Liberals marginally more intelligent than conservatives, who are considerably smarter than moderates

In response to Dennis Mangan's recent post discussing Satoshi Kanazawa's speculations as to why political liberals dominate Western institutions, I left the following comment. Rather than rehash it as a stand alone post, I'll just offer it again here. The body of Mangan's post offers fuller context if the line I excerpted from it is tough to comprehend:

---

the General Social Survey show that there's nearly an 11-point childhood IQ difference between those who identified (as adults) as "very conservative" and "very liberal", with a monotonic increase between the two.

Kanazawa isn't quite correct in this assertion (and I encourage those who can tolerate a clunky interface to verify as much for themselves). Mean wordsum scores for all white GSS respondents from the survey's inception to the present, by political orientation:

Extremely conservative -- 5.98

Conservative -- 6.35

Slightly conservative -- 6.51

Moderate -- 5.97

Slightly liberal -- 6.54

Liberal -- 6.55

Extremely liberal -- 6.57

Using a purely verbal test as a proxy measure of IQ has the effect of artificially inflating women's scores relative to men's. The effect is modest--women have a .15 point advantage over men on Wordsum--but when it comes to politics, where the gender divide is not insignificant, it shouldn't be discounted. The same, this time for men only:

Extremely conservative -- 5.98

Conservative -- 6.29

Slightly conservative -- 6.45

Moderate -- 5.72

Slightly liberal -- 6.39

Liberal -- 6.29

Extremely liberal -- 6.51

Those who describe as "extremely liberal" and "extremely conservative" together constitute just 5% of the total respondent pool. Cutting them out and looking at the remaining 95%, we see that liberals and conservatives are of about equal intelligence, with moderates coming in around 5 IQ points lower.

Of course, when it comes to dominating media institutions, we're looking at the right tail of the intelligence distribution, and those who describe as "extremely liberal" are going to cluster here more than people with other political outlooks will. Further, the Wordsum distribution is wider for liberals than it is for conservatives--liberals are more likely to score in the 0-2 and 9-10 ranges than conservatives are, while conservatives are more likely to score in the 3-8 range than liberals are. The standard deviation for white men's Wordsum scores, by political orientation (the larger the standard deviation, the more variance there is in scoring among those holding the political viewpoint):

Extremely conservative -- 2.05

Conservative -- 2.00

Slightly conservative -- 1.99

Moderate -- 2.01

Slightly liberal -- 2.19

Liberal -- 2.30

Extremely liberal -- 2.68

On so many dimensions, the political situation in the US has become one of the top and bottom in alliance against the middle.

GSS variables used: WORDSUM, RACE(1), SEX(1)(2), POLVIEWS