<p>
</p>
<p>But their study may not have anything to say about that. They claim that including the pool of colleges admitting or rejecting a given student serves as a control on bias from selection of applicants by the colleges. However, to draw conclusions they need to control the bias from selection of colleges by the admitted applicants. The low yield numbers at most colleges indicate that (contrary to what they assume in the study) there isn’t much relationship between these two sources of bias; admissions committees are indeed taking into account variables that the study can’t measure directly, but not in a way that has anything to do with controlling self-selection bias in the students. </p>
<p>One expects that for applicants with a given range of matriculation choices including some elite and non-elite schools, are there systematic factors driving both the choice and the later income levels (e.g. degree of ambition of the student) or influence of that choice on the income (lesser debt allowing for career in the nonprofit sector, etc).</p>