"I also disagree on several fronts that SAT/ACT is a direct measure of intelligence or a predictor of future performance. "
I agree with you that the tests don’t predict future performance.
“We seem to think there’s some major qualitative difference between a 1460 and a 1600”
You are correct. On the old sat a first time 2250 was considered as good as a 2400 by admissions officers because most students improve on taking the test a second time around 100+pts.
Standardized test though are IQ tests. Classes with higher IQs would most likely move faster, cover more material.
Also, high stem schools are known for covering lots of material quickly regardless of sat scores.
Does this mean that a student from an elite university would fair well at a high stem school? I would believe that a student transferring from an elite LAC to a high stem public school like NC State would see their GPA drop by a point and take a couple of semesters to adjust. My own experience hiring engineers, I’ve seen elite undergrad students with 4.0s who then obtained MS at public universities with 3.2s.
Transitioning LAC to STEM is hard.
Transitioning STEM to LAC is easy.
re #19: “This is in large part why state flagship engineering programs have such large attrition.”
The theory goes, these are some of the exceptions that are oh so necessary to prove the rule.
That rule being one of significant but imperfect correlation, not absolute relationship in every single case. Nobody whatsoever claims that it holds in every single case, or that there are no exceptions.
@Mastadon, there are three things in my run on sentence you referenced, all tied back to the original post: 1) the challenge (or lack there of) of getting good grades at an elite school 2) the academic rigor at an elite school 3) whether or not an elite school prepares their graduates to be effective at the jobs they’re training for. That’s 3.
@Greymeer, statistically, mid 1400s to 1600 are the same, regardless of first or fifth attempt, 99th percentile. As for standardized test (ACT/SAT) being IQ tests, they most certainly have been shown not to be. They neither measure intelligence nor subject aptitude, but rather but rather measure only how well people take that particular test.
You find the same ability when you look at the difference of mid 1400s to 1600 in the range of 400 - 1600. You find the difference when you limit the range in mid1400s to 1600 students only.
No you don’t. In fact, from sitting to sitting, the same score doesn’t even mean the same thing. On the math section missing one question can result in scores anywhere between 760 and 800. It’s relative to all the other test takers only on that day taking that specific test. So, if you happen to be lucky enough to sit on a day where there are more low scores, it’s actually easier to score higher. Take the test on a day where all the foreign students who were given the answers in cram school (look it up, it has, and does happen) also took the test, and you’ll likely score lower. Any suggestion otherwise, and unfortunately, that is certainly what the majority of people believe, is simply ignorance of the test’s methodology. You cannot further parse the 99th percentile.
@Greymeer, we can’t just say that a STEM-to-LAC/Humanities, or vice-versa, transfer is easy or hard.
Some kids are just better with numbers and logic than they are with qualitative concepts, and some top numbers kids simply can’t write. That kid will not find it so easy transitioning to a LAC/soft science/Hum major, class or environment with a lot of classroom discussion and writing.
Conversely, there might be a math whiz who is simply interested in the soft sciences or humanities instead. So nitvevery kid going from a qualitative major is going to find STEM hellish.
To say that hum/SS kids are weaker than STEM kids is wrong.
Re #18, don’t know about the referenced study, as I don’t care enough to follow this stuff, but IIRC it used to be the case that HS grades had some predictive power, but the best prediction of first year college grades was a combination of HS grades and SAT scores together. Which makes a lot of sense to me, since HS student populations can vary quite a bit.
I think you will find that colleges flashing the higher SAT brackets tend also have higher HS grade brackets, on average. [As a correlative matter, NOT a one-to-one relation, NOT a relation that has no exceptions]. If that’s the case, then referencing by SAT scores can be used as a rough proxy, for some purposes, for referencing by a combination of SATs + HS GPA.
After first year predictive power apparently wanes, I guess. But that makes sense, because after that, students have completed many “gen ed” requirements that don’t cater to their strengths, and can focus more on the particular areas that they are best at. Also in many cases study habits may have adjusted more to demands of college. If they’d had those study habits in HS their HS grades would have been better.
SATs are not IQ tests per se, but they were derived from army IQ tests. They Can be studied for. But not to an infinite extent. A 1400 scorer can possibly get to a 1500, but a 900 scorer is unlikely get there. IMO.
@monydad, I know we are diverging from the original topic a bit, but the interesting follow up is why did the student score 900 in the first place. It’s easy to assume it is a direct measure of their intelligence and potential, but the language and style of the SAT is known to disadvantage students whose first language wasn’t English and students of lower socioeconomic status.
The study referenced above compared students at test optional schools like Wake Forest, Pitzer and Smith, and found that when matched for other variables like GPA, AP scores, etc. that students do the same in college whether they took the SAT or not. This indicates the SAT has zero added predictive power, and due to the factors referenced above, may be counterproductive.