Crowdsourcing is an increasingly popular source of participants in studies of problem gambling. Studies with crowdsourced samples have reported prevalence rates of problem gambling between 10 to 50 times higher than traditional sources of estimates. These elevated rates may be due to study framing motivating self-selection. In this preregistered study, we examined whether study framing influences self-reported problem gambling severity and harmful alcohol use in a sample of participants recruited from a popular crowdsourcing website.
Method: Two recruitment notices for an online questionnaire were placed on Amazon Mechanical Turk (MTurk). Notices were framed as “Gambling and Health” or “Alcohol and Health”. Only participants who passed data checks were retained for confirmatory analyses (N = 564; 44% of recruited participants). Participants in the gambling framing (N = 261) and in the alcohol framing (N = 303) were compared on the Problem Gambling Severity Index (PGSI) and Alcohol Use Disorders Identification Test (AUDIT).
Results: Problem gambling rates and severity scores were significantly greater for participants in the gambling framing compared to those in the alcohol framing. Self-reported scores of harmful alcohol use were significantly greater for participants in the alcohol framing compared to those in the gambling framing, but there was no significant difference in prevalence rates for harmful alcohol use.
Conclusions: Study framing is an important consideration for gambling and alcohol research. We found that study framing may substantially increase the observed rates of problem gambling severity in crowdsourced samples, potentially via encouragement of self-selection.