Many times on this blog I have tried to estimate the IQs of various groups or individuals using simple regression, or in some cases multiple regression (a technique suggested to me by a Promethean).

Back in 2017, commenter Gypsy asked me to apply simple regression to Oxford and Cambridge students. He estimated that only one in 58 UK 18-year-olds qualify for these schools and said the correlation between IQ and GCSE was 0.7.

I replied by saying:

If one in 58 UK teens achieves A*A*A or better, then this implies that the median teen who obtains A*A*A or better is at the one in 116 level in academic achievement.  If there were a perfect correlation between IQ and academic success in the UK, we’d expect the median IQ of these academic superstars to be 36 points above the UK mean of 100 (white norms), but since the correlation is probably about 0.7 (using the correlation between IQ and GCSE as a proxy), their expected average IQ would be 0.7(36) points above 100, or IQ 125.

Well it’s very rare that one of these theoretical statistical predictions gets tested by actual empirical data but when Gypsy recently inquired about the Raven scores of UK university students in general, I stumbled upon a startling fact. The median Oxford undergrad obtained a raw score of 27 out of 36 on the Advanced Progressive Matrices Set II:

As table 16 shows, that equates to the 75th percentile among the general young adult population of both the U.S. and the U.K.(circa 1992-1993) which equates to an IQ of 110 (U.S. & U.K. norms).

However the Oxford students had a 40 minute time limit and the general population could stay almost as long as they needed to. How big a difference does the time limit make? In one study, raw scores increased by about 2 points when people had unlimited time compared to just 40 minutes.

Thus, it seems reasonable to assume that if given unlimited time, Oxford undergrads would have averaged 29 instead of 27, moving them up to the 84th and 86th percentile of U.K. and U.S. norms respectively.

I estimated how Oxford undergrads would have scored untimed because it’s hard to find a general population sample in the U.S. or U.K. who took the test in 40 minutes like the Oxford students did, however table 16 does have timed stats for UK retail managers, which seems like the kind of average level occupation that would reflect the general population. A score of 27 in 40 minutes puts one in the 85th percentile of this group (virtually identical to the percentile of their estimated untimed score in general population samples).

This equates to an IQ of 116 (UK norms). This is 9 points lower than the IQ of 125 I had predicted in 2017, however it should be noted that the the Raven is a uni-dimensional test that only correlates 0.74 with more comprehensive tests like the WAIS. Thus it’s likely that on the WAIS, instead of scoring 15 IQ points above 100, they would have scored 15/0.74 = 20 IQ points above 100.

We also don’t know when the data was obtained. I’m assuming it was in the 1978 to 1993 era, however if it was substantially before then, the Flynn effect might have greatly deflated the score.

We also don’t know how representative the Oxford undergrads were. I’m guessing they were psychology students, who are likely fairly representative because I know at least at the University of Toronto, 60% of all undergrads take psychology.

But taking things at face value, it seems the most prestigious university in Europe has an average IQ of 120. No wonder David Wechsler defined IQs of 120+ as “superior intelligence”.