<p>ch,
It’s time to retire the Revealed Preference study. Putting aside the many defects in its construction, it’s a very dated analysis. </p>
<p>As the information presented so far in this thread attests, the differences in students that are enrolling at all of these colleges are pretty darn small. A decade ago, when the RP work was done, that was not nearly as much the case. </p>
<p>For those of you who prefer comparing based on the ACT exam, here is the data:</p>
<p>The Revealed Preference Study was based on a survey conducted specifically for the study in 2000. The data come from over 3200 high school seniors who were identified as being in the top 10 or 20 students in their high school classes by counselors at about 400 high schools in 43 different states, schools that typically send some students to top colleges.</p>
<p>This study is based on ACTUAL college decisions. The results are analyzed statistically in a way similar to the way rankings are determined among chess players. There is extrapolation in the sense that students who apply to Harvard don’t apply to Podunk so there are no cross-applicants in some cases. You have to make inferences. If students prefer Harvard over Clark and students prefer Clark over North Dakota State then students would probably prefer Harvard to North Dakota State. It’s transitive inference.</p>
<p>The authors state that the Revealed Preference Study focused on top colleges because they have sufficient cross-applicants.</p>
<p>The authors also propose a way to control for financial aid. Financial aid can help a less desirable college “win” over a college that is actually more intrinsically desirable academically.</p>
<p>You should read the article. It’s pretty insightful.</p>
<p>… but if mormons prefer BYU to Harvard, and nobody but mormons even bother applying to BYU to turn it down later, then by transitive inference BYU will show up above all other colleges in revealed preference ranking. Because the people who dispreferred BYU to other colleges didn’t even bother applying there, to have their dispreference entered as a data point.</p>
<p>This methodology problem arises because the sample of people who actually applied to BYU is a biased sample of the underlying population of applicants to college in general, not an unbiased sample.</p>
<p>Besides which, there are all sorts of reasons why people choose a college which may or may not be applicable to your own situation. Location, distance from home, weather, financial aid, particular programs of study,…
A choice that is better for most people may not be better for you.</p>
<p>Right. The authors address the problem of self-selection. They also talk about Caltech which has a similar issue with BYU.</p>
<p>“Table 4 helps us to understand the results for Cal Tech, which are somewhat
problematic. Because students self-select into applying to Cal Tech based on an orientation
toward math and science, Cal Tech’s pool of admittees overlaps only slightly with that of most
other institutions, except for MIT, with which Cal Tech has substantial overlap. MIT, on the
other hand, does have substantial overlap with other top schools. Unlike the other institutions
in the top twenty, Cal Tech appears to draw a more focused group of applicants. In Section
III.D, we discussed how such self-selection might bias inference for some speciality schools,
with the possibility of some upward bias in the estimate.”</p>
<p>"if you read the study, they actually talk specifically about BYU "</p>
<p>… because that, and some other examples cited, are the most obvious examples of the generic problem with the study. All the data involves biased samples. However the nature of the bias may not be so obvious in other cases. That doesn’t mean it is absent, or that other results are not consequently distorted by it.</p>
<p>“Perhaps the single most interesting college in Table 6 is Brigham Young, which appears
in the top 10, between Princeton and Brown, in region 8 (which contains Utah). We have
checked and determined that, if we were to compute a Utah-specific ranking, Brigham Young
would rank even higher. The dramatic appearance of Brigham Young in the top 10 almost
certainly occurs because the college is particularly desirable in the eyes of Mormon students.18
We cannot verify this conjecture because we did not ask students about their religion, but this
leads us back to our general point about latent desirability and self-selection into applicant
pools. The reason that Brigham Young wins so many tournaments with Utah students is that it
is truly more desirable to them. Similarly, the reason that a bit of regionalism appears is that
University of the South, say, is truly more desirable to Southerners. This is not a problem we need to “fix” in the national ranking. It is simply an indicator that, with sufficient data, it
would be reasonable to compute sub-rankings for identifiable groups of students with welldefined
tastes. We know now that these rankings will tend to join up at the top. A benefit of
computing sub-rankings is that some colleges’ performance in the national rankings depends
on the fact that they are especially popular with a well-defined set of students who self-select
into applying (think of Cal Tech). Self-selection does not appear to be an important concern
with our national ranking, except perhaps for the engineering schools. However, we speculate
that it would be appropriate to construct sub-rankings once we got much outside of this group.”</p>
<p>CH,
Again, the study is dated. Perceptions and visibility change. The best proof we have for that is the huge increase in applications to many colleges over the past decade. </p>
<p>Far more than was the case a decade ago, many more schools are on the application lists of high school students who might rank in their school’s top 20 or 30 students. Back then, it was likely tight universe and applications out to 3-4 schools. Today it’s often 10 or more applications and all kinds of important factors coming into play in a far more significant way than in 2000, not the least of which is affordability.</p>
<p>But still, collegehelp, what do other people’s preferences matter when I’m making my own choices? </p>
<p>If more people prefer cities to rural areas, should I then prefer cities, too? If more people prefer beach vacations to skiing, should I then prefer beach vacations, too? If more people prefer vanilla over chocolate, should I then prefer vanilla too? It’s IRRELEVANT. Utterly irrelevant to anything.</p>
<p>People choose things based on all different kinds of criteria, some of which may be mine, and some of which may not be mine. </p>
<p>Don’t you have any self-confidence in your own preferences, or are you so lacking in self-esteem that you need to substitute others’ preferences for your own?</p>
<p>Competitive sets are different by region, however. In the midwest and California, for example, large state schools are part of the competitive set for the smart kid in a way that they typically aren’t in the Northeast.</p>
<p>First line of the thread: “Four universities located in the South/Southwest now stand as legitimate rivals to the Non-HYP Ivies in the competition to attract the top strain of high school applicants.”</p>
<p>What collegehelp has been showing refutes that.</p>
<p>Absolutely. It is amazing to me how insular the northeasterners on CC are. They really don’t get that yes, for smart kids in other parts of the country, the Michigans and Wisconsins and Vanderbilts are just as much a part of the competitive set for the smart kid as the Ivies. Methinks that for living in such an allegedly “sophisticated” part of the country, they’re actually well behind the times.</p>
<p>These schools already have been rivals for the top strain of high school applicants for years now, BuddyMcAwesome. A few of you from the northeast finally noticed these schools, and all of a sudden they’re being deemed as on the map. They were on the map a long, long time ago. Provincial people just didn’t notice.</p>
<p>speaking of provincial people reminded me, somebody ought to post the % of students from the midwest, and then from the South, etc. who are in attendance at each of the schools we’re talking about.</p>
<p>Of course you’ll find that schools tend to pull in from their local areas. That’s true of the midwestern and southern schools, and it’s true of the northeast schools as well.</p>
<p>New England 4.4%
Midwest 14.4%
South 44%
Middle States 13.6%
West 6.8%
Southwest 7.9%
U.S. Territories >.007%
International 8.4%
Unspecified >.4%</p>
<p>So basically half of Vandy is comprised of kids from the South + Texas</p>
<p>I’m getting at the fact that on CC, if a Northeast school pulls disproportionately from the Northeast, it’s a national school, but if a school located elsewhere pulls disproportionately from that area, it must just be a regional school. It seems that the 44% Vandy from the south and 45% Dartmouth from NEng / Mid-Atl are pretty much the same thing. BTW, I have no dog in the fight for either school.</p>