Since we're talking health outcomes here, how prudent is it to measure these outcomes based on respondents' self assessments of how healthy they are, as this study does, rather than empirically measuring those outcomes?