> If less than 40% of the population has impairment levels of THC at any given time but 40% of deceased car crash drivers have impairment levels

You're looking at two different populations in this and your other comments, drawing a false equivalence. The study is over a 6 year period, over which 103 people (40%) tested positive for THC. You're saying that because the number of people who self-reported consuming THC in the last year is 20%, that means the result of the study is eye popping and shocking because the number is 40%. But you cannot directly infer elevated risk just because a subgroup has a higher prevalence than the general population without controlling for exposure and confounders. Especially considering what we are talking about is people self-reporting they are criminals.

Moreover, fatal crashes are not randomly distributed across age groups or vehicle types, and younger people, because they are not as experienced, they drive more often, in smaller cars with fewer safety features, are more likely both to smoke THC, and die in crashes even while sober. So there's a strong sampling bias here you're not accounting for.

And this isn't downplaying the results, it's pointing out its limitations of the study and warning you not to read into it what isn't there. You seem to be shocked by the results which should cause you to dig deeper into the study. I would say the most surprising thing here is they found nothing changed before and after legalization.

> If less than 40% of the population has impairment levels of THC at any given time but 40% of deceased car crash drivers have impairment levels of THC in their blood

Yes, IF. That was my point.