Send it. An 800 math score is fantastic and should be shared. I don’t think a 10-30 point difference from the 50% is at all meaningful.
2 Likes
Indeed.
“Our research has shown that, in most cases, we cannot reliably predict students will do well at MIT unless we consider standardized test results alongside grades, coursework, and other factors. These findings are statistically robust and stable over time, and hold when you control for socioeconomic factors and look across demographic groups. And the math component of the testing turns out to be most important.”
1 Like
I have often repeated my nephew’s quote that at Yale “It can be hard to get an A, but it’s harder to get a C.”
Harvard is the same way, and the most common grade is an A-. Spend minimal effort, and you can get a B. Among my son’s friends, he only knows of one person who got a C, and that’s for accidentally not turning in a final exam (the student has all As otherwise, so that C that sticks out like a sore thumb).
In other words, it is really difficult to fail a class at Harvard. There is also plenty of help for struggling students, so either a student is completely unqualified, or they are going through something that is preventing them from doing even a minimal amount of effort for the class.
1 Like
Also, this virtual / online / distance education was an unplanned emergency switch for most, so both the schools and students may have done worse than where students chose a school which has had years of experience doing virtual / online / distance education as a planned mode of instruction.
1 Like
My child attends a highly regarded NYC private school. The school does not release information about the 25/50/75 percentile of its students (or maybe they do quietly in junior spring)
However, I noticed that Horace Mann – generally considered to be one of the better NYC privates – did release this information in its school profile. I was surprised to see that for 2022, HM’s middle 50% was 690/760, and for math it was 690/780.
The school’s self reported verbal mean was 712 and its math mean was 720. The school did not release overall composites, but if you just made some simplifying assumptions, you would conclude the mean composite score was 1430, and the top 25% score was 1540.
These scores were a lot lower than I would have thought a school like HM would have. Granted, they are one of the larger if not the largest top tier NYC private school, so it’s hard to curate a high mean/median score as the population size increases. But still, I don’t understand how HM kids place so well with such underwhelming scores (for T20 colleges).
This is consistent with what various studies have found, that certain types of high schools place better than their test scores and unadjusted GPAs alone would explain. At a high level, one or more of three things could be happening:
(1) How grades/rigor/rank is treated in an internal academic score could be relatively favorable;
(2) Other factors in holistic review (activities, recommendations, essays and so on) could skew relatively favorable even controlling for the numbers; and/or
(3) Applicants from these high schools could basically be getting special admissions consideration equivalent to common hooks, like legacy status.
From the outside, and without access to complete applications, it is really hard to control for (1) and (2) such that we could know if (3) was playing any meaningful role.
In any event, all this would support one emerging thought about test optional, which is it might be safer going test optional as an unhooked applicant at the most selective colleges if you are sufficiently confident in the perception of your high school’s rigor that having near-perfect grades in the most advanced courses offered will very likely be sufficient to get a very high internal score for academic qualifications.
But perhaps if you do not go to such a high school, a high test score might be more necessary to serve as a sort of independent, standardized marker, on the theory colleges will give even near perfect grades at your high school not as high of an internal academic score.
Other factors at elite high schools that can also affect their placement success at highly selective colleges:
- Dedicated college counseling staff have the knowledge to direct students to applying to highly selective colleges that are more likely to see each particular student as a fit from the college’s point of view, meaning that admission chances are better than for generic reach colleges that students would otherwise all apply to.
- There may be fewer issues with recommendation rationing or low quality recommendation writing.
2 Likes
I should have been more specific.
If conventional wisdom is to be believed – namely that SATs should only be submitted if at or above the median score for T20 schools – the relatively mediocre SAT profile of Horace Mann suggests that at least half the class is going test optional.
This is itself a surprising and indeed shocking conclusion. But if it’s not correct, then the inverse must be true: that large numbers of Horace Mann kids are submitting 1450-ish SAT scores and still getting in, which is also a counterintuitive conclusion.
Basically, I was surprised at the middling SAT profile of the school. I wonder if Brearley, which has even stronger placement than HM, has a similar SAT profile.
Missing information that is required in order to make this conclusion:
A. The proportion of students of the classes or 2025 and 2025 from these students on probation.
Unless the numbers are much higher for these students, the conclusion that this is due to TO is unsupported and unwarranted
B. The proportion of students who were required to withdraw or went on academic probation whose test scores were below the mid 50% of the previous years.
Again, this is not known, so no conclusion can be drawn.
This is a classic case of people pointing to a correlations and screaming “CAUSATION!!!”.
Finally - has this been seen in any other college?
Since when is Harvard the stand-in for the millions of kids at 4 year schools? It’s bad enough that harvard is worshipped as the Epitome Of Academia. Now people are extrapolation from Harvard to the rest of academia.
I would like to see their studies.
Claiming “Our research” this, or “Our research” that is not convincing, even when it is coming from MIT, when they are not actually publishing their research, or demonstrating how they came to their conclusions. Actual publicly published research from other colleges that went TO did not demonstrate any change in success rate after going TO.
Then there is the question of what “reliably predict students” even means. Their retention rates of students who were applied TO in 2021, and were accepted, was 99%. That is as high as it ever has been, and higher than during some years when admissions was not TO.
Finally, and likely the most damning, is the fact that test scores are strongly correlated to family income, and the lower the family income, the more likely a student is to need to drop out, and the mor elikely they are to have issues in their life that will negatively impact their academic success.
Since MIT is need blind, they have absolutely no way to incorporate this into their admissions decisions. The best quantitative estimate of a student’s family income is their test score. So including a student’s test score can indeed help predict student success, but not because its necessarily tells them anything at all about the student’s academic potential. They will not detect the effects of income on success of students, since they have already culled a large number of the most vulnerable students by using tests scores in admissions.
I mean, it’s indeed a way to increase the predictive abilities of the admissions. However, they should admit that they aren’t actually “need-blind” in their admissions.
That, in itself, doesn’t mean that all students did well.
In fact, in this companion piece Stuart Schmill writes (highlighting is mine):
To be clear, performance on standardized tests is not the central focus of our holistic admissions process. We do not prefer people with perfect scores; indeed, despite what some people infer from our statistics, we do not consider an applicant’s scores at all beyond the point where preparedness has been established as part of a multifactor analysis. Nor are strong scores themselves sufficient: our research shows students also need to do well in high school and have a strong match for MIT, including the resilience to rebound from its challenges, and the initiative to make use of its resources. That’s why we don’t select students solely on how well they score on the tests, but only consider scores to the extent they help us feel more confident about an applicant’s preparedness to not just to survive, but thrive, at MIT.
In the same piece linked above, Stuart Schmill writes:
we have found is that the way we use the SAT/ACT increases access to MIT for students from these groups relative to other things we can consider. The reason for this is that educational inequality impacts all aspects of a prospective student’s preparation and application, not just test-taking.
…
This may seem like a counterintuitive claim to some, given the widespread understanding that performance on the SAT/ACT is correlated with socioeconomic status. Research indeed shows some correlation, but unfortunately, research also shows correlations hold for just about every other factor admissions officers can consider, including essays, grades, access to advanced coursework (as well as opportunities to actually take notionally available coursework), and letters of recommendations, among others. Meanwhile, research has shown widespread testing can identify subaltern students who would be missed by these other measures. Of course, this area of research is complex and contested, but the main point is that for every aspect of every application, we always have to adjust for context: as one of the papers I linked above notes, “college admission protocols should attend to how social class is…encoded in non-numerical components of applications.” Meanwhile, the predictive validity of these tests for MIT, coupled with their ability to identify (some) students who would not otherwise be ‘picked up’ by other indicators, means that we are able to use them to help diversify our class more than if we did not consider them.
And according to the new Chetty study, MIT is as need-blind as they come:
The same is often true of this:
3 Likes
I don’t know the answer to who goes test optional at Horace Mann, but I do get the sense that “conventional wisdom” is rapidly breaking down in the fact of emerging data, not least as people process what we are seeing in the latest round of CDS.
I think the new consensus is heading toward something like that in the test optional world, your baseline assumption should be that if you are 25+ in terms of test scores (meaning your test scores place you at the 25th percentile or higher in the latest CDS report for enrolled students) for a highly selective college, you should probably submit those. That can be a narrow window, the difference between 50+ and 25+, but I think at least some people are beginning to believe scores in that range are helping unhooked kids get the high internal academic scores they need to be maximally competitive for these colleges.
Of course some people are getting admitted test optional too. It varies a lot by college, and you have to be careful with the CDS because it reports both SATs and ACTs and very likely some enrolled students submitted both. So you have to deduct something from the SAT+ACT percentages submitted to get to the proper percentage of those who submitted one or both, and by implication the percentage of those who submitted neither.
OK, so who is getting accepted test optional? Great question, not so great data to answer it. But at least at highly selective colleges, it could mostly be people who have some other strong characteristic.
It could, for example, mostly be hooked people–but that is complicated because hooked people can also potentially get in with lower-than-normal test scores. So, say, what is happening these days with recruited athletes getting pre-reads? Are they submitting lower than normal but still pretty high test scores? Or going test optional? Again, good question, not so good data to answer it.
But certainly another (overlapping) possibility is it is also kids from high schools like Horace Mann. Like, I am not sure someone from Horace Mann would actually submit a sub-25th score, but maybe they can get admitted based on grades where that wouldn’t necessarily work for someone who had the same grades but at a normal public HS.
Again, lots of good questions, not lots of good data. But what you are seeing is consistent with some successful Horace Mann applicants doing one or the other–either submitting low and getting admitted anyway, or going test optional and getting admitted anyway.
And so if you are not applying from Horace Mann, and if you are not hooked in some other way, how do you compete? Well, maybe you submit a higher test score. Either way, you have done something that (hypothetical but plausible) Horace Mann applicant has not done. And you hope that helps make you competitive.
I poked around a bit more to see what top boarding schools looked like. Exeter was 720/730 (2019 data); Philips Academy was 720/740 (2018-19 data but reporting “indefinitely suspended” in the current profile); Groton mean 1490 composite (2023).
My guess is that Horace Mann’s SAT profile is really pretty typical for the top “feeder” schools. After all, you’re talking about a testing pool of 100-200 kids where the average composite score is dialing into the 97th percentile or thereabouts.
The puzzle is that scores of around 1450-1490 would definitely put you in the bottom 50% of the most selective schools, and maybe at some places, closer to the bottom 25%. Yet these feeder schools continue to report great placement.
It is very counterintuitive to me that kids going to rigorous schools can go test optional without penalty. All of the ostensible reasons for TO (equity, access, lack of access to prep) go out the window with the cohort of kids who attend Groton or Horace Mann.
1 Like
Here’s another counterintuitive statement: 50% of the graduating class at these schools are in the bottom half; they’re not attending HYPSM, but they are still attending fine colleges.
I’m unsure why it matters what the median is at a certain HS. What’s evaluated is what the applicant presents. Colleges are admitting students, not a group from Horace Mann or Groton.
5 Likes
I think this is understandable if you add a very important consideration not on your list, which is that these standardized tests are just not very strong predictors of college success, particular not at the top of their ranges.
As always, it is useful to keep in mind the SAT/ACT are testing for knowledge and skills that are basically a couple years or more behind the normal expected level for the kids who will be attending these types of colleges. And they can’t test at all for things like advanced scientific knowledge, the ability to research and write a term paper, the ability to participate meaningfully in discussing a dense historical text, or all sorts of other things these colleges will be asking these kids to do.
OK, so colleges rationally want to know more about the applicants’ relative ability and preparation for all that, and in an ideal world they would get that information out of transcripts and teacher recommendations and such.
But in our imperfect world, they do not get transcripts and recommendations in an easily comparable form, because there is no standardization in US secondary school student bodies, curriculums, evaluations, and so on.
OK, so when you think about that from a highly selective college’s perspective for a while, some things start making sense, even if they are not the greatest things from the perspective of social fairness issues.
Like, some kids go to high schools that in the upper years basically function just like their colleges. The kids are doing challenging advanced science classes, researching and writing term papers to high standards, discussing dense historic texts, and so on. And the kids who can do that reasonably well in a high school with a selective student body are relatively good bets to be able to keep doing it reasonably well in a college with a selective student body.
So for kids like that, SAT/ACT scores are adding little or no important information.
OK, but other kids do not go to high schools like that. They may have few advanced classes at all. The advanced classes they do have may be mostly APs, which are OK in a few fields (like Calc, or modern languages), but really not very good when it comes to imitating college-level science or humanities classes.
And in fact, very likely your internal tracking data says something like that kids from the former sorts of high schools with high enough grades in the more advanced classes almost always at least do well in your college, but kids from the latter sort of high schools can have perfect grades in the most advanced classes in their high school and yet sometimes (not always, but sometimes) they will really struggle at your college.
All right, so you have one set of students with unweighted 3.9+ in these classes that really resemble your college classes, and another set of students with unweighted 3.9+ in these classes that don’t really resemble your college classes. In that context, the second group is plausibly at a big disadvantage.
But these colleges don’t really want to completely fill up with the former. To be clear, they take a ton of the former. In some ways, such kids really serve as the annual core of their enrolled classes, the students they can more or less count on being the sorts of classroom students AND the sorts of well-rounded activities participants they need to keep their college rolling year after year.
But they want to also take some of the second sort of student. But they get so many applicants like that. Which do they pick?
And one plausible answer to that question is that they pick the ones who also have high test scores. Not what they would prefer in an ideal world, but because they can’t get the information they really need from their grades in an imperfect world, the test scores help at least cut down on those cases where someone with perfect HS grades is admitted and then struggles.
OK, but then they notice this means they are skewing their admit classes to just a certain profile of people who attend such high schools. And they don’t like that either.
So now they are in a very delicate situation. On the one hand, test scores help them avoid cases of enrolled students who go on to really struggle. On the other, they tend to select for enrolled students from certain concentrated demographics. And somehow they have to feel their way through all that.
But in the meantime, all this is happening outside the part of their class that went to the kids from the college-like high schools. That pool is demographically skewed too, of course, but they think they need them to be sure of maintaining their college’s rolling success. And plausibly, they don’t need test scores as much with those people. Because they have been taking and are being evaluated in college-like classes.
1 Like
We were also just discussing all this in another form in another thread. These colleges have a problem in that for large groups of applicants, sufficiently high quality and unbiased information about their preparation for college classes is severely lacking. In that context, using all the data available to build multi-factor models makes sense, but it doesn’t really solve the bias issue. And then you can chip away at the bias issue in other ways–to the extent the Supreme Court will let you. But you can only do so much with the information available.
By the way, it is interesting to take a very close look at those Chetty charts (which, to be fair, are a bit old at this point). Controlling for test scores, MIT starts tracking on the low side in the top 3% or so. But if you look back, it is tracking high from like 70-97 or so.
In venues where a lot of upper-middle-class families are complaining about the overrepresentation of top 1% families once you control for numbers, this may seem perfect.
But the broader social issue include that the upper middle class is overrepresented when you don’t control for numbers. Which, among other things, is because upper middle class kids are overrepresented when it comes to very high numbers.
So MIT putting relatively high weight on high numbers, particularly Math, is plausibly pushing down some admits from the top 3% to up to the next 30% or so.
But not farther than that, because after that the general skew of the numbers is making such test scores extremely scarce.
Incidentally, Caltech is noisier, but it has a similarish pattern–it is that high outlier at 70-95 on the second chart, then the low outlier 98-100. It is hard to see the way these charts are formatted, but CMU, Hopkins, and Chicago are sort of like that too, albeit less extremely.
1 Like
But if they are correct in that lower scores hamper students’ success in their rigorous course of study (and we have no reason to suspect they are not - in fact, it makes perfect sense), then they are doing what they can - but not more.
Because the alternative of lowering their academic standards would strike at the core of their mission.
3 Likes
To oversimplify - if a student attends a rigorous known high school and has excellent grades and advanced coursework, a college will feel comfortable admitting said student without test scores.
3 Likes
Half of what you say makes sense, namely that SATs can validate GPA from high schools of unknown quality. That’s MIT’s rationale for requiring tests.
But we’re talking about test optional. Tests may not tell you much about the valedictorian from Exeter. That person has already demonstrated the ability to succeed at HYPMS.
But let’s be honest, that person isn’t going test optional, either. I have a suspicion that you’ve got lots of private school kids at Horace Mann who aren’t at the top of their class, maybe As and Bs, some hard classes, some fluff, who are going to choose test optional. They’re not going to go to HYPMS, but they are absolutely going to semiselective schools like Emory, WUSTL, Amherst etc
TO just doesn’t make sense to me for those types of kids. It seems to reverse the entire rationale for TO.
I think the TO strategy works best for private colleges with an acceptance rate between 15-30%. If the acceptance rate is lower, than you really need every hook, line, and sinker to get accepted. If the acceptance rate is mid 30’s and up, then the average ACT may be 31 so if the student goes TO clearly the student didn’t test well at all, which may be a red flag.
So to me best case, for TO kids with excellent rigor/grades are privates where the average ACT has continued to creep up to 33/34 for bottom 25%. Not tippy top selective, but not mid tier selective. There is definitely a small swath where TO can work well.
I go back to Boston College as an example. For current applicants, it is recommended you do not submit if your score is under a 34. Well, my guess is that is a huge number of kids, a lot of them very bright, who tested 31-33 but now are told not to submit are applying to BC.
1 Like
semi-selective?
4 Likes