The 325 most selective schools ranked by SAT 75th percentile

College Search & Selection
61

<p>But this would require ascertaining if the schools do not superscore for admissions purposes, but report the scores on a “best score” basis. That is simply unknown and unproven! Schools that supposedly report non-superscored numbers are also supposed to report the exact number of students in the top 10 percent of the class. Does the UC do that, or relies of best estimates! </p>

<p>The impact of superscores is an entirely trivial argument that has been thrown around as a wet rag. And for not particular reason, except to boost the ego of a few fanboys.</p>

<p>An argument that is prone to be deflated when comparing the ACT versus SAT scores to analyze the variances created by the superscores.</p>

62

<p>

</p>

<p>And why a closer look at the admission data for the UC system tells a different picture in terms of unduplicated applications.</p>

63

<p>An argument that is prone to be deflated when comparing the ACT versus SAT scores to analyze the variances created by the superscores.</p>

<p>I’m not sure I understand the above, but aren’t some schools superscoring the ACT as well?</p>

<p>Anyway…very large state schools really don’t compare well with smaller privates. The large state schools have a mission to educate “the masses”, and they usually offer a wider variety of majors that don’t really require/need high test scores. As I said earlier, they tend to weight GPA more so that their schools don’t exclude kids from underprivileged areas. And, maybe more importantly, they also need lots of “warm bodies” to occupy their 25,000+ seats.</p>

<p>Southern schools will be hurt by using the lower quartile limit. I believe that the southern state schools are still required by law to admit students who have minimum scores…a holdover law from the days of desegregation. </p>

<p>Very large state schools can have split personalities. The STEM side can be very academic with mostly high stats kids in the classes. The more creative side can be filled with right-brained kids whose talents don’t measure well in standardized tests (who cares that the highly talented Illustration or MT majors may have modest SATs? How do they affect the difficulty of the Math and Physics classes???). And, lastly, these schools do offer some lighter majors for kids (maybe some are athletes) who need to graduate in SOMETHING. lol</p>

64

<p>

</p>

<p>Actually, I don’t think we can reach that conclusion. Harvard reports 25th-75th SAT CR of 690-800, and SAT M of 700-790. USNews (and some others) then add up the two figures and report 25th-75th SATs as 1390-1590. But you can’t assume that everyone who was in the bottom quartile in CR was also in the bottom quartile in M.
There are likely some “lopsided” scorers who did better in CR than in M or vice versa.</p>

<p>Suppose Harvard enrolled just 4 students with the following scores:</p>

<p>Student A: CR 680, M 800
Student B: CR 700, M 790
Student C: CR 800, M 710
Student D: CR 800, M 700</p>

<p>Then it would report its 25th percentile CR at 690 (bottom 25% scoring below that figure) and its 75th percentile at 800 (bottom 75% scoring up to that figure); i.e., SAT CR middle 50% 690-800.</p>

<p>It would also report its SAT M middle 50% as 700-790. </p>

<p>If you add those two figures together (as US News and some others do), you’d say “SAT 25th-75th percentile 1390-1590,” which is what US News reported for Harvard for its 2010 entering class. And someone reading that might say, “I can’t believe a quarter of the people Harvard enrolls have SAT CR+M below 1390.” </p>

<p>But that would be a false inference, if an understandable one, based on faulty presentation of the data. If you go back and look at my hypothetical scores, they range from 1480 to 1510. The low score (out of 4) is 90 points higher than the figure you’d get by simply adding the CR 25th percentile to the M 25th percentile, and the high score (out of 4) is 80 points lower than you’d get by simply adding the CR 75th percentile to the M 75th percentile.</p>

<p>Of course, my hypothetical makes everyone lopsided; the real world isn’t like that. In Harvard’s actual admission pool, some people probably have balanced scores in the mid-700s, some have double 800s, others are close to that. But it’s likely that many (or even most of the people they admit with low CR scores balance that off with high M scores, and vice versa. So the 25th-75th percentile data, presented the way they’re usually presented, are potentially very misleading. It’s likely there aren’t 25% of people attending Harvard who actually scores below 1390. And at the top end, it’s likely that not as many as 25% of the entering class scored a 1590 or 1600.</p>

<p>So if you look at that 25th percentile figure and think, “Gee, I’m at least at the 25th percentile, maybe I have a chance,” you might want to think again; you’re probably not as close to the actual 25th percentile as you think, based on the faulty way the scores are reported. By the same token, if you’re toward the higher end but under the reported 75th percentile, don’t despair; the actual 75th percentile might actually be lower than you think.</p>

65

<p>In other words, the USNWR and collegehelp assumption is that the CR and M scores are completely correlated, and we can add the 25th and 75th percentile figures to get the 25th and 75th percentile figures for CR+M.</p>

<p>The bclintonk example is nearly anti-correlated, so the variance is reduced to nearly zero. In fact if purely anticorrelated, the variance is exactly zero.</p>

<p>We can’t safely assume normal distributions for the scores, particularly with Harvard, because we have many scores near the CR and M maximum, but just to capture the mathematics if we do pretend the CR and M scores are normally distributed, then the general solution for the deviation of the CR+M is sigma<em>total = sqrt(sigma</em>CR^2 + sigma<em>M^2 + 2<em>rho</em>sigma</em>CR<em>sigma<em>M). We can make the math a lot easier by assuming sigma</em>CR=sigma<em>M=sigma, collapsing the result to sigma</em>total=1.4</em>sigma*sqrt(1+rho).</p>

<p>So for purely correlated, sigma<em>total=2sigma; for purely anticorrelated, sigma</em>total=0; and for purely uncorrelated, sigma_total=1.4sigma. </p>

<p>Now purely uncorrelated is probably a poor assumption for SAT scores in general, since we would not expect equal chance that someone who scores an 800 CR would score a 200 or 800 math, and certainly not an equal chance of the score being either less than 400 math or greater than 600 math. However, for these limited distributions for school cohorts, uncorrelated is probably closer to the truth than fully anticorrelated or fully correlated.</p>

<p>So if we look at a distribution like Bryn Mawr CR 25-75 = 590-720 and M 25-75 = 580-700, the CR+M distribution might be something like the uncorrelated prediction of 1210-1380, rather than the correlated prediction of 1170-1420. Or returning to the Harvard example (conceding that the distributions are obviously not normal), the uncorrelated prediction would be 1420-1560 rather than 1390-1590.</p>

66

<p>*Actually, I don’t think we can reach that conclusion. Harvard reports 25th-75th SAT CR of 690-800, and SAT M of 700-790. USNews (and some others) then add up the two figures and report 25th-75th SATs as 1390-1590. But you can’t assume that everyone who was in the bottom quartile in CR was also in the bottom quartile in M.</p>

<p>There are likely some “lopsided” scorers who did better in CR than in M or vice versa.*</p>

<p>Very true about some having lopsided scores. I imagine that many int’ls and some ESL students have lopsided scores…high Math, lower CR. </p>

<p>That said, I do remember the H and Y recruiters saying that they set aside a certain number of seats for frosh athletes when they were talking about test scores that everyone else needed… The Y recruiter wrote on a white board and indicated lower SATs for the athletes’ …and I’m vaguely remembering that the numbers she wrote down were in the lower 1300s for M+Cr.</p>

67

<p>Why is it that though we are in 2012, colleges still put SAT scores out of 1600 rather than 2400. Are those SAT lists outdated??? Thx guys.</p>

68

<p>

Not only that, some schools are cross-superscoring the SAT and the ACT.

There are still a lot of schools that don’t really use the Writing section of the SAT, or just use it as a sort of tie-breaker.</p>

69

<p>Why is it that though we are in 2012, colleges still put SAT scores out of 1600 rather than 2400. Are those SAT lists outdated??? Thx guys.</p>

<p>Because many schools really just aren’t using the Writing section that much to determine admittance. My guess is that they don’t want a high Writing score to “make up for” lower Math or CR sections. </p>

<p>My thinking is that schools will think Student A is stronger than Student B…</p>

<p>Student A</p>

<p>Math 740
Reading 710<br>
Writing 650
totals… 1450/1600 … 2100/2400</p>

<p>Math 650
Reading 690
Writing 790
totals…1340/1600… 2130/2400</p>

70

<p>Knowing a former admissions counselor, alot of games can be played with the common data set numbers, super score / not super score, only report the better score of SAT or ACT of admitted students, etc.</p>

71

<p>The correlation between SAT CR 25th and math 25th is .89.
The correlation between SAT CR 75th and math 75th is .86.</p>

<p>CR and math are highly correlated.</p>

<p>The correlation between SAT total 25th and ACT comp 25th is .95.
The correlation between SAT total 75th and ACT comp 75th is .86.</p>

<p>SAT and ACT are highly correlated.</p>

<p>Not sure whether this sheds light on any of the previous discussion (which I did not completely follow).</p>

72

<p>

</p>

<p>Unfortunately we need to know the correlation for individual scores rather than for the group statistics in order to figure the true 25-75 spread of CR+M.</p>

73

<p>Are these scores for admitted or attending students? I couldn’t find a definitive answer in the thread.</p>

74

<p>The stats undoubtedly come from CDS, so that means they are for enrolled students.</p>

75

<p>They are from the IPEDS US Dept of Ed database and they are for the 2010 incoming freshman class (first-time, full-time). That is, attending. </p>

<p>If you are searching for a college, I would suggest avoiding schools where you would be in the top 25% (i.e. above the 75th percentile) because you would be selling yourself short. Unless, perhaps, they give you a free ride. Apply where your scores are in the middle 50% if you want a comfortable challenge. If you want a hard challenge, apply where your scores are just below the 25th percentile but not too far below. Maybe within 15 or 20 points of the 25th. This is, perhaps, the most important factor to consider(among many) when finding a good match.</p>

76

<p>

</p>

<p>I am not surprised. Lani Guinier, in describing elite colleges, said “merit often functioned simply as a handmaiden to power”. </p>

<p>What I am surprised by is the Lauren Rivera study:</p>

<p>[Brown</a> and Cornell are Second Tier - Percolator - The Chronicle of Higher Education](<a href=“http://chronicle.com/blogs/percolator/brown-and-cornell-are-second-tier/27565?sid=at&utm_source=at&utm_medium=en]Brown”>http://chronicle.com/blogs/percolator/brown-and-cornell-are-second-tier/27565?sid=at&utm_source=at&utm_medium=en)</p>

<p>Would it not make more sense to recruit students by their major than by the school they attend? The results would be a lot less “noisy”. Then again, who can say that the system is not purposely designed to be “noisy”.</p>

77

<p>The above article by Lauren Rivera, Asst Prof at Northwestern, was based on her access to one firm and interviews with recruiters from other firms. The article never actually mentions Brown or Cornell which makes you wonder who chose the headline. The article is actually an indictment of recruiters who are narrowly focused on the brand name schools of Harvard, Princeton, Yale. The article suggests, Columbia, MIT, Dartmouth, and Michigan are considered second-tier by such recruiters. From the article:
"the game is rigged. That obtaining a name-brand diploma matters more than actually learning something. That the gatekeepers at our nation’s most prestigious firms are pathetically shallow, outrageously parochial, and insufferably snobbish. "</p>

78

<p>This is why I’m not impressed when people say they went to Harvard or wherever. It’s not that hard to get in if you have connections or the right extracurricular activities. I know a few students who got in, after being rejected during regular admissions, simply because daddy made a phone call.</p>

<p>MIT and Caltech are another matter, because those students are highly specialized in the sciences. The students I’ve met from MIT and Caltech have been far more impressive than those from other schools. The lower 25th for MIT is probably because it stoops lower on the verbal section to accept foreign students who excel in math and science.</p>

79

<p>This study seems to reinforce the finding of another study I recall - that the US is below average in terms of socio-economic mobility when compared to other developed countries. That is, your future success in the US depends more (and the most) on your family background than if you were in most other developed countries (many of them are in Europe). We would like to think this is a land of opportunties and freedom but ironically, it’s the more egalitarian/“socialist” places with much longer history that provide better opportunities for the underpriviledged.</p>

80

<p>There is a strong correlation between selectivity (SAT, freshmen in top 10%, etc) and graduation rate. I noticed that Penn State again exceeded graduation rate prediction (US News overperformance) by 17%. Penn State’s grad rate and SATs are similar to some UCs but its 45% of freshmen in top 10% of HS class is way below the UCs which are in the 95-100% range of top 10% freshmen. Texas A&M has slightly better selectivity stats than Penn State but the grad rate is 7% lower.</p>

<p>Is there a flaw in the US News prediction formula for over/underperformance? The formula seems to work pretty well…the vast majority of schools are within + or - 5% of the predicted grad rate…most are + or - 2%. Is there something about Penn State that enables it to exceed by 17% the grad rate that would be predicted for a school with only 45% freshman in top 10%? Is it possible that Penn State is falsifying data like Emory, Iona, and Claremont McKenna have done?</p>

<p>Interested in hearing your thoughts…</p>