More Colleges Backing off SAT and ACT Admissions Rule

Applying to College
281

@SatchelSF

Interestingly this Ohio U report I stumbled on just now has UGGPA for admitted MS and PHD engineering students.

Higher GPA for women at both levels. So I’d guess you could then look at graduating MS students GPA (they have GRE breakouts too.) by gender 2-4 years after the study and see which what was a better predictor. Oddly, the female and male students admitted to MS and Phd programs flip the SAT script, with Men having a few point better Writing GRE and a few point lower MATH. Interesting that OU seemed to prefer higher writing score men and slightly higher GPA and math score women… Dunno what to make of it, really, but my interest in this has been revived a bit.

the have graduation GPA by gender over a couple of years too, like UCSD:

Women score about .2 to .3 higher in GPA graduating from OU engineering (and architecture too) That’s really interesting.

And women earn a higher % of honors in both schools compared to their representation… Interesting.

https://engineering.osu.edu/sites/eng.web.engadmin.ohio-state.edu/files/uploads/uess2013.pdf

So between UCSD, MIT and OU we have some evidence that:

When women are admitted with the same, or slightly less, SAT scores as men, (in UCSD admissions, since gender can’t be taken into account) over time their engineering GPA will consistently be a few 10th of a point higher than men and more women will have a 3.5+ GPA than men.

At MIT, they will also consistently earn a higher GPA and more will graduate in <6 years.

And at Ohio U the women admitted to the Engineering MS and PhD program have higher UGGPA than the men admitted (this could be due to any number of factors, but suggests, since in Engineering there are usually fewer female candidates than male, that the female candidates in general have better GPAs, since, OU is “lowering” it’s GPA standard for men, not women…

Also, at OU, the women UG engineering and architecture graduate with higher GPA and more honors (a 4 year sample.)

All pretty interesting. I’m coming more and more to the conclusion that maybe SAT Math is not a particularly good predictor of engineering educational success after all…

282

"Why would male students consistently record 30 point higher SAT math scores on average, yet not perform better than women earning equal or lower SAT scores in math-intensive programs or Colleges for GPA or graduation rates?

Is it possible at the top end, SAT Math is not a great predictor of collegate success?’

A 30 point differential is not considered significant by the college board and by most colleges. Also girls tend to do higher in the verbal and writing sections so the SATs are about the same to MIT. Again a 4 percent difference in grad rates would not be significant. If you were talking 100 pt and 10% respectively, that would be notable. Meaning girls with a 650 math sat graduated at 94% while boys with 750 graduated at 84%. And you need to find out more why the boys dropped out, if they dropped out to start Facebook, Microsoft, Apple, that’s different. I certainly don’t hold Harvard or Gates, Jobs, Zuckerberg in a worse light because they dropped out.

BTW that article was written to show that girls don’t get into MIT just because they’re girls.

283

“Your individual curiosity and GRIT is what will carry you through life. Increasingly colleges are trying to do a better job of selecting students with these qualities.”

Those are great qualities, nobody would disagree with that. However high test scores are necessary (but not sufficient) for getting in to the highly selective schools, unless you have a hook. And grit is not really one of them.

And do you know how many kids with grit and curiosity got rejected by the top schools - a lot.

284

“All pretty interesting. I’m coming more and more to the conclusion that maybe SAT Math is not a particularly good predictor of engineering educational success after all…”

I don’t think anyone is saying that one criteria will predict, well I guess you are with GPA. The studies you cite have students scoring in the 700s, you have to show that students with a 550 average in math do better than those with 750 in math. That would be like taking an average CSU kid (not the best) putting them into Cal Tech and getting a better GPA than his or her counterpart. Do you really believe that?

285

Just to correct the record: Jared is not competing against Dorothy. He is competing against his peers at Snoot School who have As.

Carry on.

286

@theloniusmonk

I don’t understand your reasoning. One argument for a higher acceptance of men than women in, for instance, engineering programs (like the UCs) is the persistent difference in SAT Math scores. The justification - and it’s all over the place out there - is that men are just “better” at math and so, therefore, should get into engineering and other math-intensive programs at a higher rate ( I’m pretty sure @SatchelSF made a similar argument in this thread. And you can easily find it, even among “mainstream” thinkers.)

The problem, of course, is college (and life) are not simply about “intelligence” and the argument goes off the rails when those who supposedly exhibit highest “intelligence” don’t exhibit highest “success.”

So, if the overall graduation GPA and graduation rate for Women in engineering is PERSISTENTLY higher than men at certain programs the obvious conclusion is that the schools are picking too many Male applicants who are not as successful as the female students they select (and my guess is that is because men’s slightly SAT scores “compensate” for lower GPA in the overall ap, but I don’t have the data for that). Then the question is: “Are they picking the wrong Males, or are Females just naturally more successful in these Stem/engineering progams” (I’m being “ironic” there as that is the argument many put out in regard the Math SAT score imblalance…")

But the reality is (for UCSD, MIT and OU at least) these schools are actually picking a less “successful” pool of Male applicants who then get lower graduating GPA and have a lower graduation rate (assuming you go with me that those are good indicators of success.)

Now the question is why is their selection of Males less successful than their selection of Females?

287

Grit is embedded in other admission factors such as higher academic and extracurricular achievements.

The same goes for applicants with high test scores.

288

@theloniusmonk @ucbalumnus

And the question again is: Does GPA predict “success (including grit)” well enough to forgo the $$ and time taken by testing.

My instinct is that standardized testing weights the pool against “grit” as GPA is a “long game” that requires persistence and consistent effort that SATs don’t (and this is from a “short game” guy who only got into a decent college because of his PSAT score and spent much of his HS doing, well… things that were not studying…)

But I think that jury is still out. I am encouraged, however, that more and more schools are suggesting maybe they can select a successful applicant pool based only on actual HS work/ECs/LORs and don’t need the artifice of standardized testing.

289

“Now the question is why is their selection of Males less successful than their selection of Females?”

So for MIT, they don’t use math scores to pick a class only to screen anyone less than a 700, as 97% of the admitted class had over 700. So if you assume the girls are at 700 and the boys at 800, that’s not a big difference to show up in a classroom. Basically the enrolled students, boys and girls, at MIT are good at math (to use your words). MIT accepted pretty much the same number of boys as girls according to their latest CDS:

girls 754 accepts, 5889 applicants - 13%, of those 511 enrolled, 68% yield
boys 757 accepts, 13131 applicants - 6%, of those 599 enrolled, 79% yield (impressive yield with just EA)

Of the 511 girls, 6% don’t graduate in 6 years, or 30
Of the 599 boys, 10% don’t graduate in 6 years, or 60

That means according to you without looking into the reasons why the 90 out of 1100 kids didn’t graduate in six years (started a company, had to support family), MIT is doing a poor job of admissions because they rely too much on the Math SAT?

290

“My instinct is that standardized testing weights the pool against “grit” as GPA is a “long game” that requires persistence and consistent effort that SATs don’t”

Well, people spend a lot of time studying for standardized tests, it’s not a short game at all. And the UCs and Stanford, maybe more colleges only use 10th and 11th to calculate GPA. It’s not a four year thing at all for admissions, it’s three max, maybe 3.5 if you include midyear grades (and not used if you apply early). So if you include sat 2s, psat, APs, the time spent studying for them is not that much lower than the 18 months that comprise 10th and 11th grade.

291

@theloniusmonk

Huh? Your MIT stats also mean 2x as many boys don’t graduate as girls in 6 years - 100%!!

You are focusing on minutia to ignore the bigger picture. So, if MIT does not think after 700 on math SAT the SATs are useful for admissions then how useful are they? Do you think a decent adcom can’t tell you if a given student is a 700 math student? And is a kind who tops out at 690 really less able to succeed at MIT than a kid who get 710 on their 4th try (a true story for one of my kids - got 690 twice. Knew they were not getting the score they could. Spent 1 month in very specific study with a tutor - hit a 750 next time out. But we were able to pay for tutor, rush scores etc.)

Again, the GPA difference is real and replicated across 3 schools - one STEM/tech focused and in the two others, in engineering (and architecture.) So, if SAT is supposed to be helping these schools “better select” a class, you have to ask why it is not helping them here? The amount that SAT itself claims their studies improve ability to predict FYGPA are tiny as well, but they (and others) happily use them to argue that SAT is critical to the selection, but you’re saying anyone who gets over a 700 math can handle MIT if GPA and other factors suggest it? Ok. That supports my position, really.

And the “GPA is only 2 years…” For UCs, yep. For the rest of the real world: 3.5 AND you can’t blow your senior year. Every school has a few kids get warning letters based on midterm senioritis grades.

Anyway, I find every argument I get from the “lottery ticket” argument for Podunk Pete to the "anyone over 700 Math can handle the top STEM school in the country continues to lead me to ask - so, what is the value of SAT/ACT again?

Obviously, some schools will move to the “test optional” basket, but the financial grip of the standardized-test industrial-complex won’t end soon (if ever.) So our Princeton friends can keep shopping for their houses in Short Hills. Good for NJ income taxes!

292

“And is a kind who tops out at 690 really less able to succeed at MIT than a kid who get 710 on their 4th try”

Statistically, there is zero difference: both are woefully unprepared to deal with MIT. Median SATM is going to be somewhere north of 790. Only 0.7% of admits presents an SATM < 700, and only 8.0% < 750. These students will be almost exclusively URM. Someone who can only score a 710 on the 4th try (or a 690 on the 1st, unless we are talking 7th grade talent search and even then it’s questionable) is going to be hopelessly outclassed at MIT, CalTech, the math intensive majors at HYPS+, etc.

Because the SATM is so easy, even a single question answered incorrectly - typically a careless misreading of a question in my experience - will knock SATM to well below the MIT median. The SAT Math Level 2 is really no better, although less prone to “sillies” as usually you can get 4 to 5 wrong and still get an 800, a score which 20% of the test takers generally achieve. That’s more than 30,000 students each year. Many of the math kids I’ve worked with have taken to knocking out the Math Level 2 achievement test in 8th or 9th grade.

One of the problems is that the math testing is way too easy for the talent that is out there. GPA is really no better - everyone gets an A anyway these days, and increasingly kids are taking AP Calculus in 7th and 8th grades, at least on the West Coast. I’d think that the vast majority of successful applicants to MIT or Caltech will need to present some real evidence of mathematics or science talent like AIME scores, or showings in science fairs and olympiads, etc. I’m with @CaliDad2020 on this - for the very elite quantitative schools SATs are meaningless.

293

you boys needa chill. this guy up here talking about how kids are taking calculus in 7th grade, chill out dude.

294

@SatchelSF @theloniusmonk

So do you think that MIT therefore simply chooses from a Female applicant pool that somehow transcends the persistent 30 point gender discrepency in Math? And if so, and given that MIT applicant math scores, regardless of gender, are likely to be “effectively” 800, why would women consistently and persistently earn a higher GPA and higher 6 year grad rate if they are “naturally” and “inherently” less capable at math?

Something in that equation don’t compute. If all students enter at the top, and some are toppier than others, even though tests suggest they should less top - well, you’ve got a fairly large pool of outliers, no?

Or maybe your test “does not mean what you think it means.”

295

@CaliDad2020 - I’m not exactly sure what you are asking regarding females at MIT. The report you linked, while helpful, is largely fluff and does not provide enough detail to understand really what is going on.

There is no “30 point gap” at MIT on the SATM between females and males who are actually at MIT. The entire IQR of the SATM distribution at MIT is 30 points (770 to 800), and with some plausible assumptions about the shape of the distributions for the various groups, it’s hard to see how the median for non-minorities could be outside the 790-800 range. So, at most, if there is any gap between females and males there in SATM, it couldn’t be more than a few points at most, and it’s quite possible that there are more 800 SATM females than males there. See the stats here: http://mitadmissions.org/apply/process/stats

Another way to think about this is that the most common score at MIT on SATM is 800. As you yourself noted, one question wrong will result in a 750-790 score (depending on particular test date), and because of random “silly” errors that occur for all test takers, the SATM is really not measuring anything meaningful for the vast majority of MIT students (preference admits aside, because their score range is low enough to get them out of the “single random silly error” issue). The overwhelming majority of students at MIT are literally off the SATM scale. If anything, at these levels the SATM is just a test of conscientiousness, ability to sit still for an hour and focus, not of any inherent math ability.

So, I’m agreeing with you that the SATM measures nothing meaningful for the MIT admission process.

Are you asking whether females at MIT are better at math than the males? That’s a different issue. Or, are you saying that because females score lower on average in the entire college population than males, than the statistics put forth by the MIT diversity report that females have higher GPA and graduation rates at MIT, proves that SATM is not a valuable measure for any school, even one where ability is quite a lot lower than at MIT (as in OU)? Or, are you saying that the higher GPA at MIT proves that there are more “top” math ability students at MIT who are female than are male?

296

@ucbalumnus - “The two cities [San Francisco and San Jose] have similar populations and similar percentage Asian (SF has higher percentage white and lower percentage others than SJ). SES is similar. So what explains the 7 times difference [in NMSF]?”

There are 7 times as many “smart” kids in San Jose as in SF (taking “smart” to mean in this context the NMSF threshold). What’s so hard to understand about that?

Who cares if SES is presumed to be similar in the two cities? (I’m not sure it really is - you do need to compare the school populations and the parents of school age children between the two cities, not simply aggregate population numbers on income, education, race, etc.) Even if SES were similar, doesn’t that just support the idea that SES is not a very big deal?

For any test with a cognitive component, the primary causative factor of results is going to be intelligence. All this back and forth about SES, race, ethnicity, income, parental educational attainment, and even SAT scores and IQ scores for that matter, etc. is just banter. Those variables are just the “shadows on the wall” from which we can only in a very rough way infer the true explanatory variable, intelligence.

I’m reminded of the famous story of Bernoulli’s challenge to the best mathematicians of his time to solve a particularly tough calculus of variations problem. Newton solved it in 12 hours, while Leibniz asked for a year’s extension of the deadline. Newton had a trusted colleague submit it on his behalf anonymously. When Bernoulli examined the solutions a year later, he immediately knew who had submitted the best work, and proclaimed, “We know the lion by his claw.”

It’s just amazing the variation in human capabilities and talents, even at the plus 5 and plus 6 standard deviation levels of a Leibniz and a Newton. I will never understand why people can’t just celebrate our differences, rather than concoct dopey stories about how we are all really the same if only… if only…

297

@SatchelSF

I am asking this:

Folks use the “persistent” 30-ish pt. difference between M and F in SAT to “prove” that women are “not naturally as good at math” and justify much of the inequality in F appearances in some top programs. (CIT for instance, or any UC engineering program, some CMU programs to call out a few.)

MIT has the pick of the litter of Math-capable M and W and likely (as you note) pick all “effectively 800” M cadidates, or at least “mostly 800-capable math candidates” - folks I think we agree are statistically in the same “lane” for math.

This assumes that, at “worst” the M and F candidates admitted to MIT are equal in math capabilities. (It is possible F have slightly lower scores on average as MIT is able, if it wants, to select lower scoring F candidates to keep gender balance. I don’t know if it does.) But at the very “least” F and M candidates are equal in regard SAT Math scores and capabilities (on average.)

So, why then are F persistently more likely to graduate in 6 years and get higher GPA?

And why do we see this at UCSD and OU engineering as well?

If Math SAT strength is supposed to be a “best” indicator of “ability to succeed.” And if F score persistently lower - yet F who score at “best” equal to males in SAT (since we’re saying MIT, by definition, is picking the “best” math students) consistently perform higher than M in college GPA and graduation rates, it seems we have to question, at least in the margins, the logic sentence: M do better than F on Math SAT persistently, SAT predicts success in math-centric College programs, Higher SAT Math scorere (ie. often M) will therefore have more success in Math-centric programs in college.

It seems to me it appears equally high math scores don’t produce equally good GPA or grad rates. Now, it may be as @theloniusmonk argues, that the 4% difference in grad rate and the few tenths difference in GPA are not signficant - or even not a good indicator of college “success” (though it’s hard to argue that, i think.)

But often these Math SAT score debates miss the forest for the trees: GPA is often as good if not better as an indicator, but as we see in this thread, folks are much more comfortable with “precise” numbers, even if those precise numbers are not more precise in predicting outcome.

Which brings this back to the original point of this thread. More schools seem to be agreeing that perhaps SAT/ACT are not critical, or even that helpful to the selection process. I’m suggesting that some things, like College Graduation GPA in subjects where SAT is used to “prove” inherent gender imbalance, supports that trend.

(Along with a whole host of other evidence that suggest SAT/ACT is marginal at best to college admissions, but I’ve done that dance here.)

298

Ability, hard work and interest need to be considered. I know many women that are very strong in math (they have the ability and they work hard) but they hate it. That’s why they aren’t pursuing STEM professions.

299

@CaliDad2020 - Good points, interesting discussion. I don’t really know too much about engineering, so I’ll stay out of that. But I do know something about math.

I’ve been looking at the MIT paper. It’s pretty clear to me that despite the presumed equal 800 mode score on SATM for the two sexes, on math ability the girls at MIT are less capable. Especially at the extreme right tail. I tried to find comparable stats for CalTech but couldn’t. I did discover that the IQR for Caltech admittees for SAT Math Level 2 is 800-800, reinforcing the idea that SATM and SAT2M2 are meaningless at this level, but I guess also effortless for the successful candidate so no harm no foul…

The gold standard for anyone who is involved in college math is the Putnam competition. Look at the MIT Putnam results - MIT gets to pick 3 team members each year and as many kids can enter independently of course. The stakes are high - MIT has been in the top 5 for like 16 out of the last 20 years, and is second only to Harvard over the last 80 years in Putnam Fellows. You can see the team members here: http://math.mit.edu/academics/undergrad/opportunities/putnam.php

You’ll be hard pressed to find any female names on that list. (I had to google many, and some of course are well known to the math competition community.) I can’t find a single one who has either been a team member or Putnam Fellow from MIT (excluding of course the girls’ competition - the Elizabeth Lowell Putnam which has a lower threshold, although takes the same exam). At least since 2003, MIT has had the chance to pick 14*3=42 team members in total. As near as I can tell, males = 42, females = 0. It’s easy to write a pat on the back diversity report, much tougher to put your money where your mouth is when real results are on the line!!

Also, plenty of clues in the survey results (see first set of responses in the Appendix) that females are less prepared than males at MIT in math (9x more males to females in hardest of 3 math sequences, 1.8x more males in intermediate sequence, but 1.3x more females in easiest sequence. Of course, many males come in as IMO gold and place entirely out of the sequence. Not many females do that I can assure you… I strongly suspect that females are getting preference in admissions at MIT based on those math sequence results that slipped out (note that the report in the discussion says that males were “statistically more likely” to take harder math than females - at 9 times the rate I’ll say that’s an understatement!).

I don’t particularly care about endogenously given variables like GPA or graduation rates (easily explainable for the males in graduation rate - more variance in male behavior and intelligence, less maturity when matched for age, and also more likely to be bold and leave school to start a company or jump to a hedge fund). GPA - I suspect choice of major is a big issue at MIT and I simply don’t believe that they have corrected for it (no methodology or data are provided in the report). I prefer exogenously given variables for results. I noticed in the survey that MIT males were 1.5x more likely to win “outside” awards for their work than females, while internal MIT awards are equal male female.

Just to wrap up on math ability being selected at MIT, I would bet the farm that if MIT released the results of real achievement milestones at this rarified level, male scores will be demonstrably higher than female scores in the admittee pool. AMC12 and AIME scores, and USAMO qualifications are the gold standard analogs in the high school world to the Putnam in college. Lore in the competition world is that 80% of USAMO qualifiers are accepted if they apply to MIT. The sex ratios are not disclosed explicitly anymore, but they skew 80%+ male at these levels.

Anyway, fun discussion on the thread, I hope some ideas get traction.

300

“So do you think that MIT therefore simply chooses from a Female applicant pool that somehow transcends the persistent 30 point gender discrepency in Math? And if so, and given that MIT applicant math scores, regardless of gender, are likely to be “effectively” 800, why would women consistently and persistently earn a higher GPA and higher 6 year grad rate if they are “naturally” and “inherently” less capable at math?”

Who is saying that females at MIT are less capable than the males there? You make up these strawmen and are the only one arguing against them. What if the males that didn’t graduate in 6 years had a lower math SAT score than the females.? You need more info about the 90 kids that didn’t finish in 6 years, what if you find that the males that didn’t graduate in 6 years all had a 4.0 gpa? Both of those would hurt your position.

And you’re the only one that is making a big deal about a 30 point discrepancy, if you have data that says that kids with 4.0/1300 do better at MIT than a 3.5/1500, that would be relevant.