BACKGROUND
Studies have found an association between problem gambling and poverty. However, there is relatively little research on social inequalities and problem gambling using population representative data.

METHODS
A population-representative self-report web-based and postal survey with register-based linkage was conducted in the three geographical areas of Finland. Participants (n = 7186, aged 18 or older) were randomly selected from the population register. Sociodemographic factors and social welfare benefits were studied among gambling groups and their statistical difference were examined by χ2 test. Seven logistic regression models were calculated, where unemployment, social security benefits and low income were treated as dependent variables and where sex, age, family structure and education were controlled as covariates. The results were presented as odds ratios (OR) with 95% confidence intervals (CIs).

RESULTS
Problem and at-risk gambling (ARG) was more common among people who were unemployed [PG: χ2=6.4 (1), P < 0.01, ARG: χ2=12.4 (1), P < 0.001] or had received social security benefits [PG: χ2=41.6 (1), P < 0.001, ARG: χ2=22.9 (1), P < 0.001]. The OR for problem gambling was high as 5.6 (CI: 3.22-9.61) among respondents who had received social assistance even when covariates were taking into count. Almost a third of those experiencing problem or at-risk gambling received at least one form of social security benefit.

CONCLUSIONS
The most important task of gambling policy should be reducing gambling-related harms and diminishing social inequality. However, even in government organized system where gambling profits are used for common good, profits come from the most socially disadvantaged people thereby exacerbating inequality.