One “urban legend” which is in common circulation among my friends is that liberals are smarter than conservatives. From my own personal experience this seems plausible, and I doubt I’m the only one as evidenced by the furious speed at which the “Bush voting states have lower IQs” meme spread around the blogosphere several years ago. But is this true? I’ve seen enough data to suggest that this really isn’t so, and my psychologist friends have told me the biggest predictor of liberalism isn’t IQ, but a strong tendency toward “openness” on personal tests.
But I just couldn’t leave it alone…I decided to look at the General Social Survey and see what it tells us about political ideology and intelligence. The GSS has a variable, POLVIEWS, where people are classified as extremely liberal to extremely conservative where those states are equivalent on a 1 to 7 scale. In other words, if you are 4 you are a moderate, 1 extremely liberal, and 7 extremely conservative. The WORDSUM variable just measures the number of correct answers on a vocabulary test of 10 words. 0 means you’re semi-sentient and 10 means you have a good vocabulary.
Instead of producing garish bar graphs I thought I would just look at correlations. Though political ideology is categorical here, it is nicely translated into a numerical scale where rank order exists. The “higher” numbers are more conservative, the “lower” ones more liberal. For vocabulary there is a natural numerical scale. The question: do political ideology and vocabulary track each other at all? Yes, but hardly. The raw results are below, but I don’t really think that the story ends here, so read on.
|Correlation between political views and correct answers on vocabulary test|
|WORDSUM 0-10||-0.03||United States|
|-0.02||East North Central|
|-0.04||West North Central|
|0.11||East South Central|
|-0.10||West South Central|
|WORDSUM 0-4||0.06||United States|
|-0.01||East North Central|
|-0.01||West North Central|
|0.11||East South Central|
|0.08||West South Central|
|WORDSUM 5-10||-0.06||United States|
|-0.07||East North Central|
|-0.06||West North Central|
|0.11||East South Central|
|0.04||West South Central|
|0.02||Less Than High School|
The table should be rather intelligible to you. I sliced & diced the data and analysis in a few different ways. The rightmost columns shows how I limited the data set by region or educational attainment, while the leftmost column shows that I constrained the vocabulary spectrum to the less intelligent and more intelligent. I wanted to see if interesting patterns were masked by throwing all the data together into one lump.
I think there are signals being obscured. Note that the overall correlation coefficient of political ideology and vocabulary results is a nearly non-existent -0.03. If it’s a linear system you’d square -0.03 to determine how much of the variation of Y you could predict from X, pretty much nothing. But look what happens when I take the less intelligent and separate them from the more intelligent; the signs switch! (a negative would indicate that conservatives tend to be duller than liberals, and a positive the inverse) Nevertheless, the correlation coefficient is really, really, small (it averages out with a negative sign because there are actually considerably more people in the 5-10 WORDSUM interval). But notice what happens as you increase educational attainment: the difference starts to show up significantly at the graduate school level. The chart to the left shows that the frequency distribution of scores for liberals and conservatives vary; liberals exhibit more variance. This was what my hunch was, extremely fiscal liberals are probably often poor and unintelligent, extremely social liberals are probably moderately affluent and intelligent. Conservatives on the other hand have a bigger mode than liberals, they’re concentrated closer to the center of the distribution.
The WORDSUM variable really doesn’t push things to the tails very much. 6.5% of the sample received a 10. When we think of pointy-headed liberals culturally I think we’re assuming Harvard or Stanford, but the student bodies there are way above 6.5%. As you know moderate differences in distribution can loom very large at the extremes, so I wouldn’t be surprised if some of the perception by elite liberals that conservatives are stupid might simply have to do with the fact that at their cognitive station the imbalance might be noticeable (it might be that at such prestigious universities political indoctrination/consciousness raising by the Leftish professoriate is especially effective). One might also consider the possibility that if half of the top 0.25% of students go to 10 elite schools which are very liberal, and the other half go to 100 less elite schools, you might get a critical mass effect so that very smart liberals socialize together and produce a subculture where cognitive status is jockeyed for and its exhibition expected. In contrast, smart conservatives who went to State U for whatever reason (e.g, perhaps they had Christian conservative qualms about applying to bastions of liberalism?) might become entrepreneurs or successful middle managers who don’t live in a subculture where their intelligence is as salient.
Shorter version: There isn’t a large difference in intelligence between liberals and conservatives, on average. But it might be that the small differences are highly significant in our day to day life.
Note: All the above applies only to whites as I restricted the data set.
Bonus: From 2006 Exit Poll of US elections