Can Intelligence Explain the Overrepresentation of Radicals and Liberals in American Sociology?

A recent study reported dramatic overrepresentation of liberals in American Sociology; intelligence can explain around 30% of this gap, leaving 70% to be explained by other factors

In a recent study, Horowitz et al. (2018) surveyed n = 479 American Sociology professors (response rate = 14%). In their sample, 21% of respondents identified as ‘radical’, 62% identified as ‘liberal’, 2% identified as ‘conservative’, and 2% identified as ‘libertarian’. If one combines the two ‘left-wing’ and the two ‘right-wing’ categories, then 83% of respondents identified as ‘liberal’ (using the American definition of that term), whereas only 4% identified as ‘conservative’.

Consistent with many earlier studies, this suggests that liberals are dramatically over-represented in American sociology, and that conservatives are dramatically under-represented. (The usual caveats about sampling bias apply.) One possible explanation is that (social) liberals tend to have higher intelligence. Indeed, Onreat et al. (2015) obtained a meta-analytic effect size of r = –.20 for the relationship between cognitive ability and socially conservative attitudes.

In order to calculate how much of the liberal over-representation in American sociology can be explained by intelligence, I utilised the method employed in my 2015 paper (see Section 2.1 for details). Specifically, I updated the figures in the second and third columns of Table A.1 by including data from the 2016 wave of the GSS (i.e., data from 2000–2016).

Results were as follows. In the general US population, 27% of people identify as ‘liberal’ and 35% identify as ‘conservative’. In the sub-population scoring 10 out of 10 on the vocabulary test, 44% of people identify as ‘liberal’ and 30% identify as ‘conservative’. Therefore, intelligence may be able to explain 30% of the gap between sociologists and the general population on ‘liberal’ identification, and 14% of the gap between sociologists and the general population on ‘conservative’ identification.

Note that the preceding figures should be considered upper limits, given that the mean IQ of American sociologists may well be lower than ~128 (corresponding to a score of 10 out of 10 on the vocabulary test). If one instead defines the high-IQ sub-population as those scoring 9 or 10 out of 10 on the vocabulary test (corresponding to a mean IQ of ~124), then intelligence can only explain 26% the gap on ‘liberal’ identification, and only 10% of the gap on ‘conservative’ identification.

See Section 4.2 of my paper for methodological limitations.