Parents and parent-age people: Compare your SAT scores to those of current high school students

… using these concordance tables for the 1995 recentering and 2016 redesign.

https://files.eric.ed.gov/fulltext/ED563025.pdf shows the effect if the 1995 recentering of SAT scores.

https://satsuite.collegeboard.org/media/pdf/higher-ed-brief-sat-concordance.pdf shows the effect of the 2016 SAT redesign.

Here are the concordances in 50 point increments of the pre-1995 scores. The concordance tables in the linked pages include scores between the 50 point increments.

Section Pre-1995 score 1995-2016 score Post-2016 score
V/CR/EBRW 300 380 420
V/CR/EBRW 350 430 480
V/CR/EBRW 400 480 520
V/CR/EBRW 450 530 580
V/CR/EBRW 500 580 620
V/CR/EBRW 550 620 660
V/CR/EBRW 600 670 720
V/CR/EBRW 650 710 740
V/CR/EBRW 700 760 780
V/CR/EBRW 750 800 800
V/CR/EBRW 800 800 800
M 300 340 380
M 350 400 440
M 400 440 480
M 450 480 510
M 500 520 550
M 550 560 580
M 600 600 620
M 650 650 670
M 700 690 720
M 750 760 780
M 800 800 800
2 Likes

Didn’t the SAT have a logic section? I seem to remember that in the early 80s… or was that part of verbal?

In the 1980s, there was a logic section (called analytical) in the GRE general, whose other two sections were very similar to the SAT at the time.

I do not recall a logic section in the SAT of the time, but there was an English grammar section that was a separate test (not part of the SAT score, but a separately scored section that seemed to be universally ignored). The English language (not literature) achievement test at the time was basically a longer version of that English grammar test. At some point in time, it was turned into a writing sample, then moved from being a separate test (by then, the achievement tests were renamed SAT II) to becoming the third section of the SAT (part of the 2004 changes; the SAT II were renamed SAT subject tests), then dropped after test prep companies figured out how to game the scoring rubric (part of the 2016 redesign).

2 Likes

Is this column supposed to say Post-2016 score?

Yes, corrected.

1 Like

Oh, maybe I’m confusing it with the GRE then. Thanks!

LSAT has a logic section.

2 Likes

Rescaling the SAT scores listed in the October 3, 1960 Life magazine (more like the time when high school students grandparents were applying to college) shows the following:

College 1960 SAT V 1960 SAT M Converted to 2016 SAT EBRW Converted to 2016 SAT M
Amherst 638 673 740 690
Columbia 657 673 760 690
Princeton 641 675 740 700
Swarthmore 650 675 740 700
Yale 640 665 740 700
Stanford 641 635 740 660
Michigan 585 600 700 620
Iowa 588 580 700 610
UCB 587 575 700 610
Grinnell 566 569 680 600
NYU 549 537 660 580
Pittsburgh 512 540 640 580
Virginia 525 554 640 590
Vanderbilt 540 550 660 590

What is interesting is that the converted math scores are lower than the converted verbal scores. Could there really have been much less emphasis in math in high school in 1960?

1 Like

Does this factor in the influx of Asian immigrants ,who as a demographic group, score higher on SAT math than the US general population in 1960?

1 Like

I poked around and this is supposedly a 1980 SAT SAT 1980.pdf | PDF Host

and there’s more here linked in a reddit post

I didn’t see any 1960s SATs, but my google search was very brief. I expect there are some out there somewhere.

1 Like

While the kids immigrants who came as PhD students and skilled workers in math-heavy areas (e.g. engineering and computing) skew today’s Asian SAT score averages upward and even more so the math section score averages (while having a much smaller effect on other racial and ethnic groups due to relative numbers of those immigrants versus the pre-existing or other immigrant populations in the other racial and ethnic groups), there does not seem to be that much of a current skew toward the EBRW section compared to the math section overall (<20 points in favor of the EBRW in every other racial/ethnic group, according to Fast Facts: SAT scores (171) , versus the much larger differences in the converted 1960 scores).

Note that some of the colleges in the table above (Amherst, Columbia, Princeton, Yale) were male-only in 1960, but they still exhibited a section skew in favor of the verbal-converted-to-EBRW score over the math score despite the current tendency of male SAT takers to do (slightly) better on the math section.

These are nit higher average SAT scores across the population. These are the average SAT scores for these colleges.

So the reason is far far simpler than trying to calculate proportion of people from different ethnic groups of SES.

There are two actual reasons. The first is that with more applicants, and a lower acceptance rate, the colleges can select students with high SAT scores.

However, there is more than that. “Holistic” meant that students of specific SES, ethnicities, race, and high schools were preferred, and college-bound students knew it. So a student for the right high school and the right family would apply to, say, Amherst, even if they didn’t have the best test scores.

So, in 1960, acceptance rates to Amherst of students with verbal SAT scores of over 700 was 36%, and for students with SAT scores 600-700 was 25%. In 1974, it was 25% and 16%, as the SAT scores of applicants started climbing.

That meant that a large number of students were being rejected while many students with lower SAT scores were being accepted.

Interestingly, Amherst seems to have held similar acceptance rates for a while. In 1960 it was around 20%, and in 1970 it was almost the same.

I just spoke of AMherst, because they have the data available:

1960:
https://acdc.amherst.edu/explore/asc:927535/asc:927547

1970:
https://acdc.amherst.edu/explore/asc:927329/asc:927330

It matters not one whit that my old SAT scores would’ve been a 1600 nowadays. I still wouldn’t have gotten into today the Ivy that took me back then, and probably not a T50 school, and I bet that most of the parents on here would’ve been in the same boat. Everything has changed in the college admissions game.

1 Like

@ucbalumnus Where’s 2005 in this chart? Why exclude that major change?

Not sure what you are trying to say with this random data chart. That scores shifted a tiny bit for some score deciles after redesigns? All of that is disclosed to students and families when they sign up to take the test. They don’t mention the redesigns, but they explain how the test is scored for the current students’ test.

Redesigning the test doesn’t make it invalid. In fact, test design has improved as a whole quite dramatically since 1960 and even the 1980s as understanding of assessment techniques has evolved.

Removing disjointed data from context is not going to help people understand or appreciate your point, whatever it is.

Hm. We could start a thread complaining about how confusing the college admissions process is and how there isn’t enough money on the planet to pay enough to people to help anyone guide us through it…wait. Hold my beer.

2 Likes

A 2005 old to new SAT conversion does not seem to exist, and there seem to be a few scattered comments about just adding or ignoring the writing score to convert.

So there were no changes to the SAT before 1995? I find it hard to believe that all pre-95 scores were on the same basis. FWIW, I think I took it in 1976. Based on your table my score would bump up some but I’m not sure what that proves.

It does not prove much of anything other than that one has to be aware of the rescaling effects when comparing today’s SAT scores to those you or others may have gotten in earlier generations.

It may also be relevant if you work in an industry where job applicants are asked for the SAT scores from high school even after they have long since graduated from college. Not accounting for the rescalings may result in age discrimination against older people whose SAT scores appear lower due to the rescalings.

That would require me to remember my SAT score :thinking:

It was a good score, and it felt important at the time that I took it, but it didn’t help me with the one flagship college I applied to. I’d already forgotten it by the time I was done with college. But somehow I still remember my friends’ childhood phone numbers which are even more useless :slightly_smiling_face:

2 Likes

It seems like this table would get you in the ballpark for comparison.

My SAT in 1989 was relatively high and helped me get into 4/5 schools where I applied. I was accepted at Johns Hopkins and Carnegie Mellon, but not Princeton. Based on this chart, it would be a bit over 1500 now, which seems fairly accurate.