After adjusting for sex and age, living in a state with legal sports betting is associated with higher probability of having mentally unhealthy days, especially among adults aged 25–44. The effect size is around 5–8 percentage points, depending on age group. Men and women are affected similarly, which doesn’t align with the Couture et al. findings, likely because we were looking at a larger dataset which included post-COVID-19 observations. Younger adults (18–24) have the highest baseline mental health risk, but legalization effects are somewhat stronger for those 25–44. These associations cannot establish causality (due to cross-sectional BRFSS data), but they indicate a robust correlation.
We are not following the same participants over time which prevents establishing causal relationships. We cannot determine whether the legalization of sports betting directly contributed to worsening health outcomes. Also, the measure of health outcomes are self-reported which may introduce recall bias and response bias. Respondents may inaccurately recall their health over the past 30 days and under-report sensitive behaviors due to stigma.
The study period includes the COVID-19 pandemic which potentially
serves as a major confounding event. Changes in health outcomes related
to the pandemic may coincide with legalization, confounding
associations. Our exposure variable, sb_legal does not
account for individual sports betting behavior or the level/intensity of
exposure, therefore, which only allowed us to address population-level
associations based on legalization exposure.
There are many additional analyses that could be done on the dataset, given more time. This includes finely tuning the regression model and addressing confounding factors, like COVID. We observed that mental health outcomes decreased overall from 2017-2024, but establishing a causal effect of sports betting would require further analysis. We did not get a chance to look at the effect of online vs. retail-only sports betting legalization (which varies state-to-state), but this would be another interesting policy effect to examine.