Using data from the 1980s and 1990s, Epenshade looked at how chance of admission to colleges varied with a vareity of factors, including SAT score and race. He found that the regression coefficient for high SAT scores become smaller in the later studies, indicating colleges were placing the most weight on SATs in the 1980s group and the least weight in the later 1990s group. He writes,
“The SAT gradient flattens over time, suggesting that SAT scores are becoming less influential in evaluating applicants to elite universities. One reason this might be happening is that applicants’ SAT scores have risen, from an average of 1323 in the 1980s…”
This trend of SAT scores becoming less influential has likely continued in more recent years at many highly selective, holistic colleges, as suggested by the Stanford numbers I posted showing little change in SAT score during a period when acceptance rate had dramatic changes. Epenshade suggested the trend of colleges giving less weight to SAT scoes might relate to increasing SAT scores among the applicant pool, which would also suggest colleges continuing to give less weight to SAT scores today.
This thread has also mentioned Epenshade’s estimates of how race converts to SAT points for college admissions, such as Epenshade stating, “Being African American instead of white is worth an average of 230 additional SAT points on a 1600-point scale,” Today there are many differences from the older data such as both SAT scores and race having a different degree of influences on admissions with different relative weightings at different colleges, elite colleges having different applicant and accepted SAT ranges on a rescaled test with a new 2400 scale, elite colleges being generally far more selective, being more common to have superscoring or take the test multiple times, more taking both SAT and ACT and choosing best score, the test optional movement, etc. This gives little meaning to the specific numbers.
Data10 interesting theory. Just found that CC has an Academic Index calculator developed by a former Dartmouth AdCom. It seems to weight SAT much more heavily than HS GPA. I tried varying combinations and the best results in the AI calculator came with high SAT.
It’s my understanding the AI’s primary use is a metric for comparing stats of athletes to non-athletes at colleges within the Ivy athletic conference. It’s quite a stretch to assume that means SAT score is the most important part of the application. College websites and admissions offers usually give quite different information. For example, Cornell’s website states:
Each year, the NACAC does a survey asking a large number of colleges what factors are “considerable importance” to admissions decisions, and each year the top two factors in their survey are grades in college prep classes and strength of curriculum… not test scores.
A is for Admission was first published 18 years ago. That means the information’s about twenty years out of date. In 1995, Dartmouth’s admit rate was around 20%; it’s now around 11%. The SAT has been changed several times since Michelle Hernandez last worked in admissions.
I’m sure she didn’t develop the AI. She wasn’t that senior.
Harvard recently stated applicants need not submit SAT IIs, if it poses a financial strain. That’s a clue that things have changed.
Please don’t cite the book as a source for today’s admissions practices.
I find it very interesting that Ivies do things in reverse.
It is always put forward on CC that if professional teams selected candidates holistically, they wouldn’t be the best as a way to counter holistic admission processes. Interestingly enough the AI prevents competing schools to have any unfair advantage by ensuring that first and foremost, athletes have to meet an academic standard and only so many players with lower academic standards can be admitted at each school. So theoretically, only athletes are expected to meet higher academic standards.
^^ You mean that theoretically, only athletes are expected to meet some sort of academic standard?
Athletes at ivies are not evaluated holistically by admissions. They are supposed to meet academic standards that are roughly inversely correlated with their athletic ability and there is an absolute floor for AI.
the admissions process is holistic but that doesn’t mean that the students are holistic (Renaissance style kids).
A tremendous athlete that is weak in all other areas - though there is limit to how academically weak an athlete can be - can be admitted through the holistic admissions process. I knew a dyslexic math whiz - he got an 800 on that section and breezed through all his math courses and he got a 490 on the Verbal. schools wrote to him but he realized from the literature he received that most schools assumed he was international and that English was his second language. Oops. He got into a top school though.
A holistic process allows well-rounded and pointy kids to get accepted. Ultimately, it allows these schools to admit who they want to admit while protecting to a degree from losing legal challenges. That is the real function and purpose of today’s admissions process.
As far as weight on the SAT, I think after 2000, it doesn’t really distinguish one persons IQ from another. The test isn’t designed that way plus superscoring, multiple retakes and prep classes make it a poor surrogate for IQ tests in that regard.
I disagree that it doesn’t distinguish IQ. When you see students who just can’t get a perfect score, who consistently miss all questions of a certain type, you begin to realize how it sifts them out up there on the higher levels.
Do they reveal any other groups’ academic numbers as required numbers to meet?
Essentially, I see that they are stating that coaches can request LLs based on their athletic ability as long as they meet a specific AI and yes a floor is enforced as a percentage as to how many they can accept with low AIs.
I’ve never heard of published AI’s on student populations at Ivies. Michelle Henandez’ explanations of AI were very helpful to our family (non-athletes) and helped shape the college list, which was very successful.
Based on most accounts, the required AIs for athletes seems to be derived from the average of the AI for each campus with the baseline number of 176 for a very very limited number of athletes.
I may be reaching some wrong conclusions since the schools don’t publish admitted student statistics by race.
One of the conclusions I have come to is that an URM defensive or offensive linesman may be better off applying to school and then try to walk on rather than appky as an athlete because they are expected to meet a higher academic standard if they need to meet 215 to 220 AI.
I have no real basis to reach that conclusion since I don’t know a calculated race based AI for the schools.
I think it’s inherent in the rules for athletic admissions that they have to calculate the AIs for all admitted students, but that doesn’t mean that they use AIs in the admissions process for non-athletes.
If anything, I think the institutional goal of maintaining a quality athletic culture may lead, at least at some colleges, to a mild bias against high-stat applicants. After you have admitted the undeniable “walk-on-water” candidates, who will tend to have ultra-high stats, there could be a tendency to look for exciting students with lower stats, because that will loosen the strictures on athletic admissions a little. Years ago, in the Hargadon era, I saw statistics on Princeton admissions that looked like they were doing that.
Say you have a student with a 600 SAT and another with a 2300 SAT. I don’t think that those SAT results would only “modestly” predict a person’s success at a certain school. The only reason SAT has just a modest correlation with GPA is that colleges often admit from a narrow range of SAT scores…it’s very tough to get into Harvard with less than a 2000 SAT, and for most people, 2200+ is necessary. The difference between a 2400 and 2200 is indeed fairly negligible.
So the correlation between GPA and SAT scores is obscured by the fact that people who wouldn’t do well at a top school are simply not admitted and those who would over-perform greatly (i.e. a 2400 SAT student going to a 3rd-tier state university) simply don’t attend these schools. Furthermore, students with higher SAT scores tend to go into more difficult majors, such as engineering and math, within schools, further obscuring the correlation. So a lot of the SAT’s usefulness as a predictor isn’t necessarily easily measured, because it acted as a preventive measure to keep unqualified students out.
But this has nothing to do with the SAT being a bad test. It provides plenty of discrimination academically across the spectrum; it makes a clear distinction between those near the top and those near the bottom, and the results of those tests do predict college performance.
The most valid criticism against the SAT is that it doesn’t discriminate well amongst the top students who are more than ~2 standard deviations above the mean (top 2.5%). In this case, the SAT IIs, AP exams, or perhaps a different test yet to be created would provide that further discrimination.
But the idea that one’s SAT score means nothing is utterly ludicrous. If you have a 1200 on the SAT (all three sections), you’d have a very difficult time at a prestigious school. I can’t see how this wouldn’t be the case.
If a correlation between SAT and college grades exists, then one would also expect a correlation between SAT and HS grades to exist. There is also a correlation between test scores and strength of HS curriculum, quality of LORs, quality of essays, chance of exceptional awards/ECs, etc. So if a college considers other portions of the application besides just SAT scores in their admissions decisions, then they are going to tend to admit higher test score applicants within the pool, even if they don’t consider test scores in admission decisions.
This correlation with other components of the application effect makes can make it awkward to distinguish between correlation and cause of college academic success. For example, suppose a college suddenly went test optional and started admitting applicants with great HS grades + high course rigor, great LORs, and everything stellar on the application except test scores. Is that unknown test score applicant likely to do poorly compared to a similar applicant who had a the same high quality rest of the application + higher test scores? In other words, do test scores add much to the prediction of college academic success beyond the other components of the application colleges use in their evaluation of students? I’ve seen several studies that analyzed this question. I’ve yet to see one suggesting SAT M/CR adds much to the prediction beyond the rest of the application. This is especially true when controlling for both a measure of strength of HS curriculum and HS GPA.
The attack on SAT, ACT etc is an attack on any objective means of assessing IQ…mainly because it is inconvenient for some segments of society. They want NO evaluation of native intelligence. This is just a part of the general attack on excellence that is invading society at all levels, in favor of a bland “equality” based on pride and entitlement. This is a culture in serious decline, and eventually we will end up simply punishing people of ability, and promoting mediocrity because that “feels good” to the majority. Grades are important, of course, but that category allows for inflated grades handed out to some people so they don’t “feel bad”…there is NO means of standardizing grades between schools in markedly different parts of the country. Standardized tests are a necessary criterion. But, of course, that may be inconvenient!
