Region & Target Population
- ‍Region: Canada (nationwide sample)‍
- Target population: Adult gamblers 18+ who gambled monthly or more in the past year (panel-screened).
Study Design
- Secondary analysis of a national online panel survey with a baseline and a follow-up
Sample Characteristics (with data-collection years)
- ‍Baseline: N = 10,199 Canadian gamblers (screened for monthly+ gambling)‍
- Follow-up: n = 4,707 re-contacted respondents (from baseline cohort)
- The sample included a large subgroup classified as problem/pathological gambling at baseline (used for “self-reported problematic types/modalities” analyses).68
Cross-National Structure & Comparison Logic
- The study’s “comparison” logic is within-country: it compares individual gambling types and modalities against each other, and then tests whether those associations hold once your account for:‍
- Breadth of involvement (how many formats someone plays), and‍
- Overall intensity (frequency/time/expenditure).
Measures Used
- ‍Gambling participation captured with the Gambling Participation Instrument (GPI) (past-year types engaged in, provider/modality, frequency, time, expenditure, etc.).‍
- Problem gambling classification (study uses a classification approach and reports categories including recreational, at-risk, and problem/pathological).‍
- Self-report item for “problematic types/modalities” among those with problem gambling (participants indicated whether particular types contributed more than others, and identified which ones; they also indicated whether problems were mostly land-based.
Research Questions
- Which specific types and modalities show strong univariate associations with problem gambling?
- Do those associations remain once you control for breadth of gambling involvement?
- Do particular types/modalities predict future problem gambling (prospective/lagged prediction)?
- Among people with gambling problems, which types/modalities are self-identified as most problematic?
Key Findings
- ‍Breadth matters more than any single product: Across analyses, breadth of gambling involvement emerged as a stronger predictor of gambling problems than participation in any one format, supporting the idea that multi-format engagement is a major risk marker.
- ‍But product-type still matters (especially EGMs): Even after accounting for breadth and intensity, the results converged on electronic gambling machines (EGMs) as the most robust “riskier type,” with casino table games and online gambling also implicated (though less strongly than EGMs).
- Self-reported “what caused my problem” points strongly to EGMs: Across multiple population surveys discussed in the paper, about 45% of people with problem gambling report that a particular type contributed more than others, and EGMs were the most frequently identified problematic type; instant lottery tickets also appeared as problematic in some surveys.‍
- Prospective tests use a stricter approach than simple correlations: For lagged prediction, the authors ran logistic regressions in blocks, first the format, then adding breadth, then adding intensity (frequency/time/expenditure), to test whether a given type predicts later problems above and beyond general involvement.
Study Conclusion
The authors conclude that while “risk” is heavily tied to how broadly and how intensely people gamble, product design still matters: EGMs show the most consistent and robust relationship with problem gambling across univariate, adjusted, and prospective analyses, with casino table games and online gambling also contributing additional risk. They frame these results as directly relevant to regulation and public health strategy, because it supports product-informed harm prevention rather than treating all gambling formats as equally risky.
