As you know, many colleges have dropped the mandatory SAT, a standardized test that has two parts: verbal comprehension and mathematics. Each part is scored from 200-800, so the lowest possible score is 400 and the highest 1600.
At many colleges, submitting SAT test scores for admissions has been eliminated or made optional—often during the pandemic—under the assumption that giving scores would disadvantage racial minorities, who don’t test as well as do white or Asian applicants. This was a way to achieve diversity—a way to enact “holistic admissions.” Even though SAT scores were good predictors not only of college achievement, and of later-life success, measures of potential achievement were considered less important than indices of diversity.
The University of California commissioned a study to test this, and, sure enough, SAT scores were found to be better predictors of “success” than were high-school grades. Nevertheless, the whole UC system eliminated standardized test scores and still won’t consider them.
Now the highly-rated Dartmouth College in New Hampshire has done a similar study, found the same correlative predication as did the UC system, and has reinstated the requirement for SATs, something it made optional during the pandemic. Not only will that facilitate the admission of students who will do well, but they found that making tests option had actually reduced diversity, because lower-income and disadvantaged students withheld scores that would have helped them get in.
Click screenshot below to access, or, if it’s paywalled, I found the article archived here.
Some excerpts:
Last summer, Sian Beilock — a cognitive scientist who had previously run Barnard College in New York — became the president of Dartmouth. After arriving, she asked a few Dartmouth professors to do an internal study on standardized tests. Like many other colleges during the Covid pandemic, Dartmouth dropped its requirement that applicants submit an SAT or ACT score. [JAC: Four-part ACTs are alternatives to SATs.] With the pandemic over and students again able to take the tests, Dartmouth’s admissions team was thinking about reinstating the requirement. Beilock wanted to know what the evidence showed.
“Our business is looking at data and research and understanding the implications it has,” she told me.
Three Dartmouth economists and a sociologist then dug into the numbers. One of their main findings did not surprise them: Test scores were a better predictor than high school grades — or student essays and teacher recommendations — of how well students would fare at Dartmouth. The evidence of this relationship is large and growing, as I explained in a recent Times article.
A second finding was more surprising. During the pandemic, Dartmouth switched to a test-optional policy, in which applicants could choose whether to submit their SAT and ACT scores. And this policy was harming lower-income applicants in a specific way.
Why?
The researchers were able to analyze the test scores even of students who had not submitted them to Dartmouth. (Colleges can see the scores after the admissions process is finished.) Many lower-income students, it turned out, had made a strategic mistake.
They withheld test scores that would have helped them get into Dartmouth. They wrongly believed that their scores were too low, when in truth the admissions office would have judged the scores to be a sign that students had overcome a difficult environment and could thrive at Dartmouth.
As the four professors — Elizabeth Cascio, Bruce Sacerdote, Doug Staiger and Michele Tine — wrote in a memo, referring to the SAT’s 1,600-point scale, “There are hundreds of less-advantaged applicants with scores in the 1,400 range who should be submitting scores to identify themselves to admissions, but do not under test-optional policies.” Some of these applicants were rejected because the admissions office could not be confident about their academic qualifications. The students would have probably been accepted had they submitted their test scores, Lee Coffin, Dartmouth’s dean of admissions, told me.
The article gives the range of test scores between 1300 and about 1550 (remember, 1600 is the highest), and the chances of students getting into Dartmouth that have a given score; this is divided into “advantaged” students and “disadvantaged” students, not really defined but implied as having come from “poor neighborhoods or troubled high schools.” Those are surely correlated with race, but “disadvantaged” is not equated to “black or brown”. The data aren’t biased because, after an admission offer is made or not made, colleges are entitled to look at the SATs of all students who took them, as they’re a matter of record in this way. Below is the graph that the NYT gives:
One thing that strikes me about this is how damn selective Dartmouth is. I did pretty well on my SATs taken in 1965 (a total of 1512: 800 in math and 712 in English), but that would put me in the “advantaged” class having only about an 8% chance of being admitted. However, with scores in the 1400 range, disadvantaged students would have doubled their chances of being admitted. Apparently disadvantaged students didn’t know that, and so withheld those scores, which are still in the upper 5% of students taking the SAT!
Dartmouth’s results also dispelled two common criticisms of the SATs:
For instance, many critics on the political left argue the tests are racially or economically biased, but Beilock said the evidence didn’t support those claims. “The research suggests this tool is helpful in finding students we might otherwise miss,” she said.
I also asked whether she was worried that conservative critics of affirmative action might use test scores to accuse Dartmouth of violating the recent Supreme Court ruling barring race-conscious admissions. She was not. Dartmouth can legally admit a diverse class while using test scores as one part of its holistic admissions process, she said. I’ve heard similar sentiments from leaders at other colleges that have reinstated the test requirement, including Georgetown and M.I.T.
Note that they’re not using race to increase students’ chances of admission, but “disadvantage,” and that is legal, even if race is associated with “disadvantage”. The evidence is, however, that had scores been mandatory, and the gap above maintained, Dartmouth would have increased its diversity.
In the end, a school has three choices: not use SAT scores or consider them for admission; make their submission optional so that they are considered as one factor for admission; or make them mandatory, and they’re considered for admission. The first choice eliminates a very important predictor of college success; the second, which was what Dartmouth used until now, partly eliminates the predictors but also may reduce diversity itself, since students don’t know what the graphs like the one above look like, ergo how their scores could affect their admissions; and the third is what Dartmouth decided to do.
I think they made the right choice. I have no beef with separating “disadvantaged” from “advantaged” students, so long as “disadvantaged” means truly disadvantaged (I’m not sure that “first generation students whose parents didn’t go to college” can be seen as a “disadvantaged” class). “Disadvantaged” should not be automatically assigned to racial minorities, though.
Dartmouth is a rigorous and highly regarded institution, so I suspect other schools who eliminated SAT requirements or made them optional may reconsider their decisions.
h/t: Greg
I think Steven Pinker said something like “test scores are a way to separate the smart poor kids from the dumb rich kids.” But at the university where I’m a trustee, even the Dean of the Honors College doesn’t use test scores. Instead he says admission is based on “who we think can succeed.” (I guess he’s saying to those not admitted, “We don’t think you can succeed.”) And lo and behold, those admitted into the Honors College, where they benefit from very small class sizes, discussion format, etc — do generally succeed. Good for them — but the students who are rejected are left wondering why,
It’s worth reminding ourselves that, if we take seriously the evidence from things like twin studies, which we should, then the effect of this sort of advantage/disadvantage on students is less than commonly supposed.
For example: “Academic achievement consistently shows some shared environmental influence, presumably due to the effect of schools, although the effect is surprisingly modest in its magnitude (about 15% for English and 10% for Mathematics) given that this result is based on siblings growing up in the same family and being taught in the same school …” [“shared environment” is defined as all aspects of environment than siblings living together would share] Link
This is an encouraging development…. I’m not really sure why I would care… it’s just maybe the mystique to a far off land seems to be returning to the Earth, some place to yearn for… in another life, anyway…
Well done Dartmouth! This is good for both the students and the school.
The results of the research are nothing new. They have been known for at least forty years. However they were “discredited” by well-wishing folks who wanted the truth to be otherwise. Let us hope the other members of the Ivy League wake up before they are swallowed in the mire.
The Dartmouth statement reflects the fact that college admissions procedures are connected with real world outcomes—how admitted students do in college—and thus cannot avoid what has been called white empiricism in some quarters. On the other hand, a “Centre for Teaching and Learning”, concerned with more ethereal matters, can employ Other Ways of Knowing, as in this statement from the Centre for Teaching and Learning at Montreal’s Concordia University: “…impressing the principles of the Two Row Wampum Belt into the strategic plan creates a path…for faculty to actively engage in evaluating their curriculum and pedagogical practices by challenging and decentering Eurocentric canons of thought.”
After “Woke Kindergarten” it’s good to see some sanity in higher education. Thanks for the post.
It will be very interesting to see how other institutions respond to this. They can respond by reinstating test scores as criteria for admission—as the data would seem to recommend—or they can continue to devalue such scores. This will test the veracity of those who claim that they reject the SAT because it is not a good predictor of college success. Either they use test scores in their evaluations, or they will need to have a good reason not to.
I think we should expect the responses of other institutions to correspond to their public/private status. Dartmouth’s decision is made easier because it is a small (4,500 students), elite private institution that competes for a smaller number of higher aptitude students. It can be selective and still maintain sufficient enrollment numbers.
Public institutions, especially lower-tier state colleges, face a number of challenges that are at least partially mitigated by being far less selective, including: reductions to state funding; state funding tied to enrollment; pressure to lower costs; and a shrinking population of college-aged, college-bound, high school graduates. It may be cynical to think so but the careers of many higher ed admin/faculty/staff workers, and the survival of the institutions themselves, depend on having student butts in seats. Opening the doors to admission ever wider is one way to, at least temporarily, preserve careers and institutions (Yes, at the expense of those students who perhaps should not be in college at all. NOTE: I didn’t say it was ethical. I’m speaking descriptively, not normatively.).
This ain’t all bad. Many very capable young minds find themselves in these schools, in a sea of mediocrity and low aptitude students. To the degree faculty identify and challenge these students (all students, really) to achieve their potential, it’s a good thing. By which I mean, those high aptitude students who nevertheless find themselves on the outside of selective institutions looking in, for whatever reason, are still given the opportunity to reach their potential. It’s an imperfect safety net for those disadvantaged students who could/would have been admitted to more selective schools had their circumstances, or in the case of Dartmouth, the admissions requirements, been different.
Well, the SATs have been “recentered” several times, so 712 in English in 1965 would presumably correspond to a much higher score nowadays.
See, for instance https://www.greenes.com/html/convert.htm
(which is just about the change from the early 90s to the late 90s.
I’ve just finished reading The Bell Curve, after hearing a Sam Harris podcast with Charles Murray.
It’s amazing how many issues discussed in the early/mid-nineties are still very much front and centre. It could quite easily have been written yesterday.
One amusing bit, and probably a red flag, was when they discussed the work of some intrepid young reporter in Washington who was looking into lowered standards for admission into the police. The reporter’s name? Tucker Carlson.
I read this just before Christmas, I found it pretty much depressing. And that was before we got to the chapters on race.
SAT recentering. Grade inflation. It’s America. Where everyone can be an above-average student.
I wonder if one reason for the extraordinarily low acceptance rates at Dartmouth and like institutions is an increase in the average number of applications submitted by each student. In an era in which everything is online, automated, and easy, I’ve heard of kids sending applications to over 30 schools, which was unheard of in my day.
Phew! That’s a damned impressive SAT score, Professor! Supposedly it indicates an IQ of 151.