Might require some investigation and modeling to figure out a good cause. For example, the stats based on drunk-driving incidents compared to total might be mi

Might require some investigation and modeling to figure out a good cause. For example, the stats based on drunk-driving incidents compared to total might be misleading as the totals might vary significantly.. we'd probably want something more like [drunk-driving-incidents]-per-[driving-hour] for a starting point. Then probably explore stuff how frequently police catch people and typical consequences, how much folks in an area drink, how people drink (late nights at bars, or earlier at dinner?), if people in an area tend to car-pool, how the roadways might play in (safety features and well lit?; twisty backroads?), etc.. Really seems like there should be research on this sorta thing already, but not immediately seeing one on Google, which is odd.. But, yeah, dunno why folks'd do it in the first place. Seems like simple self-destructive behavior. --- [This study](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3823314/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3823314/) seems to use a multi-linear-regression (rather than a causative model), but still looks neat. Overall, looks like they found that stricter regulations and beer-taxes significantly reduced drunk-driving incidents, while drivers-under-24, higher speed-limits, and unemployment had the opposite effect.