<p>“I’m not sure why you would exclude some graduates of non-PhD fields at Cornell, but not do the same for the other colleges and universities. That’s a slippery slope.”</p>
<p>The reason you’d want to do it is to compare Phd rates across colleges for people who are studying the same sorts of things, in the same types of colleges. Comparing apples to apples. A prospective liberal arts major might want to compare PhD rates for Haverford vs. Cornell’s Arts & Sciences College; probably less “apples to apples” with the irrelevance of Cornell’s College of Architecture thrown in. etc.</p>
<p>I think you can get reasonably close to this objective when you do % law school, because the vast preponderance of the law school applicants from Cornell are coming from only two colleges there, one of which is tiny. But it’s a little tougher to do that with Phds, because probably 4 and maybe 5 of the colleges-including some contract colleges- produce a reasonable %, so there’d be a lot more error when you exclude colleges from the denominator.</p>
<p>So you probably have to live with the heterogeneous, incomparable denominator. Recognizing that the result will have that much less utility, for an applicant to a particular one of its colleges which may have a profile of student destinations that differs markedly from this university-wide aggregate of disparate colleges.</p>