<p>If the vast majority of apps are qualified, then it does make sense that it comes down to minor differences between the apps. Now, these apps vary only by content, not by value. That is, one student may be a violinist and another a cellist, but that says nothing about the quality of the application itself - nothing quantifiable. That is, an admissions officer will admit someone over another because Princeton is looking for a cellist to fill out the orchestra, and not a violinist. Or some other small difference like that. Therefore, it’s an inherent characteristic that counts, not because someone applied SCEA. If the cellist applied RD and the violinist SCEA, the violinist would be denied and the cellist accepted. The results would be the same either way, if we’re talking about these differences. </p>
<p>Now, I have seen Princeton students who should not be there by virtue of their statistics. I do not know if they applied SCEA or RD. I just know that of the dozens of people I have had in-depth conversations with who had applied SCEA, NONE would have been considered “borderline” within any standards. These are people who are strong all around. I suspect that this is due to either A) the small sample size being nonrepresentative of the overall SCEA applicant pool or B) the fact that less-than-qualified applicants - your “borderline” applicants - were routinely denied during SCEA because they did not receive any extra “boost” that you allege. </p>
<p>The 2% vs. 19% stats were not what I was referring to. I was referring to the data in the study you cited which I have repeatedly questioned the validity of, due to it being outdated. And which I still have yet to see a critical defense of (because they said so is not a viable defense). </p>
<p>The most recent study you cite takes a bunch of data from a bunch of schools and makes a generalization about all the data. Due to the large size of the data set, a discontinuity at Princeton’s data would be obscured. For instance, I could take a survey of everybody in America and “discover” that Americans prefer watching football over soccer. However, in making such a generalization, I would be ignoring the fact that a subset of the population (which I chose NOT to distinguish in my data set) would rather watch soccer over football. For instance, it could be that Spanish-Americans would rather watch soccer while everybody else would rather watch football. The large size of my data obscures this fact. Second, I see nothing in the study that applies specifically to Princeton. You’re making a generalization from a lot of data and applying it to the specific. That’s a no-no.</p>