I'm pretty sure there's data of the translation from marijuana users->hard drugs users. It's super hard to control for endogeneity of those variables though, w

I'm pretty sure there's data of the translation from marijuana users->hard drugs users. It's super hard to control for endogeneity of those variables though, which would cast significant doubt on a causal interpretation of what's going on. The fact that marijuana is currently illegal means that there is an associative bias for dealers - they have access to/hang out with harder drugs/more dangerous dealers.