<p>
I agree that, of course, it’s not a perfect analogy. I was just thinking of it because the grants are judged in multiple rounds, but money is only allocated once per year – the first-cycle grants are awarded based on the amount appropriated for the year, but technically at that point the institutes don’t know the quality of applications in future rounds. Many/most applications are also “deferred” to later rounds, in effect, by giving them scores that are just below the funding threshold, leading them to be revised and resubmitted. </p>
<p>
Well, except that the numbers apparently aren’t different (whatever the trend toward significance may be), and are based on self-reported profiles. Numbers may not lie, but people certainly do. For kicks, we might as well analyze the self-reported stats posts from EA and RD admits from last year – they’re stickied up at the top of the forum.</p>
<p>For what it’s worth, I think the EA and RD pools are each composed of several distinct groups of applicants, and there’s limited benefit in analyzing the two pools as groups without considering the subgroups that make them up. The EA pool contains Questbridge applicants, but otherwise is likely to shift toward the high-income/coast-dwelling/private school end of the applicant pool. (This is why Harvard eliminated its early program several years ago – they felt the early pool lacked diversity, and was composed mainly of people who were clued in to the admissions process enough to apply early.) The RD pool contains international students, who are often highly qualified but admitted in very low numbers. All of this variation between the two pools makes it questionably useful to distill their properties to a few numbers and compare them. </p>
<p>Interestingly, the admissions office does have the complete dataset, not just some self-reported profiles from a small percentage of applicants, and they do spend considerable time analyzing it, even if not releasing it.</p>