People who study human behavior in societies need to worry about lag times and lead times. That comment results from this piece on the results of legalizing pot in Washington--most studies show little change in teenage pot use.
I'm not convinced, perhaps because I've a vivid memory of a high school teacher warning against the dangers of marijuana, probably in 1958 or so. But over the last 50 years there have been "epidemics" of use/abuse of various substances, most notably the "crack" epidemic in the 1980's. That seems to have settled down, perhaps because kids saw the adverse impacts of crack addiction and decided to avoid it. There have been others--the "date drug" scare, for one.
Such epidemics are, I think, very much like epidemics of physical disease: the flu, SARS, HIV, Ebola. The initial cases don't show up in summary statistics; there's a lead time for the disease to spread to the point where it will show up. The necessity of a lead time means there's a lag time in seeing its effects.
So I'm not convinced by a few years experience in one state.