Re: the discussion of the weight given to UG institutions - one small piece of evidence that speaks to this is this document: http://www.stanford.edu/group/SPLS/docu ... sMay08.pdf
It's a chart of the numbers for Stanford UG students who matriculated to law school. Unfortunately it's a couple years old, but the general point stands. If you peruse the numbers for different schools, you'll see that the median GPAs for matriculating students are lower than the overall medians for the law schools they attend. (E.g., the median GPA for Stanford UG kids going to HLS is 3.76, 3.84 for YLS, and 3.83 for SLS - numbers that are closer to those schools' 25th percentile numbers than to their medians). The differences are less pronounced (if evident at all) for LSAT scores, which makes sense - going to a harder UG should mean that, for two equally "smart" students, the one going to the harder UG would have a lower GPA, but an equivalent LSAT.
Anyone have comparable data for HYP? My guess is the bump is the same, if not greater.
Edit: just noticed that the GPA numbers on that chart are means, not medians, which skews the data somewhat (I'm not sure in which direction, but probably down).