I think you're wanting to understand what it is about the individual's environment that is an actor in the outcome that's depression for that individual or that group of individuals. So for example, if you would want to look at "Is there something going on in the individual's family context that results in depression or that facilitates or accentuates the opportunity for depression?" or "Is it something about their interacting with their peers, are they being bullied, what does that mean?" or "Is it that this individual lives in such a depressed environment in the neighborhood, they can't leave their homes, they're afraid, what it is?"

So you're no longer looking at what it is about that individual that they're presenting a depression syndrome. And so ordinarily you would look, "What's the individual doing? Is there something physiologically wrong?" But then we're saying, "In addition to that, what about the milieu they're in, their contextual milieu? Is that an actor? And what do we know about that?" So it's asking a broader question. It's seeing the individual, it's looking at the individual interacting with his or her environment and is that resulting, is that protecting, is that a risk factor, is that doing something to enhance or promote this behavior in the individual?

It comes out of, if you think of it, the early Bronfenbrenner ecological model, and this is an expansion of it. You can sort of look at Bronfenbrenner's work and then fast forward to George Kaplan's work in 2000 where he showed you the levels of things that impact an individual's life. So it can go from genes and biology all the way up to the family, peers, community, the broader social, structural, geopolitical and all those broader big macro pieces. So it's really taking a different perspective as opposed to just the individual and this individual is in a vacuum, but seeing the individual not in a vacuum but as part of a constellation of interactions.

And it has challenges because you're moving to multi-level analyses, and so there are, methodologically we've made strides in HLM and all the ways of modeling. In 2000 there was a book actually called Social Epidemiology, Berkman and Kawachi, and that was the seminal text on social epidemiology. And that spawned a number of other texts.

But in terms of a major methodological concern is this ecological fallacy or extrapolating from individual level to broader community context. And many people are working on solving that problem and figuring out how to credibly measure these moving from the individual to the broader level; there's an excellent model by Glass and McCahey that shows this macro, meso, and micro levels of interaction. So there's a lot of work being done in social epi that's totally applicable to what we do in drug abuse research.