Are leftists genetically more intelligent?
Two caveats about a recent study.
Written by Noah Carl.
A study on the relationship between intelligence and political views recently had people bickering on Twitter/X. The study, by Tobias Edwards and colleagues, found that intelligence was positively associated with six different measures of left-liberal attitudes in a sample of American adults. Not only that, but polygenic scores for intelligence were positively associated with all six measures too. This provides evidence that, on average, individuals with left-liberal attitudes are genetically more intelligent than their counterparts on the right.
The study itself looks to be careful and rigorous; I have no complaints about the methodology. What I will do is outline a couple of caveats.
To begin with, I would suggest the headline finding that polygenic scores for intelligence are positively associated with political views doesn’t tell us a great deal we didn’t already know. This is not to say it isn’t interesting and worth publishing; just that it’s not particularly surprising. (Unsurprising findings are often worth publishing.)
Many studies have observed that cognitive ability is positively associated with socially liberal attitudes and liberal self-identification. This could be because cognitive ability causes political attitudes, because political attitudes cause cognitive ability, or because some third variable causes them both. It seems rather unlikely that political attitudes cause cognitive ability. Your friend converts you to left-liberalism, and this bumps up your IQ score? Doesn’t sound right.
As for some third variable causing both cognitive ability and political attitudes, the only one I can come up with is education. Perhaps going to college boosts your IQ by a couple of points, while shifting you to the left politically? However, we can rule this out as the main explanation for the association between cognitive ability and liberal attitudes based on studies that measured cognitive ability before subjects went to college. Ian Deary and colleagues found that g measured at age 10 predicted liberal attitudes at age 30. Likewise, Ingrid Schoon and colleagues found that g measured at age 11 predicted liberal attitudes at age 33.
Hence by the far the most plausible explanation for the association between cognitive ability and liberal attitudes is that the former causes the latter. Now, it’s conceivable that only the non-heritable variation in cognitive ability has a causal impact, but this just doesn’t seem very likely. Most of the non-heritable variation is not shared between siblings raised in the same household, so you’d have to argue things like developmental noise and getting hit on the head are what shape our political views.
Genes causing cognitive ability and cognitive ability causing liberal attitudes was always the most plausible pathway. Again, this is not to detract from Edwards and colleagues’ study (which is excellent). The point is that conservatives indignant at the headline finding don’t need to be any more indignant than they already were.
My other caveat concerns effect sizes: Edwards and colleagues’ are at the extreme end of those that have been reported in the literature. Here’s the main figure from their paper. Focus on the violet circles for “phenotypic IQ”. These represent the unadjusted correlations between intelligence and the various measures of political views.
The effect size for social liberalism is around 0.38, while the effect size for authoritarianism is around –0.42. Averaging across the two, this is double the meta-analytic effect size of –.20 reported by Emma Onraet and colleagues, based on 67 studies.
Now, Edwards and colleagues’ study is more recent and it’s possible that more recent studies tend to yield larger effect sizes. However, Onraet and colleagues found no evidence of this in their moderator analysis: the average effect size for studies published since 1990 was lower than that for studies published before 1970. Edwards and colleagues also used a comprehensive IQ test to measure intelligence, which you’d expect to yield a larger effect size owing to greater reliability. Yet in Onraet and colleagues’ moderator analysis, the average value for studies that used general ability was only marginally higher than the overall value.
The same applies to political orientation (liberal versus conservative). Edwards and colleagues’ effect size is around 0.23, which is almost double the value you get in both the General Social Survey and the American National Election Survey – using roughly the same scale to measure political orientation and using the Wordsum vocabulary test to measure intelligence.1
True, part of this disparity may be explained by Edwards and colleagues having used a comprehensive IQ test, rather than a 10-word vocabulary test – though the aforementioned meta-analysis suggests it may not be a large part. Moreover, the association between intelligence and liberal ideology appears to be driven by verbal intelligence, so you’d expect Wordsum to slightly understate the general intelligence of conservatives.
Scan down to the bars for fiscal conservatism. Here Edwards and colleagues made a finding that’s directionally inconsistent with the prior literature: they observed a negative association between cognitive ability and fiscally conservative attitudes, whereas most previous studies have found the opposite.2 Jedinger and Burger reported a meta-analytic effect size of 0.07, as compared to about –.20 in the figure above.
Edwards and colleagues suggest this disparity may be accounted for by the recency of their data on political attitudes (which were collected between 2017 and 2022). Over the last decade or so, there has been political realignment around cultural issues, whereby educated whites have shifted toward the Democrats while working-class whites have shifted toward the Republicans. As a consequence, those two groups may have come to adopt the economic attitudes of their newfound co-partisans – meaning that whites with higher cognitive ability may be more likely to support economic leftism.
There could well be some truth in this. Yet despite similar realignments having taken place in Europe, there’s not yet evidence of a reversal in the association between cognitive ability and fiscal conservatism. One of the largest moderators in Jedinger and Burger’s meta-analysis was whether the study had been done in the US or Europe, with European studies yielding consistently larger effect sizes. For example, Johanna Mollerstrom and David Seim found a correlation of 0.25 between cognitive ability and opposition to redistribution in Sweden.
All this leads me to suspect that the effect sizes Edwards and colleagues observed may not be generalizeable to other contexts. Which doesn’t mean the authors did anything wrong – on the contrary, their methods are exemplary. It may simply be that their dataset (the Sibling Interaction and Behaviour Study) shows an unusually strong relationship between intelligence and left-liberal views. Perhaps there is something unusual about Minnesota, the state of which the sample was meant to be representative? I’m just speculating here.
All in all, Edwards and colleagues’ study is a valuable contribution to the literature on intelligence and political views. But we should keep the foregoing caveats in mind.
Noah Carl is Editor at Aporia.
Consider supporting Aporia with a paid subscription:
You can also follow us on Twitter.
The unweighted correlation between Wordsum and political orientation in the GSS, for the years 2016–2022, is r = .13. The unweighted correlation between Wordsum and political orientation in the 2016 wave of the ANES is r = .12. Both datasets include the seven-point version of the political orientation scale, whereas Edwards and colleagues used the five-point version.
I'd add a third caveat, from page 4:
"63% of the included sample is female compared to 54% at intake, suggesting males were more likely to drop out."
That's almost 2 females for every male in their sample. Anyone ever notice anything about the politics of high-IQ females?
It’s bleakly amusing that anyone takes these studies seriously. Academia runs at about 90-94% left leaning now. Any academic study of IQ in relation to political affiliation will be intending to find a correlation between high IQ and leftism. The former editor of the Lancet some time ago concluded that about 50% of scientific studies are pure horseshit, such is the level of bias and ideological distortion (as well as financial incentive) applying to these studies in order to reach predetermined conclusions. You can bet that bullshit percentage is much higher for any directly political question that will provide ammunition for leftist sneers and vanity. Not only that, but leftist attitudes to IQ testing are riddled with cognitive dissonance and obvious contradictions. When IQ testing indicates racial disparities, such testing is described as racist and clearly influenced by inherent white biases. Yet if IQ testing can be used to pretend that leftists are innately more intelligent than rightists, well then there are no biases affecting the results and the results must be unimpeachably accurate.
These studies are garbage data spitting out the results that are fed into them and intended by the authors of the study from the start. Even their conception of what defines rightism will have been framed in such a way as to lead to the conclusion they want, to be a leftists interpretation of what right wing attitudes are, rather than to be an objective and factual reflection of what right wing attitudes actually are. And the same will be applied in reverse, with a huge positive bias in the way in which leftist attitudes are defined.
Leftism does not parallel IQ and any attempt to conflate higher intelligence with leftism is laughable. Having been in academia and seen the utterly unthinking nature of many of the people thriving in that field, who NEVER question their own political prejudices, it’s my considered opinion based on experience that in many cases modern education probably lowers IQ (and empathy too).