Modelling Predictors of Gambling Harms with Stochastic Search Variable Selection (SSVS)

Abstract

The purpose of the present study was to introduce stochastic search variable selection (SSVS) as a procedure to identify a subset of important predictors of gambling harm. The target set of predictors were dimensions of trait impulsivity, gambling cognitions, and gambling motivations. Five types of gambling harm (feeling one has a personal problem; social criticism; feeling guilt; health; and, financial) were measured by the Problem Gambling Severity Index. Casino patrons completed the measures. As a first step, we identified the significant predictors that would be included in modelling an aggregate harm score. The most important predictors, the cognition that one is not able to stop gambling, and the motivation to escape or avoid life stressors, were positively associated with overall harm. Two weaker, but statistically significant, predictors were negatively associated with harm: sensation-seeking and illusion of control. Although a perceived inability to stop gambling was the most important predictor of each individual harm, the pattern of predictors varied across harms. For example, sensation-seeking was an important predictor only for the belief that one has a gambling problem, and escape/avoidance motivation was strongly predictive of financial harm. The results suggest that primary interventions designed to mitigate harm should address the belief that the gambler is unable to stop gambling, and motivations related to escape/avoid life stressors. Other interventions would be tailored to the specific harms experience by the gambler.