Should we eliminate standard testing for college and graduate-school admissions?

In one section of new post on his website Shtetl-Optimized, computer scientist Scott Aaronson decries the continuing trend of eliminating standardized tests for college admission. Click on the screenshot to read his plaint.

Scott points out that the University of California system has just eliminated the requirement of submitting SAT and ACT standardized test scores with applications. He gives some useful links about this decision:

. . . the University of California system—ignoring the advice of its own Academic Senate, and at the apparent insistence of its chancellor Janet Napolitano—will now permanently end the use of the SAT and ACT in undergraduate admissions.

That’s not exactly correct, at least according to the rules set out below, for non-resident students (those not applying from California) apparently still must take the SAT or ACT, though this isn’t completely clear (non-resident make up about 20% of UC students).

Scott does add, and I think he’s right on the money here, that this will lead to a cascading effect whereby virtually every college in America will ditch the requirement for such tests. However, the state gives itself five years to develop a standardized test for California residents only, and if it can’t do so, it’ll drop testing requirements completely. You can still submit your SAT and ACT scores for two more years, but after that they won’t be looked at. There are a few exceptions, which I’ve put in bold below:

Here are the new University of California rules from the San Diego Times:

  • For all students, the SAT and ACT will be “test-optional” in the admissions process until 2022. Students who don’t submit a test score won’t be penalized.
  • For the 2023 and 2024 school years, the UC will not consider test scores from California students for admissions purposes.
  • California students can still submit test scores to become eligible through the “statewide guarantee admissions,” which combines high school grades and test scores to give students a spot in any campus that has space if the student is in the top nine percent of applicants.
  • Students can submit their test scores for certain scholarships and placement in courses.
  • Out-of-state students will be governed by the “test-optional” rules until 2024.
  • The essay and writing portions of the SAT and ACT requirements are  dropped beginning next year.
  • Non-resident students may still have to take the SAT or ACT in 2025 and beyond. They may also be able to take UC-designed test if it’s ready in time. The rules for non-resident students, who make up nearly 20 percent of undergraduates, are an open question.

Scott objects to the new rules, as do I, but for different reasons. He sees it as eliminating a “backdoor” route for kids who are really smart but don’t have good grades for reasons not connected to their ability to succeed in college. (Maybe they weren’t interested in a high-school curriculum or put their energies into other areas.) As Scott notes:

As a result, admissions to the top US universities—and hence, most chances for social advancement in the US—will henceforth be based entirely on shifting and nebulous criteria that rich, well-connected kids and their parents spend most of their lives figuring out, rather than merely mostly based on such criteria. The last side door for smart noncomformist kids is now being slammed shut.

. . . In case it matters to anyone reading this, I feel certain that a 15-year-old me wouldn’t stand a chance in the emerging regime—any more than nerdy Jewish kids did in the USSR of the 1970s, or the US of the 1920s. (As I’ve previously recounted on this blog, the US’s “holistic” college admissions system, with its baffling-to-foreigners emphasis on “character,” “leadership,” “well-roundedness,” etc. rather than test scores, originated in a successful push a century ago by the presidents of Harvard, Princeton, and Yale to keep Jewish enrollments down. Today the system fulfills precisely the same function, except against Asian-Americans rather than Jews.)

Now everyone knows the reasons why test scores are being abandoned: because putting any weight on them could discriminate against minority students like African-Americans and Hispanics, who perform lower than do members of other groups, and also discriminate in favor of Asian-Americans, who score the highest. In other words, to achieve “diversity and inclusion”, you eliminate those factors that militate against minority admission. Scott thinks this will push out America’s academically talented underdogs.

What he doesn’t mention, and what is pretty well known, is that high school grades themselves are a marginally better predictor of success in college than are SAT or ACT scores, and in fact are the single best predictor.

My own objection to eliminating testing is twofold. First—and I’m not sure if the research has been done on this by anyone other than test companies—it seems likely that using a multivariate predictor, which incorporates grades, test scores, and other factors (perhaps an index of the high school’s difficulty?), would be even better. Which combination of factors best predicts college success? That’s what the UC is doing in the third bullet-point above.  But again, since minority students have lower test scores, that would tend to drag down their multivariate score.

If you want greater racial equity, though, it seems to me best not to eliminate test scores, but to calculate a multivariate index of “academic achievement,” and then use other criteria, like “diversity points” to increase racial balance. This is, in effect, what is being done now by schools like Harvard. The reason, as I’ve said before, is as a form of reparations for those held back by their sociopolitical history in America.

Second, using standardized tests can help single out people like Scott who probably was at the very top for SAT or ACT, but had not-so-stellar grades. This would be a way to find those who fall through the net. And although Scott is white and Jewish, it would hold for anybody of any race whose grades don’t align with their test scores.

I believe, though I can’t be arsed to look it up, that the value of graduate exams like the GRE in predicting success in graduate school is palpably greater than are the value of SATs and ACTs for undergraduate school. But I predict, based on what’s happening, that graduate schools too will gradually phase out standardized testing as a way to evaluate applicants—and for the same reason. I would oppose that as well.

h/t: Paul



  1. boudiccadylis
    Posted May 26, 2020 at 1:29 pm | Permalink

    I don’t have a suggestion regarding all the hullabaloo to getting in college today. When I went, I chose the school and applied. To the best of my knowledge the only thing required was high school transcripts.
    Should I apply in today’s educational system I’m afraid I wouldn’t make it.

  2. Posted May 26, 2020 at 1:40 pm | Permalink

    “If you want greater racial equity, though, it seems to me best not to eliminate test scores, but” … [may I suggest] to get into primary and secondary schools and make sure these African-American and hispanic students get the training they need to do well on the tests.

    • Adam M.
      Posted May 26, 2020 at 2:03 pm | Permalink

      While I doubt the possibility of eliminating those disparities, I too would much rather they try to raise underperforming groups up to the standard rather than lower the standard for everyone.

  3. eric
    Posted May 26, 2020 at 1:57 pm | Permalink

    He sees it as eliminating a “backdoor” route for kids who are really smart but don’t have good grades for reasons not connected to their ability to succeed in college. (Maybe they weren’t interested in a high-school curriculum or put their energies into other areas.)

    I was one of those kids (though the “really smart” label may not be right…).

    I went to tough schools in Australia where getting a B meant something. I was in the top 5% of my class with a GPA of something around 3.0 (out of 4.0…I’m a fossil…). I came to the US for junior and senior year, got mostly straight A’s taking AP classes and college prep courses, and again scored in the cop 5% of my class. My high school counselor, in their infinite wisdom, averaged my GPAs for all four years and declared I was in the top 30% of the class. I’m very, very glad there was the SAT to show the schools I applied to that this was inaccurate.

    So yes Jerry, it can happen, for various good reasons not just ‘putting their energies into other areas.’


    Any state that uses this as a case to get rid of scoring is probably not reading the fine print well enough. The Califoria ‘special admissions’ program that guarantees automatic admission to the Ca kids who meet the requirements is AIUI a very big deal. Thus, even though they are omitting the standardized tests from the general admission, in practice I would bet that this means nothing, and that most Ca students still take them, because they want that guaranteed admission track.

    • Posted May 26, 2020 at 6:15 pm | Permalink

      in their infinite wisdom, averaged my GPAs for all four years

      Hilarious! Usually it takes a computer algorithm to reach that level of blind stupidity. I mean … I’m sure any admissions process that is workable in practice will fail to be perfectly fair … but still.

  4. DrBrydon
    Posted May 26, 2020 at 2:01 pm | Permalink

    Well, I think it’s just a way to interject more subjectivity into admissions. To the extent that college is no longer objectively about an education, that’s probably fine. It makes it easier to obscure inquiries into graduation rates.

  5. Charles A Sawicki
    Posted May 26, 2020 at 2:10 pm | Permalink

    Getting rid of standardized tests is a bad idea, though it does make it easier for admissions officers to avoid the work of finding exceptional students with lower scores. The actual problem which needs to be addressed is the terrible educations many poor kids get before college. Fixing this inequality will be difficult, long term and expensive. It’s the only way to achieve true equality of opportunity.

    • Deodand
      Posted May 27, 2020 at 2:24 am | Permalink

      They don’t want equality of opportunity, they want equality of outcome. As with offense, subjective criteria are the best criteria to use in these cases.

  6. Ken Pidcock
    Posted May 26, 2020 at 2:10 pm | Permalink

    Elite universities can certainly succeed with test-optional undergraduate admissions. Though acknowledging it is not always comfortable, elite colleges and universities are elite for their admissions standards more than for their academic standards, which aren’t graduated all that steeply across the hierarchy. Hence the number of Podunk U graduates in allopathic medical colleges. Fact is that there are many more adults capable of high academic achievement than demonstrated it as teenagers.

    • eric
      Posted May 26, 2020 at 6:51 pm | Permalink

      Fact is that there are many more adults capable of high academic achievement than demonstrated it as teenagers.

      I agree with this. After setting a reasonably high bar for academics, there’s probably no justification for setting it higher – other than as an artificial means of making an entry decision. But if you don’t, you’ll need non-academic criteria, which puts the pressure on the kids to do loads of extra-curricular activities, which puts us back in the same problem.

      IMO there’s a supply problem here (not enough good school spots for many good kids). This is exacerbated by problem of kids applying to loads of schools, which AIUI makes it difficult for the schools to figure out how many they need to accept to get the incoming class size they want.

  7. Adam M.
    Posted May 26, 2020 at 2:15 pm | Permalink

    I think things would be a lot simpler if they just honestly and straightforwardly pursued their objectives – although their objectives are probably illegal and they’re trying to disguise them.

    They clearly have some desired demographic mix in mind that trumps most other considerations, which is why they can say there are too few of this group and too many of that group… Just mandate that mix rather than performing contortions to arrive at it through backhanded methods. Then we can keep the SAT and ACT and they can take the top N% of students in each racial group, etc. Seems better and simpler all around.

    Sure, it may be illegal, but it seems better for them to openly violate the law than to sneakily violate the law.

    • Posted May 26, 2020 at 3:10 pm | Permalink

      If only honesty were still the best policy.

      • Jon Gallant
        Posted May 26, 2020 at 5:41 pm | Permalink

        Ever since the Supreme Court banned outright
        racial quotas in pursuit of “affirmative action”, college admissions offices have devised endless, devious kinds of sophistry and dishonesty to contrive work-around equivalents of racial quotas. One could suspect that this strategy is related, as both cause and effect, to the atmosphere created by overt sophists (like Stanley Fish), and of the fashionable post-modern notion that no simple statement actually means what it states, but only what an “interpretive community” can twist it into.

    • eric
      Posted May 26, 2020 at 6:56 pm | Permalink

      Set reasonable academic requirements predictive of success, and put every applicant that achieves them in a lottery. Your demographics will thus likely reflect the demographics of your applicant pool, but that’s very likely much better than selecting ‘best of the best’, which tends to privilege wealthy and therefore predominantly white families because they have the time and money for all the extracurriculars, tutors, etc. their kid might need to look good on paper.

  8. Posted May 26, 2020 at 2:21 pm | Permalink

    Actually, according to the UC’s own study, standardized test are better than high school grades, although not decisively so; the predictive advantage of test scores over high school grades is increasing; test scores are the best predictor for disadvantaged students; and high school grading standards vary widely, and are inflated:

    The STTF found that standardized test scores aid in predicting important aspects of student success, including undergraduate grade point average (UGPA), retention, and completion. At UC, test scores are currently better predictors of first-year GPA than high school grade point average (HSGPA), and about as good at predicting first-year retention, UGPA, and graduation.(3) For students within any given (HSGPA) band, higher standardized test scores correlate with a higher freshman UGPA, a higher graduation UGPA, and higher likelihood of graduating within either four years (for transfers) or seven years (for freshmen). Further, the amount of variance in student outcomes explained by test scores has increased since 2007, while variance explained by high school grades has decreased, although altogether does not exceed 26%. Test scores are predictive for all demographic groups and disciplines, even after controlling for HSGPA. In fact, test scores are better predictors of success for students who are Underrepresented Minority students (URMs), who are first-generation, or whose families are low-income: that is, test scores explain more of the variance in UGPA and completion rates for students in these groups. One consequence of dropping test scores would be increased reliance on HSGPA in admissions. The STTF found that California high schools vary greatly in grading standards, and that grade inflation is part of why the predictive power of HSGPA has decreased since the last UC study.

    The report’s really a pretty devastating indictment of the Regents’ action; it was avaialble to them before they took it.

  9. denise
    Posted May 26, 2020 at 2:29 pm | Permalink

    I have a problem with a public college rating people’s “character,” “leadership” or “well-roundedness”. I know they already do it, but I don’t like it. I don’t think the State of California should be making subjective judgments on their personalities before providing services to residents who have paid for them through their taxes.

    Especially in a school the size of UC. With 40k students at Berkeley, what difference does it make to anyone if they’re all well-rounded or future leaders? There is room there for every kind of individual – and there will be every kind, if you just admit them based on academic ability.

  10. enl
    Posted May 26, 2020 at 2:34 pm | Permalink

    Well, Rutgers has joined the trend (though there is little surprise in this, as the writing has been on the wall for several years)

    The problem I see with this is that it puts students in challenging secondary programs at a disadvantage, and puts the programs themselves at a disadvantage in recruiting strong students.

    I stopped reading the correlation studies re grades, standardized scores, and secondary grades a while back, because I got tired of just listing the flaws, such as not controlling for difficulty of program of study, to pick an obvious one (you can’t tell me that an major in management or art history, to name a couple, is even close to Physics, Microbiology, or most engineering in difficulty)

  11. Roo
    Posted May 26, 2020 at 2:41 pm | Permalink

    I think the issue here is that this hands a clear advantage to kids in families where mom and dad are always on the school portal, emailing the teacher, asking about extra credit, and heads will roll if their darlings don’t get straight A’s. This is, from what I understand, a very common parent-teacher dynamic in middle class and upper middle class communities. Children from impoverished communities where parents are struggling to make ends meet, working night shifts, etc., may not have that advantage.

    My prediction is that if they are phased out (which they may not be, if the first few years prove to be problematic in figuring out admissions) something will replace the SAT/ACT tests within a decade, probably sooner. It may start out as something like a test to assess areas of interest and talent in order to help students to decide what they should major in as they prepare to college, then move to “a lot of colleges actually like you to include those test results in your application now”, to “I hear your chances are really better if you include those test results”, and so on.

  12. Curtis
    Posted May 26, 2020 at 3:34 pm | Permalink

    On thing that is often overlooked is the use of SATs for students evaluating colleges. My daughter is an excellent student, wants to be a mechanical engineer and hopes to be a Division 2 or 3 college athlete. She is often recruited by schools that we are unfamiliar with.

    In less than 5 minutes we can eliminate most colleges by looking at their SAT scores and whether they have an engineering school. It allows her to concentrate her efforts on schools that are appropriate for her. We have been surprised by the quality of some of the schools that we had never heard of.

  13. Posted May 26, 2020 at 4:41 pm | Permalink

    I’m applying to grad schools this year and the GRE and other standardized tests are already starting to become either unnecessary due to easy-to-meet requirements making them easy to waive or, in many cases, schools looking at them as a formality and not caring how well you do(as long as you are in the 50th and above percentiles). This may not be the case for many of the top tier schools but for the middle tier schools where my humble gifts leave me it definitely is.

    • eric
      Posted May 26, 2020 at 7:09 pm | Permalink

      My limited experience discussing it with my fellow graduate students, a looooong time ago, was that “grades, general scores, subject score, recommendations – pick three” was a good approximation. As long as you had three of them well in hand, lagging in the fourth didn’t hurt you too much.

  14. Historian
    Posted May 26, 2020 at 5:00 pm | Permalink

    Among the variables that need to be taken into consideration when discussing the value of test scores in determining college admission is test preparation. It seems common that wealthier parents send their children to test preparation schools. I’m not sure how much difference these classes make in upping scores on average. But, many years ago, I took what was called the Preliminary Scholastic Aptitude Test (PSAT). My scores were terrible. I did not go to a test preparation class, but I bought several test preparation books and studied them diligently, particularly the section on doing the math problems. As a result, when I took the regular SAT, my scores shot up dramatically. What I took away from this is that these tests do not measure aptitude, but achievement. At the time I took the test, the College Board claimed that these tests could not be studied for. That was pure bullshit. I do not know if it still holds to that position. In any case, if these tests do, in fact, measure achievement instead of some inherent aptitude, then those who study for them are at a distinct advantage.

    • Jon Gallant
      Posted May 26, 2020 at 6:15 pm | Permalink

      That is an interesting, and not entirely surprising, account, Historian. I understand that in Japan, college admissions are wholly dependent on scores in highly competitive national and college-administered exams of the same kind—and that students spend their last year(s) of secondary school frantically preparing for these tests. Those who do not
      do adequately on them spend an extra year after high school preparing for the next round of exams again. See: .

      If tertiary education in the US goes in the
      opposite direction, eliminating the SAT and ACT test requirements, this will increase the importance of grades, and thus the authority of secondary schools; and thus, in turn, the status of the Schools of Ed which monopolize the credentialing and training of the teachers. Could this have something to do with the current trend to de-emphasize SAT?

    • enl
      Posted May 26, 2020 at 6:34 pm | Permalink

      Even without prep, the SAT scores are generally much higher than PSAT.

      The PSAT is not really a “practice” or “preliminary” SAT. It is a contest exam. It is the qualifier for the national merit scholarship, and at the top end it is spread for differentiating, rather than a good fit to the normal curve (like the SAT and ACT nominally are)

    • enl
      Posted May 26, 2020 at 6:38 pm | Permalink

      This is normal.

      They stopped the claim studying doesn’t help thirty years ago (roughly), and the PSAT is not a “preliminary” SAT as a predictor. It is a contest- scholarship- exam. It is the qualifier for the National Merit Scholarships, and, at the top end, the scale is spread for differentiation, unlike the SAT and ACT.

      • enl
        Posted May 26, 2020 at 6:39 pm | Permalink

        (sorry for the repeat… Windows decide to be peevish so I retyped…. fifteen minutes later after restart)

      • Posted May 27, 2020 at 12:55 am | Permalink

        Mensa, for one, used to take all manners of tests as proof of IQ when those tests were more highly correlated to IQ but stopped taking those tests once that correlation faded. They will not accept ACT scores newer than 1989 for example, PSAT post-1993 or SAT post 1994. This probably is also a reflection of the ability to study for the test as you shouldn’t be able to study for an IQ test.

        On a personal note, I always could crush any standardized test except the PSAT where I assume I got the base points for spelling my name correctly and nothing else.

    • eric
      Posted May 26, 2020 at 7:19 pm | Permalink

      Agreed, they measure ability to learn how to take the test.

      But what you did to achieve those higher scores was self-motivated study. Which, IMO, IS a good skill colleges want to see in their applicants, and (IMO) should be reasonably predictive of success in college classes, graduation, etc. So as long as a standardized test isn’t too arbitrary in it’s content, using it to predict performance seems to be okay to me, because it’s really measuring the student’s willingness and ability to have a new academic challenge placed in front of them and work at it to meet it. What to do with kids who get high scores because their parents harangue them into it, but have lousy study habits on their own? I have no idea. There’s likely no proxy metric that gets everything right.

  15. Roo
    Posted May 26, 2020 at 7:31 pm | Permalink

    On a more cynical note, a part of me wonders if there won’t be a concerted effort on the part of test makers and the test prep industry to woo colleges back. Again, a cynical take, but you know the saying – money talks.

  16. Jimbo
    Posted May 26, 2020 at 7:43 pm | Permalink

    So if SATs and ACTs are eliminated and GREs follow, does that mean LSATs and MCATs will too? Surely members of the ABA and the AMA in prestigious university law schools and medical schools will raise a stink within their departments about softening these admission standards, no?

  17. Gareth Price
    Posted May 26, 2020 at 8:59 pm | Permalink

    This a slight tangent but I have come up against a big barrier to returning to grad school in the US. My degrees, including a PhD, are from UK universities which did not produce US-style transcripts (or anything that is recognizably a transcript) when I was a student. Nearly everywhere that I have enquired has told me that they couldn’t accept me without transcripts. As pretty much the only people who wouldn’t be able to produce transcripts are foreign nationals, I am not sure that such a policy is even legal.

    • David Harper
      Posted May 27, 2020 at 6:18 am | Permalink

      I was a graduate admissions tutor at one of the colleges of the University of London for a few years in the early 1980s, and we took students from around the world, so we needed some way to compare bachelor-level degrees from different countries. There was a big reference book which listed, by country, all of the types of qualification, and their equivalent in the British degree classification system.

      These days, the European Credit Transfer and Accumulation System (ECTS) allows degrees from universities across Europe to be directly compared. Every module/course has a level and a number of credits associated with it.

  18. Mike
    Posted May 27, 2020 at 5:25 am | Permalink

    Steve Hsu has blogged about the predictive validity of GRE and SAT previously. I have seen some places also dropping the GRE which appears to be a way of manipulating admissions to achieve an arbitrary racial bean counting requirement.

  19. Bruce Kendall
    Posted May 27, 2020 at 2:29 pm | Permalink

    As possibly the only person who translates SAT and ACT tests, maybe. But, before eliminating them, the public should first understand them. And not in the way the College Board or ACT wants you to. My job as a translator is to show and tell everything to the stakeholder, that is normally not told or shown.

%d bloggers like this: