Unless you blind this, I’m not sure it’s possible to get past the correlation or causation problem. Weed use is not yet so destigmatized for teens that usage itself is not a marker of deviance (in the math sense, not any kind of judgement).