problem gambling recovery centres). A multi-level, mixed-effects logistic regression is used which permits controlling for unobserved heterogeneity across different problem gambling associations. The results seem to indicate that individual aspects such as age, family history, marital status or work situation, but also behavioural factors (previous dropouts, relapses during treatment, or consumption of other substances) are found to affect the effectiveness of treatment for gambling disorders. The analysis of the predictors for treatment efficacy may help treatment techniques to be adapted depending on the characteristics of individual patients and to evaluate programmes designed by treatment centres.