SAT concordance table - compare old and new SAT scores

@mdphd92 wrote:

Can’t speak for Bowdoin, but, Wesleyan requires all frosh to submit their scores once they decide to matriculate.

@wisteria100 looks like Bowdoin did convert their new SAT based on concordance. See the CDS below… They even highlighted it in yellow. Also, u mention an over 90% stat of students submitting SAT or ACT at Bowdoin… Where did u see that. I’ll look more closely but CDS says ~50% submitted for each test but many submit both and some submit none… not sure if they provide the figure of % who don’t submit either one.

https://www.bowdoin.edu/ir/data/cds-table.shtml

Sorry but those numbers for Bowdoin and Wesleyan are just crazy.

@suzyQ7 yeah the Bowdoin CDS does say they convert the SAT scores, but the thing is that was the CDS for class of 2020, and the new SAT didn’t start until March of 2016, so that class wouldn’t be taking that new version. Right?
As for the 90% number, I know I heard it somewhere, but when pressed I can’t recall the source. But I do think looking at the numbers for Wes and Bowdoin, that it is a very high percentage submitting. If you look at schools like Midd, Swat, Colby that require scores, the CDS % for submitters of both tests added together range from about 109 to 116. So while there is some overlap of kids submitting both tests, it’s not all that high. So if Bowdoin and Wes are reporting approx 95% for both combined, there may be some overlap, but probably not more than 10-15 points.

Swat’s numbers for class of 2021 were recently posted. They show 1370-1540 for their middle 50%. Although they do not specify if it is for enrolled or admitted, it is almost certainly admitted, not enrolled. Otherwise, they are just 10 points under Princeton’s 1380-1540, which I think we can agree is highly unlikely. Moreover, the 16-17 CDS shows 1305-1530 for enrolled students. While it is unclear if they are using the Concordance or not, these numbers are probably a pretty good indicator of an honest range for the top LAC’s. Again, Wes, Bowdoin, et. al., have some explaining to do. Regardless of whether they are test optional, no way those are actual numbers.

http://www.swarthmore.edu/sites/default/files/assets/documents/admissions-aid/2017%20Fact%20Sheet.pdf .

BTW, Colby also reflects lower new SAT scores. https://www.colby.edu/admission/college-profile/
Middle 50-percent Old SAT scores 1350-1520
Middle 50-percent New SAT scores 1330-1510

Not dramatically lower but the Concordance table predicted the old 1350-1520 would result in a new SAT of 1410-1540. That’s a big dif.

Again, I’ve yet to see new scores that are higher than old.

@KennyT Yes. That’s a big difference between 1520 and 1540 for sure. And Bowdoin’s 770? No words.

Amherst broke it down for the ATTENDING freshmen (not the accepted freshmen, yet, presumably those numbers would be higher).

  • assuming those are the 50 figures.

https://www.amherst.edu/news/news_releases/2017/8-2017/welcome-new-students

2017 CDS is not out yet.

A new 1469 is not quite the 1530 the Concordance predicted a 2232 would yield.

Good grief, that’s a huge difference!

It also looks like the ACT range at a lot of colleges got bumped up by one or two points. It could be that a lot of students took only the ACT OR that it was easier for a student to hit a 31-32 in the ACT than the concorded SAT number. Of course now that students had more practice time and more chances for super scoring those SAT numbers will rise. Interesting to see next years data.

Pomona and Williams also posted: https://www.questbridge.org/college-partners/pomona-college and https://www.questbridge.org/college-partners/williams-college

Williams:
SAT EBRW Middle 50%: 710-780
SAT M Middle 50%: 690-780
ACT Middle 50%: 31-34

Pomona:
SAT EBRW Middle 50%: 680-750
SAT M Middle 50%: 670-760
ACT Middle 50%: 31-34

Pomona and Williams had the exact same scores for enrolled students last year (670-770 per section SAT, 31-34 ACT), so it seems Pomona is going by actual data (their new SAT ranges are slightly lower) and Williams has used the concordance tables.

For comparison with Princeton and Stanford-

https://admission.princeton.edu/how-apply/admission-statistics
Math: 700-780
Evidenced-based Reading and Writing: 680-760
ACT Composite Score: 31-35

http://admission.stanford.edu/apply/selection/profile.html
SAT Math Section: 700-780
SAT Evidence-Based Reading and Writing: 690-760
ACT Composite: 32-35

Both of these are lower than previous old SAT ranges.

Vassar has an interesting breakdown for both: https://admissions.vassar.edu/about/statistics/

The old SAT range is 1350-1470. The new SAT range is 1380-1470. So technically, the New SAT is scoring higher. However, if you take a look at the percent scoring a 750+:

Old CR: 31.5%
Old W: 26.1%
Old M: 31.5%
New EBRW: 23.9%
New M: 27%

There are also far fewer scores with a 640 or below:

Old CR: 16.9%
Old W: 18.5%
Old M: 17%
New EBRW: 6.1%
New M: 9.8%

Note that there were twice as many enrolled students submitting new SAT scores compared to old ones, suggesting that they weren’t disadvantaged in the process.

I do appreciate Vassar’s detail. So far, this is the only one I’ve seen where the new SAT range is higher than the old SAT. However, it is on somewhat lower numbers. With all of these, it’s important to use the Concordance table to see what the old SAT was predicted to be in the new SAT. With Vassar, the old SAT range of 2030-2210 was predicted to a 1430-1520 range. The actual scores still came out some 50 points lower than predicted.

@nostalgicwisdom That’s interesting. Wiliams appeared not to use the concordance tables in actual admissions decisions because it explicitly separated out Old and New SAT scores when it reported data last spring (with New scores averaging lower than Old).

https://communications.williams.edu/news-releases/3_23_2017_admittedstudents/

Perhaps the middle 50 data concorded the Old scores to New - I suspect that the Common Data Set might require that approach although I think that would render the middle 50 data unhelpful.

The Vassar distribution shows that scores of 750 and above are less common on verbal (31.5% on old CR, 23.9% on new EBRW) and about the same in math (31.5% on old Math, 27.0% on new Math). However, scores of 700 or above are more common on verbal (60.0% on old CR, 66.5% on new EBRW) and about the same frequency on math (57.6% on old Math, 58.6% on new Math).

For the new Math test, at least, the scores seem to be about the same as the old Math test. This is also supported by my post #332 (on page 23 of this topic), where the middle 50% of scores are about the same over two successive years for all 9 University of California campuses.

I think the evidence shows that the new SAT is easier than the old for students in the middle range of scores (under 1400) and harder at the tippity top.

Not to quibble, but 1400 is almost certainly above 95th percentile which is not quite middle. However, from what I’ve seen I agree that above around 1380, the new SAT is harder than the old, is about even between 1340 and 1380 and a little easier under 1340 or so. Under every scenario I’ve seen, the Concordance table predicts much higher new SAT scores than actually occurred.

Have you seen other schools that only list middle 50 for the old SAT for the Class of 2021 profile? No mention of the new SAT.

http://www.admissions.caltech.edu/content/class-profile