Dartmouth reinstates the SAT

[jimbogumbo]

This study is a better description than College Board statements. And before anyone slams me, I do indeed know that the ACT is not the SAT, nor even an ETS product.

https://news.uchicago.edu/story/test-scores-dont-stack-gpas-predicting-college-success

One thing to note:

They examined 55,084 students who graduated from Chicago Public Schools of varying academic profiles between 2006 and 2009, and who then immediately enrolled in a four-year college. At the time of the study, all Illinois students took the ACT in the spring of 11th grade.

So, in some important ways, it's a relatively homogeneous group. The value in using standardized test scores is greatest when there aren't other metrics that are consistent for the entire population.

I'd suggest it is not only relatively homogenous, but comprise for the most part by the groups I cited earlier. What isn't emphasized is that the ACT would under predict their chance of success in a four year regional public or CC.

As someone (Hibush?) pointed out, Dartmouth is trying to distinguish among highly prepared prep school students. The CPS group is disadvantaged if forced to take the SAT or ACT as an equally weighted entrance requirement to attend a public that isn't R1 (say Illinois State), and probably even at a flagship R1 (such as University of Illinois).

[kaysixteen]

Good point about Dartmouth's heavy reliance on uberelite prep schools, but the same poster notes that most of the other students come from elite pub hss, such as the Greenwich, CTs of the country.   People who spend enormous amounts of dough to buy into such communities do so because they well know that they can expect elite prep school-quality pub schools there, and they get 'em.  Meanwhile most American kids go to schools that are very different, some almost parallel universe-level different.

[Stockmann]

There's a lot wrong with SATs, but one big thing going for them is that everyone takes the same exam, allowing for meaningful comparison. In the absence of some standardization, HS grades are basically meaningless, especially given grade inflation (A is Average).

I'm not in the US so my place doesn't do SATs. We do have an entrance exam (so it's standardized in that everyone takes the same test), and we do have hard evidence that performance in the test correlates with performance of admitted students. HS grades, on the other hand, basically don't correlate. My campus' students have basically the lowest socioeconomic profile of any major, accredited HE institution in the country (and this is a developing country) so our test isn't about keeping poor students out.

[fizzycist]

SATs are an example of standardized tests mostly working.

Close to universal access to the test for all high school students without major travel or financial requirements.

It doesn't require a huge amount of student prep time and there is little incentive for teachers to waste class time "teaching to the test".

The fancy tutors and Kaplan classes create some class inequity, but college admissions committees can adjust for it (if they wanted) because the difference they make is not huge.

I think the heated debates arise from ppl worrying that SATs will be all that colleges look at and that admissions committees will be unwilling to adjust based on other factors including socioeconomic status. Maybe worth fearing, but it doesn't play out like that in the vast majority of cases I am aware of.

[marshwiggle]

I think the heated debates arise from ppl worrying that SATs will be all that colleges look at and that admissions committees will be unwilling to adjust based on other factors including socioeconomic status.

Among other things, the margin for error on these measurements is often overestimated by people worrying about "other factors". Even for my won course, I'd agree that two students who wound up 5% apart might not be as different as that indicates, but a difference of 10% reliably indicates something significant.

So, while an absolute numerical cutoff may be statistically unsupportable, the realistic range for adjustment based on "other factors" will still be pretty small, like maybe a few percent.
 

[fizzycist]

I think the heated debates arise from ppl worrying that SATs will be all that colleges look at and that admissions committees will be unwilling to adjust based on other factors including socioeconomic status.

Among other things, the margin for error on these measurements is often overestimated by people worrying about "other factors". Even for my won course, I'd agree that two students who wound up 5% apart might not be as different as that indicates, but a difference of 10% reliably indicates something significant.

So, while an absolute numerical cutoff may be statistically unsupportable, the realistic range for adjustment based on "other factors" will still be pretty small, like maybe a few percent.
 

I'm not sure what is meant by "a few percent".

If you were an admissions officer, would a kid with a 1400 SAT, but a B- average in high school, be on equal ground with a kid with a 1300 SAT and straight A's?

Or a kid with a 1200 SAT, high quantitative portion, straight A's and strong work ethic, from a poor zip code who wants to go into an under-enrolled physics major should be denied for a kid with 1300 SAT, rich zip code, who wants to be in an over-enrolled sociology major?

It's rare that all things are exactly equal except SAT and race or whatever. There are plenty of useful signals throughout the packet. At the R1 state schools I'm familiar with I suspect they do a decent job with admissions overall because most ppl are pretty reasonable when it comes to this stuff in practice.

[kaysixteen]

The main factor in favor of continued use of the SAT is indeed the vast (and I do mean vast) differences amongst hss in this country, not only in terms of grading policies but even in terms of mere titling of courses.

[marshwiggle]

I think the heated debates arise from ppl worrying that SATs will be all that colleges look at and that admissions committees will be unwilling to adjust based on other factors including socioeconomic status.

Among other things, the margin for error on these measurements is often overestimated by people worrying about "other factors". Even for my own course, I'd agree that two students who wound up 5% apart might not be as different as that indicates, but a difference of 10% reliably indicates something significant.

So, while an absolute numerical cutoff may be statistically unsupportable, the realistic range for adjustment based on "other factors" will still be pretty small, like maybe a few percent.
 

I'm not sure what is meant by "a few percent".

Since the SAT isn't a big thing in Canada, my knowledge of it is pretty basic.

Nevertheless, there are no doubt lots of stats out there about what type of variation there is between regions, high schools, etc. so that it can be "calibrated" with other things.
 

If you were an admissions officer, would a kid with a 1400 SAT, but a B- average in high school, be on equal ground with a kid with a 1300 SAT and straight A's?

I certainly wouldn't know, but as I indicated above I'm sure there are lots of people who have analysed the data to determine whether that difference (from 1300 to 1400) is more reflective of the individual or the school. (If they were from the same school, and took the same courses, then the GPA may be a better indicator, but otherwise whether that's a significant difference or not should be pretty well established by lots of data.)

Or a kid with a 1200 SAT, high quantitative portion, straight A's and strong work ethic, from a poor zip code who wants to go into an under-enrolled physics major should be denied for a kid with 1300 SAT, rich zip code, who wants to be in an over-enrolled sociology major?

Since those are completely different programs, I'd imagine they'd have different requirements and cutoffs, so it wouldn't require a choice between those two students.

It's rare that all things are exactly equal except SAT and race or whatever. There are plenty of useful signals throughout the packet. At the R1 state schools I'm familiar with I suspect they do a decent job with admissions overall because most ppl are pretty reasonable when it comes to this stuff in practice.

That's why it always going to be useful as, at least, one more piece of information. If it seems to be wildly out of synch with other information, then there's something else going on that needs to be looked at.

[ciao_yall]

The main factor in favor of continued use of the SAT is indeed the vast (and I do mean vast) differences amongst hss in this country, not only in terms of grading policies but even in terms of mere titling of courses.

Somehow this came up in conversation last night. My friend's son had a 4.xx GPA from a very prestigious high school, a 33 on the ACT (which I guess is pretty good) but a "very low" SAT score. Is that a fluke?

Anyway, he went to Duke and does just fine for himself.

[kaysixteen]

I am totally unfamiliar with the content, mechanics, etc., of the ACT-- exactly how different is it from the SAT, and in what respect does it differ?

[hmaria1609]

Here's an explanatory note from the ACT website.

[spork]

Dartmouth's decision got attention only because Dartmouth is one of the Ivy League universities that U.S. media and economic elites are fixated on. MIT resumed requiring that applicants submit SAT scores two years ago, after a two-year pandemic-driven hiatus. MIT's decision didn't generate nearly as much media coverage.

[kaysixteen]

Thanks for the ACT article, which is of course an ACT house organ piece.  If indeed the ACT is as good relative to the SAT, as this piece claims, why has it not been adopted in lieu of the ACT by most non-Southern unis?

BTW, one of the claims the ACT makes in the piece, as a reason to prefer it to the SAT, concerns the writing sample test it has-- is it not true that the SAT did eventually adopt such a component at some point, only to eventually ditch it?

[Ruralguy]

Yes, there was a writing section on the SAT for maybe 10-15 years?

[apl68]

I took them both back in the day.  But that was a very long time ago, and I couldn't really tell you what the difference was, let alone how they differ today.