Even assuming that data is good, its an open question how much of that is causal, how much if it is "people who would use hard drugs anyway tend to stop by marijuana on the way because its more accessible", and how much of it is "people who develop relationships with illegal drug dealers through marijuana are likely to be pushed by them to other drugs at some point".
The first is exacerbated by legalization, the second indifferent to it, and the third mitigated by it, so simple correlation data doesn't really provide useful information for policy without some clear indication of why the correlation exists.
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.
The data for oxygen is far more alarming.
100% of hard drug users users first started out on oxygen.
Furthermore, if nobody was allowed oxygen, all illegal drug addiction would stop.
Oxygen is the ultimate gateway drug.
As a note, I remember an article somewhere about how it was meth the drug that gave more profit to drug mafias. So it'd still be a problem. The question is, bigger or smaller? I'm pretty sure there's data of the translation from marijuana users->hard drugs users.