SAT tests what a student has learned. It is not an aptitude test. CollegeBoard doesn’t even believe it’s an aptitude test.
You must have not seen a lot of student transcripts or college apps. Most high school students have more As than Bs. The average ACT for Class of 2023 was 19.5. Average SAT was 1028. Plenty of students have high grades and relatively low test scores. Clearly some haven’t learned the material they have been taught. Some aren’t good test takers under time. Some have undiagnosed learning differences.
Lot of issues. Lots of blame to go all around, especially for the many who are getting As in AP courses and 1 or 2 on the AP tests.
In fact once you think about it, the absolute hardest cognitive tasks (in the sense few people can do them even with all the time in the world) are naturally least amenable to testing this way.
Like, math problems that are so hard many people can’t solve them–there is little point in testing that sort of thing under time pressure.
Exactly. That is why they have to give you so many problems to solve in so little time. Too many people would get too many answers right if you let them take the time they needed.
So managers and educators who understand these issues then understand if they want the best quality of work, they should set deadlines to give people enough time to do their best work.
But we are digressing. So I will just link a couple scientific papers on this subject and leave it at that:
The recent Field’s medalist, June Huh, is noted for being extremely “slow”, but I’ve no doubt he would still ace the SAT math section with little trouble since he got into Seoul National University.
There’s actually a huge difference in academic output even between 99th and 99.9 percentile SAT Math scores. As a whole, people who are “talented” do well on standardized tests.
Do you think it’s okay to destroy a working system just because 1 out of a hundred talented people is slow and does poorly on the SAT? What about the other 99% of talented people who are accurately measured by the SAT?
Do you have a source for this statement? You have made it several times. What does unqualified mean? Dartmouth said nothing about having a higher than expected drop out rate for the recent TO classes, and D adopting a test required policy does not infer that they have enrolled ‘unqualified’ students.
The 2023-24, 2022-23 and 2021-22 CDSs show 98% retention rates, which is actually higher than the previous five years (the farthest I checked). Are these ‘unqualified’ people still there?
The paper shows that the TO cohort has an average SAT score of 1400 and they’re performing right where the regression line predicts that someone of that score would perform: In the 33rd percentile.
I am not sure what system you think I want to destroy.
I have no problem with Dartmouth wanting to require standardized tests again so it can use them in its initial academic screen on a contextual basis to help identify some academically qualified applicants it would not otherwise identify. In fact, apparently Dartmouth has been developing an AI/Big Data modeling approach to initial screening which currently uses something like 64 different types of data available to it. If it wants test scores from everyone to include as one of those 64 because they are finding that is improving that model’s accuracy in some cases, great, go for it Dartmouth.
But I am also explaining why Dartmouth reportedly does not do much with test scores after that initial screen. Which makes sense including because of all these issues surrounding what they are really measuring.
IMO, Dartmouth is wasting their time and money with AI/Big Data. That type of sophisticated analysis would only work if the explanatory variables were hidden and multi-faceted, but their study pretty conclusively proves that nothing except for the SAT predicts student performance.
I found it rather funny that they saw that a 4.0 HS GPA student had only a .1 higher GPA than a 3.2 HS GPA student. If their AI/Big Data solution works as intended, it’ll come to the same conclusion as this study and they’ll probably end up scraping the project and never talking about it again.
In fact Dartmouth is making the exact opposite sort of argument. It isn’t arguing it needs tests to eliminate unqualified applicants slipping through the cracks. It is arguing it needs tests to allow in more qualified-in-fact applicants of certain disadvantaged sorts it believes it is currently unable to identify without tests.
Of course necessarily if more qualified-in-fact disadvantaged applicants are identified and admitted, some of the advantaged applicants Dartmouth was previously able to identify as qualified are not going to be admitted. But Dartmouth isn’t stating that as a goal, the goal is just to admit more disadvantaged applicants.
Of course we know most low income students don’t take a test, and if they do have a 1300 or 1400 we know data show they (generally) won’t apply to a school with high test score ranges like Dartmouth’s. So not sure how D’s plan is going to work out.
The counselor pages are spicy indeed with this news.
ETA: My biggest question for D is why weren’t they admitting these types of students for all those years through 2020 when they required tests?
I apologize for jumping to conclusions. It’s just that whenever people talk about the validity of SAT scores, it’s like people saying BMI doesn’t matter overly muscular guys aren’t accurately measured by BMI.
The SAT and other standardized tests accurately measure math and reading ability for 99.9% of students as evidenced by the following studies:
You should reflect about whether a sophisticated entity like Dartmouth, with many millions to spend on studying these things, would really be likely to make such an obvious mistake.
And in fact, your premise is wrong. Dartmouth did not claim “nothing except for the SAT predicts student performance.” Instead, they claimed, “SAT scores have significant predictive value for academic achievement over and above other measures such as high school GPA.”
That is a carefully constructed sentence. It implies SAT scores alone are more predictive than high school GPA, or any measures “such as” high school GPA–taken alone.
It does not implies SAT scores alone are more predictive than a model made up of 64 different types of data being used intelligently together. For that matter, it did not imply SAT scores alone are more predictive than other factors that are not “such as” high school GPA.
But again I digress. You may believe Dartmouth is getting this wrong, but I am just trying to help explain what Dartmouth is actually doing and why.
[quote=“NiceUnparticularMan, post:198, topic:3657570”]
You should reflect about whether a sophisticated entity like Dartmouth, with many millions to spend on studying these things, would really be likely to make such an obvious mistake.
And in fact, your premise is wrong. Dartmouth did not claim “nothing except for the SAT predicts student performance.” Instead, they claimed, “SAT scores have significant predictive value for academic achievement over and above other measures such as high school GPA.”
That is a carefully constructed sentence. It implies SAT scores alone are more predictive than high school GPA, or any measures “such as” high school GPA–taken alone.[/quote]
There’s Dartmouth’s PR and then there’s actual data. Their data is telling us that only the SAT predicts performance. Even HS GPA was effectively worthless (4.0 HS GPA students have same college freshmen GPA as 3.2 HS GPA students). I suspect that as a public-facing institution, Dartmouth is already in hot water for bringing back tests, so they’d want to avoid anything controversial if they can avoid it. That includes saying the obvious and politically incorrect thing out loud.
We’ll have to wait and see, but I can’t imagine any other variable or combination of variables that will be more predictive than the SAT. Dartmouth could just be doing this as a “just in case” experiment, but usually when the initial rigorous analysis comes out with a result like Dartmouth’s study, it’s pretty conclusive and no kitchen-sink approach will improve their predictive power.
Dartmouth, for many years was a true leader in CS. The term “AI” was coined at a Dartmouth convening of experts. They invented BASIC there, students there all had computers (and used them for everything) before most schools.
I think it’s likely both, it’s just more palatable to say “we are missing qualified disadvantaged students,” than to say, “we are disappointed with the performance of some test optional admits.”
You aren’t making the point you think you are making.
BMI measures one thing and one thing only. It does NOT measure “how healthy is this person”- someone with end stage liver cancer can have a perfectly “healthy” BMI. It does not measure “What is this person’s expected longevity”- many people with severe and acute illnesses have “healthy” BMI’s up to the day they die.
BMI doesn’t matter in the way that people think it matters- as a snapshot for “is this person healthy”. It doesn’t do that. It doesn’t measure “health”, longevity, life expectancy. It measures one thing-- which in context with about 29 other indicators, can be used to determine “is this person healthy”. But without at least ten of those 29 indicators, it is a pretty useless number.