Does Cannabis Legalization Really Lead To More Traffic Deaths?

One of the most infamous anti-legalization talking points is that legalizing cannabis would inevitably cause carnage on the highways.  And we have thoroughly debunked that specious claim before.

But what about the recent pair of newer studies that appear to find a modest uptick in traffic deaths after legalization?  In contrast to previous studies that generally found no link with traffic casualities in the first three years after recreational legalization, one of the newer ones found an increase in traffic fatalities in the most recent year in which data were available, though still no shorter-term increase (echoing another study from last year that didn't find any uptick until five years later), while the second one found an increase (compared to synthetic controls) in Colorado but not in Washington for some reason.

Far from clarifying the issue, these newer studies leave the reader with more questions than answers.  Why so much heterogeneity and inconsistency between various studies and locations?  Why such a long time lag for the apparent effect to occur, especially given that repealing alcohol Prohibition in 1933 was associated with short-term increase in traffic fatalities per VMT that went back down to 1930-1931 levels by 1936?  And why have medical cannabis legalization laws been consistently associated with long-term decreases in traffic deaths, given the increased availability and vast gray area between recreational and medical use in practice?

Three things come to mind:  changes in tourism (thus skewing the numerator but not the denominator in per-capita crash death calculations), changes in gas prices (lower prices lead to more crashes and deaths), and changes in cannabis prices (lower prices leading to more use and possibly more stoned driving).  The first two can produce specious and spurious inferences when they are not controlled for, while the third factor as we have seen takes several years for prices to fall after recreational legalization, possibly explaining the apparent time lag with fatalities.  (That can, of course, be resolved simply by raising the taxes on cannabis.)  Or most likely of all, as per Occam's Razor, the vast inconsistencies simply mean that any supposed causal link between legalization and carnage on the highways was spurious all along.

Also, it is notable there does not seem to be such an effect seen in Canada, despite their lower age limits for cannabis (18 or 19, instead of 21 in the US states that legalized).  While Quebec recently raised it to 21 on January 1, 2020, there would still be over a year's worth of data for 2019 and late 2018 when it was still 18.  And Alberta's age limit still remains 18 for all three substances--alcohol, tobacco, and cannabis.

Thus, as per the overall weight of the evidence, Twenty-One Debunked will continue to declare this specious claim debunked for the time being.