56 Comments

I assume this was using SAT scores from the past 20 +/- years. When I tested in the early 1980's, a score of 1600 would have been regional news and more times than not resulted in allegations of cheating. With today's relaxed scoring, there were at least a handful of perfect scores in my children's school each year, making intergenerational comparisons difficult.

Expand full comment
Apr 15, 2022·edited Apr 15, 2022

How many times was the SAT re-centered? Many of us have scores that are from 30 or 40 years ago. Wondering how those scores are relative to today’s scores.

Expand full comment

What!? Failure to separate physics and philosophy? Fatal error! Both of these disciplines should be merged together and separated from all others to form a single field called LORDS OF HUMANKIND

Expand full comment

What about SAT and COMPLETED undergraduate degree?

Expand full comment

The minimum score on the SAT is 400 - you get that for signing your name. The second chart that has scores going down to 200 should actually cut off at 400.

Expand full comment

I have followed the blog for some time now and usually enjoy your analysis.

However, the SAT-IQ Z score comparison exemplified a mathematically correct but theoretically wrong estimate. Such is the nature of psychometrics, though. I would think of those IQ correlations as the minimum cognitive requirement to attain that specific SAT score — not the average.

The reason why is that most intelligent students (arbitrarily defined as having an IQ >120) take either the SAT or ACT — sometimes both. This elite sample skews the cohort and distorts the results in comparison to the general population. Therefore, it makes little sense to assume — even if mathematically correct — that the average person who scores a 1400 on the SAT (97.4 percentile) has an IQ of less than 125.

Any score above a 1400 allows most students to gain admission to competitive schools. I would comfortably estimate that an SAT 1400= IQ 130+, on average. It becomes difficult to calculate a precise number insofar as the correlation is imperfect; however, the Z score correction is theoretically implausible.

Expand full comment

Theres a reason they measure IQ via different testing means. IQ is no predictor of ethics or morals tho and those seem to be whats gone missing in the last couple decades.

Expand full comment

I’m a classics student haha!

Expand full comment

You can do a maths degree with a 116 IQ??

Expand full comment

"1600 SAT is about 135 IQ" I don't think this is correct. The article that provides this conversion has estimates of one-time SAT scores and one-time IQ scores. IQ scores' reliabilities are estimated using the same test at two different occasions. (Perhaps two versions of the same test.) But this isn't how we understand IQ/g. A better estimate of g reliability would be obtained using two different tests (e.g., Ravens and WAIS) the same time distance apart as before. I'll bet reliability will be much lower–correct me if I'm wrong. Also correct for SAT reliability. People who get 99th percentile in various standardized tests consistently are smarter, on average, than those who get it once.

Expand full comment

SAT has become more studiable basically every year since the 90s, I doubt it is 0.8 at this point. I know a lot of kids who went hundreds of points up on their second go-around just from studying, or got hundreds of points lower on their PSAT than on their SAT. In the 80s and 90s a score of 1250 was enough to get you into MENSA, while today a score of 1250 is not particularly significant at all.

Expand full comment

Every smart kid takes the SAT. There is not some hidden cache of geniuses not taking this test.

Nobody with a 145 IQ is botching a standardized test.

Each year only about a thousand kids get a perfect SAT score nationwide. Fewer than one in a thousand. It is that rare.

Every year there are surely far more than 1,000 kids with an IQ of 135 + (that’s only a shade over two standard deviations above the norm - still sub genius - more like a one in a hundred IQ). In fact there are more than 10,000 kids with that IQ. As measure by IQ tests. Not SAT.

You’re pretty obviously mistaken about the avg intelligence of a perfect SAT scorer. It’s closer to 145.

Expand full comment

Dear Dr Kirkegaard, SAT is an extremely preppable test. Therefore any interference with IQ can only be made with the first attempt, unprepped score that is unavailable to you. The actual scores students receive only has a tentative link to their IQ, and are a much greater reflection of their educational opportunities and the prep company they used. That said, I do believe SAT correlates well with ability to succeed in college, where prepatation and a strong background are a tremendous help in absorbing knowledge. I would hesitate, however, to link it to any inherent inborn intelligence. :-)

Expand full comment
Apr 17, 2022·edited Apr 17, 2022

If we get extremely technical, you are partially correct about minimum IQs. However, research on how grit and IQ correlates to scholastic achievement shows no improvement for those below 85. We can at least interpret that to mean no person with an IQ below 85 could become a surgeon, regardless of effort.

An IQ of 123 compared to 130 appears minors, true. The difference, in reality, is quite sizable. But I still believe the math distorts the IQ of high-scoring test takers. I cannot imagine that many students with an IQ >125 scored below a 1350.

Nevertheless, these are extremely technical disagreements. Even more so when considering the fact that many believe the test measures only wealth, etc. Thanks for responding!

Expand full comment

I'm OK with these being the means. Having worked with various professionals for many years, it sounds about right. Especially when taking into account inflation in test scores.

40 years ago, a score above 1500 on the SAT was pretty much unheard of.

What would be interesting is the 95th/5th percentile IQ's by score.

Expand full comment

Mensa has refused to accept SAT and ACT scores for membership about a quarter century now. The correlation between those and IQ isn't what it used to be.

Expand full comment