A few interesting points are in this article: I strongly recommend it. The 'off the record' tone of it is interesting. Adding to the interest, the author apparently didn't actually mean for this data to be disseminated widely, according to this page, hosted at yale.edu. So if you have difficulty with the concept of reading 'off the record' remarks, or remarks aimed at a selected few, I recommend not reading the first article.
My thanks to tenzil for calling my attention to this interesting pair of pages, which he found in Scott Rosenberg's blog on Salon.com.