Separating Behaviours and Adverse Consequences in the Problem Gambling Severity Index (PGSI): A Confirmatory Factor Analysis and Rasch Analysis

Abstract

The Problem Gambling Severity Index (PGSI) is considered the “gold standard” for measuring problem gambling. The PGSI provides a single score summed across nine items. The nine items of the PGSI comprise two subdomains: problematic behaviours and adverse consequences. The aim of the present study was to compare evidence of a one-factor structure to evidence of a two-factor structure representing the two subdomains. With a sample of 1,251 bettors, we conducted confirmatory factor analyses and Rasch analyses to assess evidence supporting the one-factor and the two-factor structures. In addition, stochastic search variable selection was conducted with the total PGSI score, PGSI behaviour score, and PGSI adverse consequences score as separate outcomes to examine whether information is lost when collapsing the two subdomains into a single factor. Overall, there was stronger support of a two-factor structure than a one-factor structure. However, the two-factors were highly correlated with one another and shared most predictors except for one. We recommend continued use of the one-factor structure of the PGSI unless one aims to better understand the relationship between problematic behaviours and adverse consequences.