Estimating the IQ of an Eminent Scientist
Roe has already tried this - she sampled the IQs of 64 eminent scientists using a test normed on 10 PhD students to try to make a very high ceiling test. Given the small sample sizes of the groups, this isn’t a particularly reliable estimate, but it’s much better than no information. The averages on the 3 tests she created were 166 on the verbal, 137 on the spatial, and 154 on the mathematical. Averaging out these 3 subtests yields an estimate of 152 for the IQ of an eminent scientist.
If you know the correlation between job performance and IQ within a field, it should also be possible to estimate what the average IQ of an eminent individual in that field is. If we assume that the cutoff for an eminent scientist is roughly a 200 in a billion, then they are above 5.1 in z-score notation (according to my simulation, a these would average 5.2 in scientific eminence). There is one way you can estimate the correlation between scientific achievement and IQ → assume it’s equivalent to the correlation between job performance and IQ in highly complex jobs. According to the Hunter and Schmidt meta, the correlation between IQ and job performance in the most complex jobs is 0.58. Multiplying both numbers (5.2 x 0.58) leads to an estimate of 145.
Averaging out these two imperfect methods yields an estimate of 148.5 for an eminent scientist (round up to 149 for funsies). Seems a little high to me personally, I think it’s closer to the lower 40s or so.
g <- rnorm(900000000)
uzi <- rep(0, 900000000)
uzi
subby1 <- data.frame(g, uzi)
subby2 <- subset(subby1, subby1$g > 5.2)
mean(subby2$g)