You know the story, but here we go again. The standard account of sex differences in intelligence is that there aren’t any. Or not significant ones, or perhaps some slight ones, but they counter-balance each other. The standard account usually goes on to concede that males are more variable than females, that is to say, they are more widely dispersed around the mean. Although this is an oft-repeated finding, in some circles it is still referred to a merely a hypothesis. There is a standardisation sample in Ro mania which did not show this difference, and others epidemiological samples where the differences are slight, but usual finding is that men show a wider standard deviation of ability.
Against this orthodoxy, Irwing and Lynn (2006) have argued that boys and girls mature at different speeds, with girls ahead till about age 16 and with boys moving ahead thereafter, such that men are 2-4 IQ points ahead of women throughout adult life.
Lynn further argues that if men are 4 points ahead, and have a standard deviation of 15 as opposed to women’s standard deviation of 14, those two findings almost fully explain the higher number of men in intellectually demanding occupations. There is no glass ceiling. Fewer women are capable of the higher levels required for the glittering prizes. Furthermore, this explains why men know more things. At the very highest levels of ability there are more men, and they have more knowledge, which is why they win general knowledge competitions.
This, the seditious faction suggest, is just a fact of sexual dimorphism. Male brains are very, very much bigger than women’s, and each of the component regions of the male brain are bigger than the same regions in women, and also more variable in size.
Standardization samples ought to be good, and often are so, but they are not as good as birth cohorts or major epidemiological samples, so the latter are to be favoured when looking for reliable sex differences.
However, here is another paper on standardization samples confirming the same pattern of male advantage, though not greater male variability in one of the samples.
Sex Differences on the WAIS-III in Taiwan and the United States
Hsin-Yi Chen and Richard Lynn. Pages 324-328.
Sex differences are reported in the standardization samples of the WAIS-III in Taiwan and the United States. In Taiwan, men obtained a significantly higher Full Scale IQ than women of 4.35 IQ points and in the United States men obtained a significantly higher Full Scale IQ than women of 2.78 IQ points. The sex differences on the 14 subtests are generally similar with a correlation between the two of .65. In the Taiwan sample there were no consistent sex differences in variability.
The authors say:
There are three points of interest in the results. First, in the Taiwan sample males obtained a higher Full Scale IQ of .29d, the equivalent of 4.35 IQ points. This confirms the thesis advanced by Lynn (1994, 1998, 1999) that in adults, males have a higher average IQ than females of around 4-5 IQ points. Males obtained a higher Full Scale IQ in the American standardization sample of the WAIS-III of .185d (2.78 IQ points). These two results disconfirm the assertions of Haier et al. (2004) and Halpern (2012) that “Comparisons of general intelligence assessed with standard measures like the WAIS show essentially no differences between men and women” (Halpern, 2012, p. 115).
Second, the sex differences in the Taiwan and American WAIS-III are generally similar. On the 14 subtests the correlation between the two is .65 (p <.001). Thus, in both samples men obtained their greatest advantage on Information and their lowest advantage on Digit Symbol – Coding.
Third, there was no consistent sex difference in variability. On the Taiwan Full Scale IQ the VR of 1.02 is negligible, and males had greater variability in 9 of the 14 subtests while females had greater variability in 5 of the subtests. These results do not confirm the greater variability of males reported in numerous previous studies e.g., Arden and Plomin (2006) and Dykiert, Gale and Deary (2009).
This study, on the gold standard Wechsler test, seems to confirm a male advantage in general intelligence. As discussed, standardisation samples are designed to be an excellent representation of the population on which the test will be used (with changes to make it culturally accurate), and there is no reason to believe that this balanced selection would favour males. Birth samples would be even better, but this is a good test of the male advantage proposal.
The Information subtest is a measure of very general General Knowledge, not requiring any specialist interests, but asking about the things which would generally be known in the general population. A .44 sd advantage on this subtest is enormous. The greater male representation in high level general knowledge competitions seems well founded. On the US sample there is almost as big a male advantage for Maths, and a large deficit for the digit symbol coding task, which measures simple processing speed.
The lack of a greater standard deviation in the Taiwanese sample goes against the general finding, as did the standardisation sample for Romania. Standardisation samples are not as representative as larger epidemiological surveys, but it is interesting nonetheless, in that it suggests some sampling restriction.