CNN reported last week on a new study showing that liberalism, atheism and sexual exclusivity in males are linked to higher IQ scores. The findings are intriguing, for all the obvious reasons.
Evolutionary psychologist Satoshi Kanazawa at the the London School of Economics and Political Science correlated data on these behaviors with IQ from a large national U.S. sample and found that, on average, people who identified as liberal and atheist had higher IQs. This applied also to sexual exclusivity in men, but not in women. The findings will be published in the March 2010 issue of Social Psychology Quarterly.
Reactions have been all over the place, but there’s been strong suspicion of the findings from both “liberal” and “conservative” corners (especially conservative, as you’d expect). Which is good. These kinds of results may tell us something important, but we’re always advised to proceed cautiously and critically, especially when the findings of science are reported in the popular media. And double-dog especially when that popular media outlet is FOX or CNN. Understand – their criteria for reporting on research (there are thousands of studies published each month, and if you’re not an academic you hear about maybe three of them) have nothing to do with the social value of the research itself and everything to do with whether or not they think you might click on the link (and perhaps even on one of the ads on the page).
So, the critical reader should automatically pause and consider the following with respect to this story:
- Who is the researcher? What’s his expertise? Is he a pure academic or does he receive funding from sources with an axe to grind? Has his past research been unduly driven by concerns that appear, to the informed observer, to be more ideological than scientific? And so on.
- Is the story written by a reporter who understands science and research and statistics? (The answer here is usually no.) If not, then we need to find the actual study and see what it really says.
- Further, has the reporter bothered to ask him or herself any of the questions in that first bullet point? (Again, the answer is almost always no.) If not, what does it mean for the story (and the reader’s understanding of it) that the reporter can’t tell the difference between a Nobel laureate and a corporate PR hack?
- In this case, the story addresses IQ, but what does this really tell us? IQ is not a comprehensive measure of intelligence. It tells us some things (and these are important things) but it comes nowhere near telling us everything that we’d want to know when considering the “intelligence” of an individual or population.
- The definitions used here are beyond useless. “Conservative” and “liberal” are as artificial as labels come, for starters (the Political Compass test illustrates a small part of the problem), and when you add in the fact that the study probably relied on self-identification (hardly the most objective measure in the world) there is every reason to be cautious about the very way in which the two groups were constructed. What would it mean for the results if we learned that a good number of the liberals were gun owners or that a significant portion of the conservative group had serious misgivings about the Bush administration’s pro-torture activities?
This last point is crucial, because while self-report in studies like this tends to problematic under the best of circumstances, your margin for error explodes when the researchers and the participants don’t agree on the terminology.
The study takes the American view of liberal vs. conservative. It defines “liberal” in terms of concern for genetically nonrelated people and support for private resources that help those people. It does not look at other factors that play into American political beliefs, such as abortion, gun control and gay rights.
“Liberals are more likely to be concerned about total strangers; conservatives are likely to be concerned with people they associate with,” he said.
Now, is that what you think of when someone asks you if you’re conservative or liberal? Do a less educated and a more educated subject define those terms for themselves in the same way? Even if you explain what you mean by the term, do they each process it and respond the same way (after all, regardless of whether they’re conservative, liberal, libertarian, green or fascist, a less educated respondent is less likely to have the sophistication needed to parse a definition that’s not really like any they’ve encountered before).
Not to belabor the point, but we’re talking to Americans here, and we’re trying to exclude abortion, gun control and gay rights from how these respondents evaluate whether they’re conservative or liberal? Seriously? I’d argue that for huge portions of the population, abortion, gun control and gay rights are what the words liberal and conservative mean.
Hopefully by now it’s clear that I have significant reservations about the actual study and that I don’t trust the CNN story to get the story right, regardless of the actual findings of the study or the actual objective reality that the study may or may not have accurately described. As it turns out, my hesitation may be justified.
As I snooped around some other commentary on the study, I came across further reason for skepticism (interestingly enough, from an apparently “liberal” source that was linked by another liberal source). Dr. PZ Myers, a bio professor in the Minnesota system, stomps a mudhole in Kanazawa and walks it dry.
And then look at the source: Satoshi Kanazawa, the Fenimore Cooper of Sociobiology, the professional fantasist of Psychology Today. He’s like the poster boy for the stupidity and groundlessness of freakishly fact-free evolutionary psychology. Just ignore anything with Kanazawa’s name on it.
By all means, click on the links Myers embeds in that passage at his site, because he’s just getting warmed up. I don’t know much about Myers as a source himself, but he’s an academic, he’s a self-described agnostic and he links to the Richard Dawkins network (Dawkins being the Great Liberal Evolutionist Atheist Satan from Hell), so we might at least view his assault on Kanazawa as worth exploring, being as neither is exactly coming off as a conservative apologist.
So, to the question: are liberals smarter than conservatives (or vice versa)? Somewhere out there is an answer, and I for one would love to know what it is. I have my suspicions, based on my own experiences, but those suspicions are hardly science. If I’m right, I’d welcome the support of hard research, and if I’m wrong I’d like to know so I can reevaluate and get my opinions more in line with the facts. Hopefully you feel the same way.
In order to find that answer, though, we’re going to need a better study than Kanazawa’s (which seems horribly flawed, although I won’t know for sure just how much so until I see the actual study). Here’s what I think a more conclusive study would look like.
- For starters, it would need a more comprehensive measure of intelligence. IQ is a piece of the puzzle, but we’d also want to factor in creativity, associative thinking, critical thinking and problem solving. We’d like to be clear about the importance of memory vs. processing power in the equation, and before we get started we’ll want to decide whether to integrate newer concerns like “social intelligence” or whether social skills are better classified as something other than intelligence.
- We’ll want a much better handle on that whole conservative vs. liberal quagmire. Doing the study so as to render a verdict on those two categories is useless. We’d be better served by evaluating intelligence according to which political party people identify with, and even this would be problematic (what do you do with all those independents who are independent for wildly divergent reasons, for instance). I don’t have a satisfying frame in mind right now, but unless we can get to some meaningful definitions about political beliefs (definitions that make sense to the participants as well as the researchers) we’re wasting our time and money.
- It needs to be longitudinal and will ideally have mechanisms for evaluating how perspectives shift over time. More to the point, it would be important to know what factors shift those positions. Does education make you more X? If so, are there certain kinds of education that do so?
- It would be nice to know how these factors vary according to demographic variables. Are you more prone to the liberalizing effects of education if you’re working class from the South than if you’re middle class from the Upper Midwest?
- This study needs to be funded by a non-partisan entity of some sort and should be conducted by researchers with no particular ideological master. Under no circumstances should it receive funds from corporate sources. Whether there’s any actual biasing effect or not (and by the way, there is – research most often serves the interests of those writing the check), the value of such a study would be badly kneecapped by the appearance that its results were bought. It goes without saying that the study should be headed by a person or team with a track record that makes clear their commitment to academic rigor and uncompromising ethics.
- Methodologically, the study should employ both quantitative and qualitative instruments. You’ll obviously need the quant to generate a broad statistical basis, but this should be augmented by interview and observation phases to add depth and texture to the findings.
- For fun, it would be nice if there were an intercultural component. Is what we see happening in the US like what happens in other countries? If not, how are we different and what factors seem to account for the variance?
There are probably more issues we’d want to see addressed, but these represent at least a decent foundation for discussion. If we conduct such a study, and if it produces results similar to those reported by Kanazawa, then we’ll have something interesting to factor into our policy making.
One note, though. Let me call your attention to this passage from the CNN story:
The IQ differences, while statistically significant, are not stunning — on the order of 6 to 11 points — and the data should not be used to stereotype or make assumptions about people, experts say.
This is among the most ludicrous statements I’ve heard in some time. Assume we were to find that intelligence between two political groups varied by as much as 10 points, and assume that these findings were significant at the .95 level (and assume for the heck of it that the qualitative segments of the study supported the findings and provided richer insights into them) – you’re going to suggest that an overall intelligence difference of 10%, considered across a population of 300 million, isn’t stunning? I beg to differ. A variance of that magnitude would be positively staggering.
A difference of 10% between individuals is the difference between an A and a B, a B and a C. It’s the difference, in many cases, between the guy you want operating on your child and a guy you wouldn’t let anywhere near your child. In a financial advisor it could be the difference between comfort and borderline insolvency. If you’d like your teenager to go to the best school possible, it’s the difference between a highly ranked national university and a good, but not spectacular state system school.
What if half the population suddenly became 10% smarter? When you think about highly competitive business deals, for instance, deals where one company gets the contract by a hair’s breadth, would you take a 10% boost?
Make no mistake, the degree of difference we’re talking about here, even if it’s at the low end of the variance instead of the high end, is massively significant when we’re talking about the collective intelligence of a society the size of the United States.
In the end, I don’t know what, if anything, we really learn from Kanazawa’s study. But it’s an interesting question, and knowing the actual answer could do us a lot of good. It’s just a shame that we can’t count on our intrepid press to get the damned story right, if and when it ever happens.