Lost in Definition: Forecasting Gambling Harm When Measures Disagree
Session Title
Public Health: Harm Measurement & Indicators
Presentation Type
Paper Presentation
Start Date
26-5-2026 12:00 AM
Abstract
Existing subjective measures of problem gambling — such as voluntary self-exclusion registers, survey-based screening tools, and behavioral thresholds optimized to predict such survey scores — are widely used in research, industry, and policy. However, these proxies differ significantly in whom they classify as a person experiencing gambling-related harm, overlap only partially, and are often poorly justified as measures of objectively observable harm. Using comprehensive behavioral player tracking data, we document the limited agreement between common measures and demonstrate substantial variation in their out-of-sample forecastability based on past gambling activity. We then introduce a new, well-reasoned measure: the Probabilistic Humane Life Affordability Threshold (PHLAT). PHLAT defines harmful gambling as expenditure that likely forces individuals, given the income distribution, to forgo essential spending needed for a humane life or to incur debt. This approach directly links gambling behavior to affordability and essential needs, avoiding reliance on self-reports or narrow behavioral heuristics. Our findings show that PHLAT identifies a substantially different and, arguably, more policy-relevant group than existing proxies, with important implications for prevalence estimates, prevention strategies, and regulatory design.
Lost in Definition: Forecasting Gambling Harm When Measures Disagree
Existing subjective measures of problem gambling — such as voluntary self-exclusion registers, survey-based screening tools, and behavioral thresholds optimized to predict such survey scores — are widely used in research, industry, and policy. However, these proxies differ significantly in whom they classify as a person experiencing gambling-related harm, overlap only partially, and are often poorly justified as measures of objectively observable harm. Using comprehensive behavioral player tracking data, we document the limited agreement between common measures and demonstrate substantial variation in their out-of-sample forecastability based on past gambling activity. We then introduce a new, well-reasoned measure: the Probabilistic Humane Life Affordability Threshold (PHLAT). PHLAT defines harmful gambling as expenditure that likely forces individuals, given the income distribution, to forgo essential spending needed for a humane life or to incur debt. This approach directly links gambling behavior to affordability and essential needs, avoiding reliance on self-reports or narrow behavioral heuristics. Our findings show that PHLAT identifies a substantially different and, arguably, more policy-relevant group than existing proxies, with important implications for prevalence estimates, prevention strategies, and regulatory design.