and by mr. rubin. critically important, we do not just look at national averages. there is a huge amount of variation in what is likely to work for different types of people, what that will necessarily entail, i did read the papers for the conference last night. they were very good. we are likely going to have to adopt different types of evidence procedures for this research effort and randomized controlled try and cannot be the only standard by which we judge things. in order to get a wider array of evidence brought to bear we are likely going to have to struggle with different types of evidence. what is fascinating is, at this same time the field of economics is moving away from data econ e econometri econometrics, causal relationships from larger databases, moving towards randomized controlled trials whenever possible, the medical profession is going to have to move to some degree in the opposite direction and rely on the panel data that will come out of a more expensive set of health information technology, precisely to examine the, be able to get exact individ