USNWR 2009: Looking at the Data XVI (Avg. SAT CR Scores & 25/75)

College Search & Selection
1

<p>The publication of the 2009 USNWR College Rankings provides an opportunity to compare schools based on a wide variety of data points. In this and in other threads, I urge the reader to think less about the absolute rankings and more about the nature and value-added of the data point being discussed.</p>

<p>Avg SAT CR , SAT CR 25 - SAT CR 75 , National University</p>

<p>750 , 700 - 800 , Harvard
750 , 700 - 800 , Yale
740 , 690 - 790 , Princeton
740 , 700 - 780 , Caltech</p>

<p>720 , 680 - 760 , Columbia
720 , 670 - 770 , U Chicago
715 , 660 - 770 , Dartmouth
715 , 680 - 750 , Wash U
710 , 660 - 760 , MIT
710 , 660 - 760 , Stanford
710 , 670 - 750 , Northwestern
710 , 660 - 760 , Brown
710 , 670 - 750 , Tufts
705 , 660 - 750 , Duke
700 , 650 - 750 , U Penn
700 , 650 - 750 , Georgetown</p>

<p>695 , 640 - 750 , Rice
695 , 640 - 750 , Notre Dame
690 , 640 - 740 , Vanderbilt
685 , 640 - 730 , Emory
685 , 630 - 740 , W&M
680 , 630 - 730 , Cornell
680 , 630 - 730 , Johns Hopkins
675 , 630 - 720 , Brandeis</p>

<p>670 , 620 - 720 , USC
665 , 620 - 710 , NYU
660 , 610 - 710 , Carnegie Mellon
660 , 610 - 710 , Boston Coll
655 , 610 - 700 , Wake Forest
650 , 590 - 710 , UC Berkeley
650 , 600 - 700 , U North Carolina
650 , 600 - 700 , U Rochester</p>

<p>645 , 590 - 700 , U Virginia
645 , 600 - 690 , Rensselaer
645 , 600 - 690 , Tulane
640 , 590 - 690 , U Michigan
640 , 590 - 690 , Georgia Tech
640 , 600 - 680 , Lehigh
635 , 580 - 690 , Case Western
630 , 570 - 690 , UCLA</p>

<p>615 , 560 - 670 , U Florida
615 , 550 - 680 , Yeshiva
610 , 550 - 670 , U Wisconsin
605 , 540 - 670 , U Illinois
605 , 540 - 670 , U Texas
600 , 540 - 660 , UCSD</p>

<p>590 , 530 - 650 , U Washington
590 , 530 - 650 , UC Santa Barbara
580 , 530 - 630 , Penn State
570 , 510 - 630 , UC Irvine
560 , 490 - 630 , UC Davis</p>

<p>Avg SAT CR , SAT CR 25 - SAT CR 75 , LAC</p>

<p>730 , 680 - 780 , Swarthmore
730 , 690 - 770 , Pomona
725 , 690 - 760 , Harvey Mudd</p>

<p>720 , 670 - 770 , Amherst
715 , 670 - 760 , Williams
705 , 660 - 750 , Wellesley
705 , 660 - 750 , Vassar
700 , 650 - 750 , Middlebury
700 , 650 - 750 , Carleton
700 , 650 - 750 , Haverford
700 , 650 - 750 , Claremont McK
700 , 650 - 750 , Wesleyan
700 , 660 - 740 , W&L
700 , 650 - 750 , Oberlin</p>

<p>695 , 650 - 740 , Bowdoin
690 , 640 - 740 , Hamilton
680 , 630 - 730 , Davidson
680 , 610 - 750 , Grinnell
680 , 640 - 720 , Colby
680 , 630 - 730 , Macalester
675 , 620 - 730 , Bryn Mawr
670 , 620 - 720 , Colgate
670 , 630 - 710 , Bates
650 , 590 - 710 , Smith</p>

<p>625 , 570 - 680 , US Military Acad
610 , 560 - 660 , US Naval Acad</p>

2

<p>The caveat about ACTs applies here, too, but it’s not as great an issue. It’s my belief (correct me if I’m wrong) that SAT is the test most commonly submitted to most selective colleges. Exceptions might be in the midwest, especially to large publics (like Wisconsin and Michigan, for two examples).</p>

3

<p>can you post the math SAT scores…why only CR?</p>

4

<p>Thank you, hawkette.
As an aside, the 75th percentile mark always has to do with the standard for those attending based upon their brain power. The 25th percentile, at least at private schools, reflects the numbers who are admitted based upon athletics and legacy status. Swathmore has no football team, and I can’t believe Pomona gives many a boost for sports prowess either. Neither school, nor the Chicagos, Cal Techs, and Case Westerns of the world are society schools the way Amherst and Dartmouth are, or Princeton, for that matter.
I think the 75th percentile mark is the one to watch. Others may disagree.</p>

5

<p>danas,
Using your hypothesis that the 75th percentile represents intellectual strength of the student body, below is how the schools compare to one another and divided in 25-point groups. </p>

<p>The 25-point groupings look about right to me. I also think that schools could easily be moved up or down one group. However, I don’t think that any could make the claim that they belong with colleges two groups above or below them. </p>

<p>SAT CR 75th , National University</p>

<p>800 , Harvard
800 , Yale
790 , Princeton
780 , Caltech</p>

<p>770 , U Chicago
770 , Dartmouth
760 , MIT
760 , Stanford
760 , Columbia
760 , Brown</p>

<p>750 , U Penn
750 , Duke
750 , Northwestern
750 , Wash U
750 , Rice
750 , Notre Dame
750 , Georgetown
750 , Tufts
740 , Vanderbilt
740 , W&M
730 , Cornell
730 , Johns Hopkins
730 , Emory</p>

<p>720 , USC
720 , Brandeis
710 , UC Berkeley
710 , Carnegie Mellon
710 , NYU
710 , Boston Coll
700 , U Virginia
700 , Wake Forest
700 , U North Carolina
700 , U Rochester</p>

<p>690 , UCLA
690 , U Michigan
690 , Georgia Tech
690 , Case Western
690 , Rensselaer
690 , Tulane
680 , Lehigh
680 , Yeshiva</p>

<p>670 , U Wisconsin
670 , U Illinois
670 , U Texas
670 , U Florida
660 , UCSD
650 , U Washington
650 , UC Santa Barbara</p>

<p>630 , UC Davis
630 , UC Irvine
630 , Penn State</p>

<p>SAT CR 75th , LAC</p>

<p>780 , Swarthmore</p>

<p>770 , Amherst
770 , Pomona
760 , Williams
760 , Harvey Mudd
750 , Wellesley
750 , Middlebury
750 , Carleton
750 , Haverford
750 , Claremont McK
750 , Vassar
750 , Wesleyan
750 , Grinnell
750 , Oberlin</p>

<p>740 , Bowdoin
740 , W&L
740 , Hamilton
730 , Davidson
730 , Bryn Mawr
730 , Macalester</p>

<p>720 , Colgate
720 , Colby
710 , Smith
710 , Bates</p>

<p>680 , US Military Acad
660 , US Naval Acad</p>

6

<p>Except that Middlebury, Bowdoin, Hamilton don’t include the lowest 20-30% of scorers. If you assume the excluded contingent are at the 25%ile, which is probably a generous assumption, it resultingtly lowers their medians by 20-30 points.</p>

7

<p>I’m skeptical that SAT scores taken alone tell us all that much about a school. I’m even more skeptical that the SAT Critical Reading score taken alone, separate from the SAT math score, tells us anything at all useful. Different schools will attract different mixes of student strengths. Schools that are particularly strong in math, science, and engineering—MIT for example—will draw a student body that is disproportionately strong in math and science. Not surprisingly, the MIT student body’s 25th percentile, 75th percentile, and median SAT math scores are all a full 60 points above the comparable SAT CR scores at that school. Separating out the SAT CR score strikes as at best a pointless and at worst a pernicious exercise that is bound to lead to misleading results. Any ranking that says Wash U’s students are better qualified than MIT’s is pure horse manure.</p>

<p>I also disagree with danas’s argument that you should throw out the 25th percentile score on grounds that this reflects those admitted “based upon athletics and legacy status.” There’s a certain truth implicit in this claim: schools have a lot of leeway to admit athletes, legacies, “development cases” and other favored categories at the bottom of the entering class. But they can do so precisely because the SAT scores of the bottom 25% of the class don’t materially affect the 25th percentile score, and consequently don’t adversely affect the school’s US News ranking. Every school focuses on two scores, the 25th and 75th percentile. Consider an entering class of 1000 students, with a normal distribution of SAT scores, arranged in rank order from #1 (highest score) to #1000 (lowest score). The only scores that really affect the 25th and 75th percentile medians are scores #250 and #750. It doesn’t matter at all how high the scores are in the top quartile: a class with scores #1 through #249 at 1600 and #250 at 1500 will have a 75th percentile score of 1500—just the same as a class with scores #1 through #249 at 1510 and #250 at 1500. Same at the 25th percentile level: the goals is to get score #750 as high as possible. Once you’ve got that figure set, it doesn’t matter how low the bottom quartiles scores go, they simply won’t affect the 25th percentile median. It’s in that bottom quartile that you’re likely to find a lot of athletes, legacies, and other favored groups. But every school is going to be just as concerned about keeping up its 25th percentile score as it is about keeping up its 75th percentile score. </p>

<p>Of course, I don’t mean to imply that all athletes or legacies will end up in the bottom quartile; an athlete or legacy who can help pull up the 25th or 75th percentile will be an especially attractive catch.</p>

<p>Also, given the reality of the admissions process, the adcom will be unable to determine in advance exactly which SAT score will end up as #250 and which as #750; therefore, they need to be concerned about the cluster of scores immediately above and immediately below their anticipated 25th/75th medians. But that still leaves a lot of room for “freebies” in the lower end of the bottom quartile whose scores won’t materially affect the 25th percentile. (This, incidentally, is also why perfect SAT scores are not nearly as important to adcoms as applicants often imagine; in general, any score solidly above the school’s anticipated 75th percentile median is just as attractive, and in some ways scores very close to but just above the expected 75th percentile median—or in some cases, just above the anticipated 25th percentile median—will have a greater impact on the schools ultimate reported outcome, and a certain amount of merit aid will gravitate toward those scores, not necessarily to the highest scores).</p>

8

<p>This does not take superscoring into consideration. Also, many private universities report admitted scores rather than enrolled scores.</p>

9

<p>

</p>

<p>Don’t many schools report mean SAT scores too? According to what you’re saying mean SAT scores would be potentially significantly below median SAT scores. In addition, do you really think colleges have changed their admissions approach so much in the last 20 years because they have an eye on what their mid SAT range will be for USNWR?</p>

<p>

</p>

<p>Since that is such a substantial difference, I can’t imagine that such a basic concept would be overlooked that it would not be on an apples to apples basis.</p>

10

<p>

</p>

<p>In a word, yes. I think the US News rankings have had an enormous distorting effect on the college admissions process, from top to bottom. It’s a national scandal, and a national disgrace. It’s a classic example of “teaching to the test.” US News is widely read and extremely influential. College administrators by and large hate it, but they feel captive to it and most feel they have no choice but to play the game. Of course they’d like to have the strongest class possible, but when push comes to shove it’s the relative handful of marginal scores that determine their SAT 25th and 75th percentile medians that matter the most to them, and in many cases financial aid policies are geared toward manipulating those numbers.</p>

<p>Do they report their SAT means? I don’t know. Where would they report these? I never see them published. The only numbers I ever see published are 25th and 75th percentile medians, and sometimes 50th percentile median (though often this is just estimated by averaging the 25th and 75th percentile medians—which really doesn’t work, mathematically speaking).</p>

<p>Alexandre’s right about superscoring, though. Colleges superscore not because they’re trying to be kind to applicants and giving them the benefit of the doubt, but because it’s in their self-interest to report the highest scores they plausibly can. Many publics do not do this, consequently their reported scores are lower than the reported scores of a superscoring private school with an identical applicant pool/admit pool/enrolled student profile. Apples to oranges, if you will. I don’t know about reporting admitted scores v. enrolled scores, but I wouldn’t be surprised if some do, if they can get away with it. Self-interest, you know.</p>

11

<p>

</p>

<p>Not sure about “many”, but I’ve found some. I am also not sure if this is only a “private” thing.</p>

12

<p>While standardized test scores are but a single data point in the evaluations by college adcomms, they are probably the single best clue that independent observers have in evaluating the academic strength/preparedness of a school’s entering student body. The National Association of College Admissions Counselors (NACAC) has consistently indicated that standardized test scores, in combination with rigor of curriculum and grade performance in college preparatory classes, are the most important factors in their evaluations.</p>

<p>As for the criticism above about the usefulness of comparing only CR data or only Math data, I am presenting in this manner as some have criticized me for presenting material on a combined basis. I can certainly post in combined fashion for CR and Math if some would prefer to see the numbers in this format.</p>

<p>As for the criticism about the 25/75 presentation, I plan additional threads that will detail the % of students scoring above a certain threshold. My personal view is that this data is far more telling about student body strength/depth than the 25/75 data. Perhaps you will consider this a better reflection of the student body strength/depth (although I doubt you will ever accept any data that somehow doesn’t place your school in a favorable ranking position). </p>

<p>Concerning the matter of superscoring and how this affects reported scores, I would like to stress two very important points:</p>

<ol>
<li><p>While some colleges will not use SAT superscoring in their admissions process, I strongly believe that they will still present their data in the CDS in superscored form. This is very easily done. As fiduciaries, the school is obligated to present its data in a manner consistent with the marketplace and in a manner that accurately reflects the school’s standing in the marketplace. Unless a college makes an explicit statement otherwise, I will always assume that a CDS will follow the established pattern of the marketplace and present superscored data. </p></li>
<li><p>If as some claim, superscoring leads to inflated SAT scores relative to ACT scores (where no superscoring takes place), then this should be evident in a comparison of colleges with representative numbers of enrollees who submitted SAT and ACT scores, eg, 25% or more of the students. If superscoring leads to higher SAT scores vis-</p></li>
</ol>

13

<p>

</p>

<p>What’s your basis for the 25% figure? Why is this level representative?</p>

<p>Also, what is the “Institute for Higher Education?” I don’t know that organization, but their concordance table differs from that provided by the College Board. (i.e the College Board considers an ACT of 30 to be comparable to an SAT 1340). The College Board concordance is based on 300,000 students who took both exams.</p>

14

<p>hawkette:</p>

<p>Using the UCs for ACT comparison is just plain silly since Calif is THE biggest CB customer, and the common data sets do not break out ACT scores. It probably doesn’t work much at Stanford either, since Calif is test crazy and Stanford wants a geographicaly diverse student body…</p>

<p>The midwestern colleges are probably the only privates that have a good mix of both ACT and SAT students.</p>