> The one I've focused on first that you seem so keen to dispel, is an assumption that there are smart kids overcoming a challenge. 'Roy' is an extreme, but there is a whole spectrum of low-oversight conditions that are likely to lead to kids getting access to alcohol, tobacco, drugs, having sex etc, which are nothing to do with smartness or challenges and are much more to do with shitty parenting.
Let's consider the four combinations of the two variables here. You have dumber and smarter kids, and worse and better parents. The kids with the worse parents will have access to the vice regardless of whether they're dumb or smart, but the kids with the better parents will only have access if they're smart enough to figure out how against parents actively trying to prevent it. Therefore the two of the four quadrants with smarter kids can get access but the dumber kids only can when they have worse parents, implying that two thirds of the quadrants with the ability to do it are the smarter kids.
> even if we allow that children getting access to tobacco are 'smarter' than those who don't figure it out, and are overcoming various obstacles, that doesn't actually imply that they'll be better able to deal with the consequences.
That's assuming the way they deal with it better is by trying the drug and then somehow not getting addicted rather than by choosing not to try the drug to begin with even though they could access it if they wanted to, or otherwise making more measured choices if they do decide to try something, like finding a source more likely to be providing the expected amount of the expected substance instead of who knows how much of who knows what. Or just hesitating a while so their first time comes at an older age.