<p>I think a list like this is only useful for "broad brush’ characterizations. For example, rating very high on this list obviously means, among other things, a very highly qualified student body academically AND a high percentage of the student body interested in an academic career. It also may say something about the degree of interaction between undergrads and professor/mentors.</p>
<p>I think the best use of this kind of list would be as one of many data points in comparing two colleges with similar overall admissions standards. For example, the difference between Swarthmore and Williams or between Yale and Harvard may indicate some real underlying difference in the “typical” student. I don’t think that small differences in ranking on a chart like this indicate a “qualitative” difference – although very large differences (#3 versus #300) surely would.</p>
<p>I agree with Driver that it would be nice to have a data set that include MD and Law degrees. I’ve seen no evidence that anyone collects that data. The PhD data is actually collected under a decades-old National Science Foundation grant.</p>