Are conservatives less intelligent? Let’s ask a liberal…

Here is an interesting article from Discover Magazine, about some recent research into the association between intelligence and social attitudes. The study was conducted by some psychologists from Canada, and published in the prestigious journal Psychological Science. It represents a newly burgeoning tradition of investigating whether social conservatives (such as religious folks) are actually dumb. Or, in the more sophisticated examples, looking at exactly how dumb it is that they are.

However, the Discover Magazine article points out how such research can be interpreted in different ways depending on your social value system. The author argues that the Canadian study was reported from a somewhat liberal perspective.

image
A nice picture

For example, he raises a number of questions about how important research choices were informed by the investigators’ own political persuasions. He points to things like how the mediation effect is described (i.e., the way intervening factors are viewed as affecting the overall cause-effect storyline), and how the researchers defined and measured key variables, such as ‘conservatism’. These are excellent points.

However, a much simpler question can be asked about this type of finding: have a look at the diagram and ask yourself, why aren’t the arrows pointing in the opposite direction?

Never mess with the arrow people (via lolroflmao.com)

The data are cross-sectional, so the direction of the arrows is entirely the choice of the researchers. Conservatives may say that the interpretation offered by the researchers — that poor abstract reasoning abilities contributes to a person’s prejudice against minorities — reflects a liberal bias. They may feel the causality is shown the wrong way around.

Maybe people with these types of prejudices lead lives that deprive them of the opportunity to cultivate abstract reasoning abilities. For example, maybe they stay indoors or socialise only with people who agree with them, thereby avoiding the cognitive challenge of having to analyse and defend their own views.

As it happens, we know from other research that precisely this happens to just about every group of people in society. Smokers hang out with other smokers, computer geeks hang out with other computer geeks, and, well, homophobes hang out with other homophobes. All this leads to an outcome called the social false consensus effect, where everyone ends up overestimating the prevalence of their own beliefs, because they are never meaningfully exposed to contrary perspectives.

Studies have even shown that smokers actually numerically overestimate the numbers of smokers in the general population (which represents something of an attitudinal barrier to many anti-smoking campaigns). And non-smokers correspondingly under-estimate the prevalence of smoking (which can make some of them very self-righteous).

In other words, the diagram could reflect the effect of social prejudice on reasoning abilities, rather than the other way round. (Note that he researchers actually find — and even show in the diagram — that social contact with minorities plays a part in the statistical association).

More arrows, please

I’m guessing that liberal commentators may simply respond by saying that their interpretation — that poor reasoning abilities cause prejudice — is just more parsimonious (i.e., it makes more sense because it’s less complex). They may also point out that, whatever the direction of causality, the dataset confirms pretty clearly (a) that social prejudices are really nasty things and (b) that socially prejudiced people are not very clever.

But while that’s true, do we know what causes what here? Simply put, no we don’t.

Without collecting additional information that conclusively shows poor reasoning abilities to temporally precede the development of social prejudices, the interpretation of this type of cross-sectional research data will always be susceptible to the interpreters’ own personal biases.

And no combination of boxes, arrows, beta-weights, and asterisks can deal with that particular problem.

Although they might distract you from it.

[Adapted/expanded from my earlier post elsewhere]

Share this:

Leave a Reply