If your aim is evidence-based policy, then I think it's very ambitious to think you could test out some programs, some interventions, and see a clear link with changes in weight. I think the time scale and the variability are so great that that sort of measure is too challenging. Weight is very good if you want to see a general trend in changes, but not to inform your evidence-based policy. Probably the best you could do is to look at observed behaviour. That is something that is much shorter-scale. In time scale, it's much easier to see and manage in a regular study.
On February 26th, 2007. See this statement in context.