Ranking the Ivies

College Search & Selection
81

<p>laptoplover and Alexandre-
The total freshman class at Dartmouth (about 1100) was about 60% that of Cornell Arts and Sciences and Engineering (about 1850). So, to do a proper comparison you would have to compare Dartmouth with the top 1100 out of 1850. Those 1100 would have much higher SATs than the Dartmouth class.</p>

<p>slipper-
It does not make sense to say that the bottom third of the students at Cornell are the worst in the Ivy League because no other Ivy League school has 1000 in their bottom third. Cornell has no counterpart in the Ivy League. Dartmouth's top 1000 are the same as its bottom 1000. It would not be possible to compare bottom 1000 until all the Ivy League schools had a freshman class of 3000, like Cornell. When Dartmouth adds 2000 to its freshman class, then we can compare the bottom 1000 at Cornell with the bottom 1000 at Dartmouth.</p>

<p>Let's say that we were talking about Ivy League football teams instead of Ivy League freshmen. Say Cornell's roster had 33 players and Dartmouth's roster had 11 players (in proportion to the number of freshmen). When you say that the bottom 1000 of Cornell's freshmen are worse that the bottom 1000 at Dartmouth, it is like saying that the first team at Dartmouth is better than the third team at Cornell. It doesn't make sense.</p>

<p>hawkette's "pound for pound" concept is interesting. I too have wondered about the "population density" of top students and the role it plays in the quality of campus life. I would say that, if you were going to apply the pound for pound concept, then you should compare pounds of the same thing which would be Dartmouth Arts and Sciences and Engineering with Cornell's Arts and Sciences and Engineering. </p>

<p>Finally, I just want to reiterate that my point has validity although it is hard to get your head around it. It is a revelation to apply statistical concepts when comparing schools.</p>

<p>One more thing...I don't think yield is a good indicator of quality. Nor is acceptance rate. The number of applications is partly a function of marketing. Schools know what their yield will be if they accept a certain number of students so they increase acceptances to yield the class they need.</p>

82

<p>This is exactly why you can't do what you are doing!!! Dartmouth's ENTIRE class isn't 1100 students. You can't have it both ways!! How is this difficult?</p>

<br>


<br>

83

<p>slipper-all schools try to enroll the best students they can. Some enroll 1000 each year, some enroll 5000 each year. To make comparisons fair, you have to set a number and compare the top X number of students with the top X number of students. That makes it apples and apples. The top 1000 at a certain school is not equivalent to the top 5000 at another school. It is a hard concept to grasp.</p>

<p>yield rates are quirky. here is a rank list of the top 75 schools in yield.
Harvard University 83
Brigham Young University 77
Yale University 70
Princeton University 69
Stanford University 67
Massachusetts Institute of Technology 66
Yeshiva University 66
University of Pennsylvania 66
University of Florida 62
Thomas Aquinas College 60
Principia College 59
University of Georgia 58
University of Notre Dame 58
Columbia University in the City of New York 58
Brown University 58
Texas A & M University 58
The University of Texas at Austin 55
University of Illinois at Urbana-Champaign 53
University of Virginia-Main Campus 51
Ohio State University-Main Campus 50
Wheaton College 49
Dartmouth College 49
Saint Johns University 48
Virginia Military Institute 48
Georgetown University 47
Williams College 47
University of Missouri-Columbia 47
Barnard College 47
The University of Tennessee 47
Georgia Institute of Technology-Main Campus 46
Iowa State University 46
Cornell University 46
University of Washington-Seattle Campus 45
University of Michigan-Ann Arbor 44
College of Saint Benedict 44
Michigan State University 43
Duke University 43
Middlebury College 42
University of Wisconsin-Madison 42
University of California-Berkeley 41
Northwestern University 41
Bowdoin College 41
Wellesley College 41
Wabash College 40
Swarthmore College 40
Clemson University 40
Sweet Briar College 40
Virginia Polytechnic Institute and State University 40
University of California-Los Angeles 39
Pomona College 39
Indiana University-Bloomington 39
Stevens Institute of Technology 39
Davidson College 39
Presbyterian College 39
Vanderbilt University 39
College of William and Mary 39
Washington and Lee University 39
Wesleyan University 38
Union College 38
University of Maryland-College Park 38
Amherst College 38
University of Minnesota-Twin Cities 38
Bryn Mawr College 38
Hendrix College 37
California Institute of Technology 37
Smith College 37
New York University 37
Vassar College 37
Claremont McKenna College 36
University of Iowa 36
Bates College 36
Hope College 36
Carleton College 36
Wake Forest University 36
Haverford College 36
Purdue University-Main Campus 36</p>

84

<p>The concept is that Cornell cannot fill a class full of the strongest students so overall it is not as strong. Just as you can argue that there are more top scorers at cornell, the opposite arguement is that you are only as good as your weakest link. Also, a class is recruited not purely based on stats, but rather diversity, athletic needs, institutional needs (legacies), etc. This gives a larger school like Cornell an advantage in terms of accessing the largest number of top scorers.</p>

85

<p>If you had to pick one statistic to compare two schools, IMO it would be cross admit yield. Simply put, when students have the luxury of choosing both schools as an option, which school do students select more often?</p>

<p>According to the Revealed Preferences survey, its Dartmouth over Cornell: 67% vs. 33%.</p>

<p><a href="http://img46.imageshack.us/my.php?image=nytcrossadmitup5.jpg%5B/url%5D"&gt;http://img46.imageshack.us/my.php?image=nytcrossadmitup5.jpg&lt;/a&gt;&lt;/p>

86

<p>the_prestige-
The proportion is probably quite small of applicants who apply to both Dartmouth and Cornell and then choose either Dartmouth or Cornell rather than one of the other 8 schools to which they applied. So, I think cross-admit yield is limited. I think the best statistic for comparing schools is SAT scores because they are standardized, almost everybody has them, and they are not subject to idiosyncratic personal preference factors. </p>

<p>slipper-
I don't understand your point regarding the recruitment for diversity, athletes, legacies. How does that factor into the discussion? Are you saying that Cornell has an advantage enrolling top SAT scorers because Dartmouth enrolls a higher proportion of less-qualified minorities, athletes, and legacies?</p>

87

<p>collegehelp,
I agree that SAT scores are probably the best measuring stick and that the cross-admit data is more a reflection of school prestige than any statistical measurements. When looking at SAT data and overall student body quality, however, your comparisons as applied to only 1000 students is flawed. At Cornell, you don't attend with only 1000 other freshman (last year, Cornell enrolled 3188 students while Dartmouth enrolled 1086). You need to measure the whole class. </p>

<p>Cherry picking the class (on the upside or the downside) badly distorts any conclusions that one can draw about student body strength at a college. Many schools can do this and claim excellent student quality, eg, U Florida enrolled 1000 students that scored at the 1400 level or higher and this group likely had a SAT average equal to or better than Dartmouth or Cornell. Would that make U Florida a valid comp to Dartmouth or Cornell? Much as I admire the strides that U Florida has made the last few years, I think we'd all be hard-pressed to accept that university and its students as a true peer to these schools. I also think that this is pretty obvious stuff for anyone not making a partisan defense of a school.</p>

88

<p>hawkette-
Comparing the top 1000 students is a legitimate way to compare schools. I know that schools are conventionally compared based on their entire freshman class but this basis for comparing schools is not the only way to compare schools. </p>

<p>This debate is really about the most legitimate basis for comparing schools. Is it the individual student or the entire class? I am arguing that a head-to-head comparison of individual students, taking class size into account, is the most legitimate basis. Compare student #1 at school A with student #1 at school B, then #2, and so on. When you compare based on entire class units you fail to control for class size. Limiting comparisons to the top 1000 is a way to keep class size constant...apples to apples. </p>

<p>You point out that students attend a school in the midst of an entire student body, not just the top 1000. There is some truth to that. But, in the college of engineering at Cornell, maybe 80%-90% of your overall interactions are with other engineering students or with other students in your department. Furthermore, students are admitted only partly based on completing the class profile. Students are primarily admitted based on individual accomplishments, they study individually, they take exams individually, they are graded as individuals, they receive their individual diploma, and they scatter after graduation to lead individual lives.</p>

<p>Perhaps you can appreciate my earlier analogy with football. The team that wins is the team that puts the best 11 players on the field regardless of the size of the roster.</p>

<p>I think slipper and I are having trouble understanding each other because we are starting with different basis for comparison: individual students versus the entire class. Comparing the entire class is conventional but it is not the only legitimate basis for comparison. I hope that everybody continues to think about this without tunnel vision. It is an interesting problem.</p>

<p>Anyway, here are 100 universities ranked according to my estimate of the SAT score of the 1000th freshman.
school, freshman class size, sat 25th percentile, sat 75th percentile, estimated SAT of 1000th freshman</p>

<p>University of Illinois at Urbana-Champaign 7570 1170 1400 1476.1
Cornell University 3188 1280 1490 1461.3
Harvard University 1684 1390 1590 1454.4
University of Michigan-Ann Arbor 5356 1210 1420 1453.6
University of California-Berkeley 4059 1220 1450 1452.7
Duke University 1724 1380 1560 1443.3
The University of Texas at Austin 7369 1100 1350 1429.0
New York University 4703 1210 1410 1428.7
University of California-Los Angeles 4386 1170 1410 1423.5
University of Pennsylvania 2385 1300 1490 1423.2
University of Florida 6641 1140 1360 1419.7
University of Wisconsin-Madison 6118 1160 1370 1417.7
University of Southern California 2763 1280 1460 1416.7
Northwestern University 1952 1320 1500 1406.0
University of Minnesota-Twin Cities 5424 1120 1360 1400.2
Stanford University 1648 1340 1540 1399.9
University of Notre Dame 2037 1290 1500 1398.1
University of Virginia-Main Campus 3088 1220 1430 1396.7
Yale University 1321 1400 1580 1396.5
Washington University in St Louis 1473 1370 1530 1394.2
Brown University 1469 1350 1530 1390.6
University of Maryland-College Park 3945 1170 1390 1389.3
Brigham Young University 5286 1120 1350 1385.1
University of North Carolina at Chapel Hill 3749 1210 1390 1384.1
Boston University 4209 1180 1370 1375.1
Michigan State University 7308 1030 1290 1372.2
Georgia Institute of Technology-Main Campus 2837 1230 1400 1362.9
Ohio State University-Main Campus 6266 1090 1310 1361.6
Texas A & M University 7471 1080 1290 1357.9
Boston College 2284 1250 1420 1355.2
Pennsylvania State University-Penn State Main Campus 6489 1100 1300 1351.3
University of California-San Diego 3434 1150 1370 1349.8
University of Georgia 4669 1130 1330 1347.2
University of Washington-Seattle Campus 4857 1100 1320 1343.8
Iowa State University 3752 1080 1350 1341.2
Georgetown University 1591 1290 1490 1341.0
Columbia University in the City of New York 1338 1340 1540 1340.6
Rutgers University-New Brunswick/Piscataway 5086 1080 1310 1340.0
Princeton University 1240 1370 1590 1339.6
Tufts University 1282 1340 1480 1330.0
Vanderbilt University 1622 1280 1460 1329.9
University of Chicago 1203 1370 1560 1329.7
University of Iowa 4256 1060 1320 1328.9
Indiana University-Bloomington 8099 1000 1240 1326.5
George Washington University 2408 1200 1390 1324.6
Purdue University-Main Campus 7476 1020 1250 1324.4
University of Missouri-Columbia 4786 1070 1300 1323.2
Carnegie Mellon University 1424 1300 1490 1320.3
University of Colorado at Boulder 5595 1060 1280 1320.1
Virginia Polytechnic Institute and State University 5081 1100 1290 1314.8
University of California-Santa Barbara 3798 1080 1320 1312.2
University of California-Irvine 4314 1100 1300 1308.3
University of Pittsburgh-Main Campus 3249 1130 1330 1304.2
Syracuse University 3054 1120 1330 1295.1
Miami University-Oxford 3606 1110 1300 1288.2
Emory University 1259 1300 1470 1281.6
The University of Tennessee 4243 1050 1270 1277.5
University of Miami 2047 1180 1360 1274.0
University of California-Davis 3182 1060 1300 1267.2
Clemson University 2813 1120 1310 1267.2
University of Connecticut 3241 1090 1290 1264.2
University of Delaware 3202 1100 1290 1264.1
SUNY at Binghamton 2079 1160 1350 1262.0
Baylor University 2782 1100 1310 1261.1
Johns Hopkins University 1304 1270 1490 1260.9
College of William and Mary 1350 1240 1440 1243.6
Dartmouth College 1081 1350 1550 1236.4
Auburn University Main Campus 4069 1020 1230 1232.5
University of California-Santa Cruz 2976 1050 1270 1230.2
Rensselaer Polytechnic Institute 1270 1220 1420 1199.8
University of Rochester 1233 1230 1420 1199.6
Lehigh University 1217 1220 1390 1189.0
American University 1387 1170 1370 1182.5
Marquette University 1852 1090 1300 1179.4
Wake Forest University 1125 1240 1400 1175.2
Fordham University 1702 1100 1290 1164.0
Southern Methodist University 1371 1140 1320 1148.5
University of California-Riverside 2660 950 1200 1134.3
Saint Louis University-Main Campus 1564 1080 1300 1131.2
Massachusetts Institute of Technology 1002 1380 1560 1085.4
Case Western Reserve University 1014 1230 1430 1003.6
University of Denver 1142 1070 1290 989.1</p>

89

<p>collegehelp,
You state above,</p>

<p>"You point out that students attend a school in the midst of an entire student body, not just the top 1000. There is some truth to that."</p>

<p>I think there is more than "some" truth to that. This is not like Penn State and having campuses all over the state of Pennsylvania. They all go to school together in the same place and on the same campus and their lives intersect in many, many ways both in and out of the classroom. </p>

<p>Re your argument that there is value in the SAT level of the 1000th student, no one can stop you from having this view, but there are some pretty large holes and IMO you make a pretty shallow argument. Using your approach, EVERY school in the Big 10 conference would rank in the Top 50 and the best would be U Illinois. </p>

<p>Some of the notable "losers" according to your methodology would be </p>

<ol>
<li>Case Western</li>
<li>MIT</li>
<li>Wake Forest</li>
<li>Lehigh</li>
<li>U Rochester</li>
<li>Rensselaer</li>
<li>Dartmouth</li>
<li>William & Mary</li>
<li>Johns Hopkins</li>
</ol>

<p>And your methodology also produces the following:</p>

<p>Rutgers (1340) = Columbia (1340) = Princeton (1340) and</p>

<p>Iowa State (1341) = Georgetown (1341) and</p>

<p>U Tennessee (1278) = Emory (1282) </p>

<p>There are plenty more ridiculous examples. It would be extremely hard to swallow your implied conclusion that this 1000th student measurement is an accurate reflection of student quality strength at these various schools.</p>

90

<p>"The middle 50 SAT range was 1310-1510"</p>

<p>OK, then, Collegehelp, as a check/point of reference you know how many students scored above 1510 at these three schools, anyway. For that one entering class. You can use your normal assumption to estimate how many additional would be between 1500 and 1510. This still doesn't help perfectly, since you are lumping in all these separate colleges together. Because the other four colleges, taken together, are quite large and undoubtedly have high scorers as well. But it still might be interesting.</p>

<p>There are probably other schools out there having a 75 %ile of exactly 1500. You can benchmark your estimates on some of these schools. Or, if the 75 %ile is close to 1500 for many other schools, you can use the procedure I outlined above to estimate just the gap between the 75%ile and 1500, and then use the real 75%ile and benchmark from that.</p>

<p>To the extent that your distribution assumption may not have much basis, it would be far less objectionable to use it only to estimate the gap between 1500 and the real 75%ile, and then use the real 75 %ile for the rest. Particularly since the size of the high-end tail is, in part, what is most in question.</p>

<p>As for this thread, ranking depends on criteria. If your sole criteria is aggregate selectivity of entering classes that is a perfectly legitimate position. (Assuming the data for each college having separate admissions is broken out, if you feel that is appropriate). However, in that case there is no point whatsoever to this thread; just get the data and display it.</p>

<p>I make no opinion as to "ranking"; however, as an aside I find Collegehelp's approach, if based on data, to be potentially instructive to various consituents, for various purposes. Whether ranking is one of those purposes is up to the ranker.</p>

<p>High-capability students may want to know whether there will be many others like them at a particular university, even if the majority there have a somewhat lesser profile. Choice of majors, schools, Honors programs, etc. may in some cases result in their being lumped together in an identifiable cohort within the university, which segregates them from the weaker masses to varying degrees. </p>

<p>Employers may want to know whether there are large numbers of high-capability students to recruit. They can, and often do, seek to segregate out the weaker students, by selecting particular colleges of a university to recruit from, and/or by establishing minimum GPA requirements or otherwise pre-screening applicants they will talk to. This is precisely the approach used by one of my employers to recruit at certain flagship state universities. Where we found great people.</p>

<p>Finally people seeking to understand trends can use this information to explain why certain schools have large representation, in absolute numbers, at top graduate and professional programs. Though their 25-75% numbers may be lower than some other schools.</p>

<p>The issue for each applicant is were can he/she best meet individual objectives. That will require an individual ranking, based on you. Each person will only be one data point on these SAT distributions. In the end, you will each be evaluated individually.
IMO.</p>

91

<p>Collegehelp, there is actually a pretty large overlap between Cornell and Dartmouth and the other Ivies. You are rejecting every argument against this silly way you have decided to make Cornell look better than Harvard and all the other Ivies.</p>

<p>Cornell published an internal report a couple of years ago where itself it admitted that it loses about 85-90% of cross admits to HYP, 2/3-75% to Dartmouth, Brown, Columbia, and Penn and splits 50/50 with Chicago and Northwestern.</p>

92

<p>"a couple of years ago "</p>

<p>yes, and a couple of years ago Cornell was barely getting 19,000 undergrad applicants while now the school is over 30,000. Data changes, big deal.</p>

93

<p>All the Ivies have gone up significantly in applications. The study was from 2004, Cornell apps have not increased 11K since then.</p>

94

<p>" Cornell apps have not increased 11K since then."</p>

<p>They've gone up around 7,000 since then (it's been something like 4K more apps a year for the past 2 years, it was 20K, 24K, 28K, then something like 31K). If the study was from 2004, then the latest data would have been from 2003. </p>

<p>Again, you just have a personal agenda against Cornell. Countless data has been thrown at you in the past, and you refuse to recognize any of it.</p>

95

<p>slipper-
I don't think this method is silly. It answers a particular question: "Which schools have the most smart students in terms of raw numbers?" This is a meaningful question. It allows comparisons based on raw numbers of smart students. (By implication, the higher the SAT of the 1000th freshman, the higher the raw number of smart students defined as students with SATs above a certain level.)</p>

<p>hawkette-
The results are surprising. The data suggest that there are just as many smart students at Rutgers, Columbia, and Princeton. This is surprising but probably true. That's what I find so cool about this way of looking at things. It is a new way of comparing schools (and legitimate).</p>

96

<p>If you believe in a methodology that equates Rutgers to Columbia and Princeton and that is what you're hanging your argument on, then God help you because you are going to need it....... :rolleyes:</p>

97

<p>hawkette-
You are not thinking hard enough about this. The point has escaped you.</p>

98

<p>hawkette, just to follow up, what my method is saying is that schools which are different in overall selectivity can actually be equivalent in terms of the numbers of outstanding students enrolled. In fact, schools which have a lesser overall reputation may actually be superior in terms of the numbers of outstanding students. This intrigues me. I am pondering the implications.</p>

99

<p>collegehelp,
I got your point. I just think that the measurement has extremely low utility. It is sort of like comparing Hershey to Godiva. Hershey makes a heckuva lot more chocolate and some of it is probably very good, but that does not make it as good as Godiva which has a true luxury brand (not to mention a taste to die for :) ).</p>

100

<p>collegehelp,</p>

<p>while i understand your argument, i don't really agree with it.</p>

<p>in order to compare apples to apples between two schools with significantly different class sizes would be to compare things proportionately - either per capita or by percentage. </p>

<p>To take an extreme example -- the way you are measuring things, you could conclude that China is a richer country than America because its top 300,000,000th citizen is worth more than America's top 300,000,000th citizen... but when you realize that America's entire population is 300,000,000 --> meaning that we are actually talking about the very last person in America who has a net worth of ZERO (or more to the point negative net worth) vs. the 300,000,000th citizen of China whose position granted may be very far from the richest person in China, it is still much closer to the richest person vs. that person's position vis a vis the poorest person in China -- in fact, the 300 millionth person still places in the top 25th percentile given China's gargantuan population of 1.3 billion... I hope that made some sense... </p>

<p>At any rate, though I have seen the numbers myself, i'd say, instead of taking the 1,000th student --> a fairer comparison would be to, say, match up the midpoint averages of both and see how do they match up.</p>