Work

Problems with the Problem Gambling Severity Index (PGSI): Is it fit for public health monitoring of gambling harms? An analysis with Rasch Measurement Theory.

Year: 2026
Type: article
Abstract: OBJECTIVE: The Problem Gambling Severity Index (PGSI) is a widely used tool for measuring gambling-related issues, playing a central role in service delivery and national monitoring, especially in the British context. However, it was developed a quarter of a century ago under clinical paradigms of "problem gambling," raising questions about alignment with contemporary public health approaches. This study evaluates the suitability of PGSI for clinical and population-level monitoring, using Rasch Measurement Theory. METHOD: = 15,716). Metrics assessed included person and item reliability and separation, rating-scale functionality, dimensionality, and relationship between raw and Rasch Measurement Theory scores. RESULTS: As a clinical screen tool for problem gambling, the PGSI is broadly fit for purpose, exhibiting acceptable binary discrimination and high precision at the clinical threshold. However, person reliability (0.19-0.58) was suboptimal for discriminating between low-, medium-, and high-risk categories. For general population samples, the targeting (mean person measure = -3.18 to -2.69 logits) indicated items were too difficult. Dimensionality suggested two constructs: behaviors and impacts. Most items demonstrate item parameter drift in difficulty levels. Response categories "Most of the time" and "Almost always" exhibited redundancy. CONCLUSIONS: While the PGSI functions well for its designed purpose (a clinical screen), limitations in targeting, reliability, and dimensionality undermine effectiveness for public health/national monitoring. Alternative tools are required, aligned with contemporary public health strategies. (PsycInfo Database Record (c) 2026 APA, all rights reserved).
Language: English
Field: Psychology
Open Access status: hybrid
Funder: GambleAware
OpenAlex ID: W7224387831
FWCI:
Cites:
Cited by:
Related to: 10