But I thought HYP were national universities! Why are ALL schools so regional??

Parents Forum
561

<p>"Why is Stanford at 81 for the NE while 45 and 44 for the other two out of regions? Why is WashU 130 for the NE while 58 and 71 for the other two out of regions? In both of these cases, most students had to travel farther from the NE to get to the school than the other two regions. "</p>

<p>Well, obviously Stanford’s got relatively more reach in the NE than in the midwest and south. And obviously WashU’s got relatively more reach in the NE than in the south and west. And part of this is because NE students in general are more likely to fill the pool of elite-school-appliers. I’ve said this all before. But you can’t compare the Stanford 81 to the WashU 130 and conclude anything about Stanford’s strength, selectivity, prestige, etc relative to WashU. Each school’s indexes are within / unto itself. </p>

562

<p>I didn’t know that about the NMF–didn’t know that each state was allocated a fixed amount.</p>

563

<p>"If you could study the movements of a large pool of “elite students,” if would be pretty interesting. But you can’t do this by just looking at top 20 research universities and LACs, because there are other strong draws for some of these elite students, such as honors colleges at flagships, and merit opportunities. "</p>

<p>Absolutely. Let’s say we could all agree that a score of X on a standardized test was a decent proxy for elite-school-worthy. So you could “star” these students and follow their movements (whether locally or around the country, whether to a state flagship or to an elite school),</p>

<p>Given that I can’t actually do that,
I have the choice between:</p>

<p>Option A) Assuming that such students are reasonably equitably distributed among students in all regions of the country, and index to the pop found in each region (the approach I used) OR</p>

<p>Option B) Assuming that the pool of those-who-attended-these-particular-20-elite-schools is the universe and the only set of students that I care about, and index to the distribution of that pool (which was BP’s approach).</p>

<p>The reason I prefer Option A, with its faults, to Option B, with its faults, is that it’s evident that the pool of those-who-attend-these-particular-20-elite (and all private)-schools isn’t a representative sample of all “elite-school-WORTHY” candidates. Because the elite-school-WORTHY crowd in many parts of the country applies to and gets “siphoned off” by Berkeley, Michigan, UCLA, Wisconsin, UVA, UNC, and so forth. (BTW, I think this concept is something that gets a lot of head-nods on CC, but I don’t think people who don’t live in those regions truly get the high caliber of kids who don’t consider elite private universities.)</p>

<p>Option A at least counts these kids by proxy by assuming they exist, but they don’t make their way to these particular private schools. Option B acts as though those kids don’t even exist. </p>

<p>And that’s my big problem with Option B. It “rewards” private schools that are in regions with relatively poorer public options and it penalizes private schools that are in regions with relatively stronger public options – because it focuses solely on those kids who decided to forego publics in the first place. Who foregoes publics in the first place? Why, people in regions where the publics are viewed as relatively weak. What region might that be? </p>

564

<p>I’d also be curious if the same states from each region are the ones that are supplying most of the representative for that region. I’d imagine that each school would show something like ‘home state’ + the same 5 states supplying most of the students. But I could be surprised as well.</p>

<p>In addition, since I assume that these schools all draw from similar areas, but varying amounts from those areas, we might not be able to conclude much about national or regional draw. So if for both School A and B, their ‘Midwest’ population is predominately from Chicago and St. Louis, while their ‘Northeast’ population is predominately from Boston and New York, but School A has lots more kids from Chicago and St Louis than NY or Boston, while B shows the reverse pattern, does that really mean anything with respect to draw form a region?</p>

<p>Also, if school A draws all of its kids from Region X from one state but school B draws its kids from all the states in the region, but School A has more kids from that one state than the total number that B draws from the various states in the region, it will show up as school A is more ‘representative’ or ‘over represented’ in that region. But those situations are very different.</p>

565

<p>“And part of this is because NE students in general are more likely to fill the pool of elite-school-appliers. I’ve said this all before.” This is all I was saying, and it is what makes the conclusions drawn inaccurate.</p>

566

<p>@skrlvr - check post #201 for a link to the number of NMSFs allocated to each State. Allocation is based on percentage of graduating seniors in State compared to the total for the nation out of 16000.</p>

567

<p>Pizzagirl, Why not use a variant of Option B? Take the top-20 universities, both public and private, and that would solve the problem of kids getting siphoned off to top public universities, right? Now, if you are inclined to use the US News rankings, there is only one public university in top-20, so you will need to go to top-30 to pick up the the likes of “Michigan, UCLA, Wisconsin, UVA, UNC” etc.</p>

<p>I would be interested in the result.</p>

568

<p>'Who are the non-NE schools admitting from the NE? When you look at your pool of schools not from the NE, the index is >100 for the NE for these schools. "</p>

<p>That’s simply not true, BP. Under my methodology, if I look at the non-NE schools, I have a mix of NE representation.<br>
For example, UChicago and WashU overindex around 130 from the NE, Northwestern, Notre Dame and Vandy index at fair share around 100, Rice and USC around 60. </p>

<p>Under your methodology, <em>every</em> non-NE school indexes below 100 in the NE because you’ve deflated it. Specifically, a non-NE school would have to have a *FULL 41% of its student body come from the NE * in order to “break” 100 using your methodology. Even though only 23% of the actual population of the country is found in the NE.</p>

<p>I think most normal people would say - if 23% of the population of the country is found in the NE, and 41% of College X student body from the NE, then College X skews to the NE. No one would say that College X is only “average” against the NE. But *that’s what your methodology demands. * Own it.</p>

569

<p>Start quote:</p>

<p>Me: What if someone doesn’t want to apply to WashU because they <em>correctly</em> think that WashU has a far high proportion of students from the midwest than an NE school (or a west coast school, or a school in St. Louis, France), and that’s what they mean when they say WashU is too midwestern for them?</p>

<p>Pizzagirl: Then whatever, since that would be a correct statement. It’s the incorrect statements I object to."</p>

<p>End Quote</p>

<p>That’s good. Have you or Sally ever asked people what they mean by WashU being “too midwestern”? Perhaps they would just have said that it means WashU has higher proportion of midwestern students than other schools of the same/better calibre, or that WashU is in the midwest so the surrounding culture would be midwestern as well, and the discussion would have ended there instead of having to prove that WashU is more “national” than the Ivies. </p>

570

<p>

Somebody, I forget who, suggested upthread that the students at Wash U are likely to be those who didn’t get into Ivies, etc. That may be true for some, but it’s probably not at all true for many of them, particularly those in Wash U’s region–and I think this is true for a lot of the other schools mentioned in the quote above. U.Va. always has and still does get some of the most qualified students from Virginia, for lots of reasons. Indeed, it’s possible that one reason some of the selective schools we’re talking about don’t have as many students from out of region as one might think is that a substantial number of the very best students in those regions don’t apply out of region at all. Thus, it’s not just about the relative quality of students in the different regions, but also the degree to which they are willing to go out of region.</p>

<p>This is another context in which the relative weakness of the public universities in the Northeast may have an impact–highly able students from the Northeast may be less likely than those in some other states to choose state schools for financial reasons. You’re not giving up much, if anything, if you go to U.Va., or Michigan, or Wisconsin at in-state rates as opposed to a costly private school. If you live in Massachusetts, New Jersey, or New York, you may make a different calculation.</p>

571

<p>Start quote: </p>

<p>Me: But I believe Pizzagirl and Poetgrl wouldn’t conider Columbia to be properly international till that number climbs to 95%. </p>

<p>Pizzagirl: I never said anything about the % of internationals. I am not indistinguishable from Poetgrl.</p>

<p>End quote:</p>

<p>I am merely extrapolating your national methodology to international. As you can see, in the international realm, it would lead to completely nonsensical conclusions. Ergo, it is leading to completely nonsensical conclusions in the national realm too. In numerical analysis, consistency is important.</p>

572

<p>“Pizzagirl, Why not use a variant of Option B? Take the top-20 universities, both public and private, and that would solve the problem of kids getting siphoned off to top public universities, right? Now, if you are inclined to use the US News rankings, there is only one public university in top-20, so you will need to go to top-30 to pick up the the likes of “Michigan, UCLA, Wisconsin, UVA, UNC” etc.”</p>

<p>Because I don’t have the data. The data that I have was a simple spreadsheet that looks as such:</p>

<p>Column A: Northeast, MW, S and W (with definitions of the regions footnoted).
Column B: % of US pop found in these regions (23%, 22%, 32%, 23%).
Columns C and on … The 40 or so schools listed across the top, and the % of their student body that came from each region. 4 numbers per school, summing to 100%. International students had deliberately been excluded and everything repercentaged back to 100%.</p>

<p>I then added in what PCHope provided re the actual student body at each school so I could figure out the raw # of people from each region at each school, which gave me what I need to figure out 1) the number / distribution of elite SEATS per region, 2) the number / distribution of elite STUDENTS per region, and 3) the number / distribution of students from each region who stayed within home region versus went elsewhere. I always kept unis and LACs separate and did each analysis twice. </p>

573

<p>"I think most normal people would say - if 23% of the population of the country is found in the NE, and 41% of College X student body from the NE, then College X skews to the NE. "</p>

<p>That’s because most normal people are not trained in statistics. You are assuming an uniform distribution in the population which simply doesn’t exist. SES varies sharply by region and college admissions are more linked to SES than anything else.</p>

<p>Put another way, most normal people may say that if 23% of the population is in NE, then no more than 23% of people earning a million dollars or more per year should be from NE as well. But that would ignore the Wall Street effect, and would be a wrong conclusion. </p>

<p>This is why statistical analysis needs to start by understanding the underlying distribution of the population. Which, frankly, you have not, which is why your analysis is giving suprious results.</p>

574

<p>I still would rather assume uniform distribution of elite-school-worthiness in the population, than deliberately tilt towards a region just because state schools in that particular region happen to be viewed as a less-viable option for elite-school-worthy students than they are elsewhere. The NE is the “outlier” region in terms of “worthiness of state schools for elite-school-worthy candidates” - why should its outlier-ness be rewarded as your methodology does? </p>

575

<p>I’m scratching my head, wondering why people are so upset with the results that started off this thread. It seems to me that they show a pretty simple pattern strongly enough so that most of the concerns we’ve raised are mainly quibbles. All they show, I think, is that pretty much all selective colleges (except maybe MIT) draw more heavily from their home regions that many of us would have suspected before looking at the data. Does anybody really think that conclusion is “spurious?” I’m as sensitive to Ivy-bashing as anybody, and there has been a bit in this thread (and lots elsewhere), but I think it’s interesting to compare the extent to which different colleges draw students from all around the country. Examining that doesn’t necessarily say anything about how much colleges should value that kind of distribution.</p>

576

<p>“I think most normal people would say”</p>

<p>I’m glad I’m not normal…</p>

577

<p>“I still would rather assume uniform distribution of elite-school-worthiness in the population, than deliberately tilt towards a region just because state schools in that particular region happen to be viewed as a less-viable option for elite-school-worthy students than they are elsewhere. The NE is the “outlier” region in terms of “worthiness of state schools for elite-school-worthy candidates” - why should its outlier-ness be rewarded as your methodology does?”</p>

<p>When doing satistical analysis you do not “reward” distribution trends, you <em>analyze</em> distribution trends. If the distribution in the population is not uniform, and you force fit an uniform distribution in your analysis because you don’t want to “reward” cretain population segments that deviate from the uniform distribution, then your analysis will be wrong.</p>

<p>It is really as simple as that.</p>

578

<p>“Because I don’t have the data. The data that I have was a simple spreadsheet that looks as such: …”</p>

<p>That is enough data to calculate what I suggested.</p>

579

<p>Choosing your assumptions has implications for your results. That, to me, does constitute “rewarding” certain outlooks.</p>

580

<p>Hunt, the results themselves were not what sparked many of the reactions so much as the not-so-veiled intent in presenting the data (see post 461 for a brief summary), and the secondary interpretations of the data which led to questionable and sometimes insulting conclusions about regional attitudes and college choice behavior.</p>