On June 13th Ronald Bailey wrote an article for reason.com entitled “Are Conservatives Dumber Than Liberals?” In the article Bailey overviews some social science studies that seem to “confirm” that liberals are more “intelligent” than conservatives, then he suggests that classically liberal people (i.e., libertarians- in his definitions) are actually more intelligent than democrats (or modern liberals). Though well intentioned as Mr. Bailey may be, I’m afraid all of these studies are, full of holes and provide no significant evidence in any direction.
First lets overview the studies that “prove” liberals are more intelligent than conservatives in Bailey’s own words.
A 2010 study using data from the National Longitudinal Study of Adolescent Health, for example, found that the IQs of young adults who described themselves as “very liberal” averaged 106.42, whereas the mean of those who identified as “very conservative” was 94.82.
For this proposition to be true a few components have to be demonstrated to be reliable. One, is IQ an appropriate test of “intelligence”? Two, do the scores found on IQ differ in a meaningful way that answers the research question of intelligence and political views? Three, is self-report reliable?
The answers to all of these questions are, unfortunately for Ronald Bailey – and the researchers who conducted the study, no. IQ tests do not measure intelligence per say, unless you define intelligence as having a history with situations that result in you picking the associations and problem solving methods used in each portion of the test. This is why IQ tests are well known to be culturally biased. It should also be noted that their use in the U.S. in education system was to determine which students had deficits in academics and standard Western style problem solving instruction. Not to mention that the test is measuring a construct, i.e., intelligence. According to the Oxford dictionaries a construct is “an idea or theory containing various conceptual elements, typically one considered to be subjective and not based on empirical evidence”. Constructs are not real entities and cannot be examined as if they are one, all of which makes the use of IQ to determine who is more intelligent a pretty ridiculous way of getting to what I believe is the implicit question of this article and studies associated with it, “do people with certain political views may only hold them because they don’t know any better?”
Two, even if IQ was a good way to go about answering this question of intelligence and political views, do the differing scores found on IQ answers this question in a meaningful way? The answer as stated earlier is no, but why? It is because of the way that IQ is scored. IQ is scored by a deviation method. A deviation method is a statistical method by which a range of scores plus or minus a certain, main score (in IQ it is 100- average intelligence) are considered to be the variability of getting score that appear to be different, but are actually equivalent, because of the natural variability of the measure. More plainly, average intelligence is considered 100 and its deviation that +/- are 15. That means that any score between 85 and 115 are essentially equivalent to getting a score of 100. Any variability between these numbers is simply the margin of error. So, when the 2010 study showed scores of approximately 95 and 106, they were both within the margin of error of the same score. This can be clarified by any basic statics book, especially one for the social sciences.
Third, is self-report reliable? Again I say no, so why not? It is commonly known that what people report (i.e., what they say) and what is observed about the person (i.e., what they do) does not line up. There can be many reasons for this, people may lie or be unaware of the truth. (One explanation for this may be found in B. F. Skinner’s suggestion that the variables that result in our actions may not be the same ones that result in what we say to others, but is another topic.) Regardless, the use of self-report harms the reliability of this study to even answer the question reliably.
Luckily, Bailey does not rely on only one study to make his point. Unfortunately for Bailey the other studies he cites don’t have any better data. First, the 2009 study he cites provides a score that is simply what person marked on the various questions and again this type of information does not necessarily say anything other than that – according to the measures used – the participants that fit the construct of conservatisms used for this study correlated (i.e., their answers point to an unknown relationship) with the construct for cognitive ability (i.e., the performance on the task given to participants, which don’t necessarily mean anything about cognitive ability in the typical decisions of everyday life). This holds all of the same flaws as the 2010 study discussed above.
As far as the 2012 study that Bailey presents – the study suggests that people provided more conservative answers to questions when they were drunk and/or have been performing difficult cognitive tests – it’s just irrelevant. If it were not irrelevant, then the inherent proposition of Bailey and the researchers would have to be that the best way to determine a people’s level of intelligence and political views is to impair them through performance of difficult task or by providing them with substances of abuse. I don’t believe Mr. Bailey is under that impression.
Next, the 2008 study describing agreement between SAT verbal scores and answers to social questions is rightly identified as a problem with the construct that the researcher was using, but then Bailey goes on to reformulate it along the lines that befit his own position. It is bad reasoning and bad science to get your results and then formulate the lines upon which you will judge them. You are working from your conclusion to find your question. Not to mention the fact that as a construct it doesn’t tell you anything about a real event, per say, at all.
Lastly, Ronald Bailey discusses a study in 2014 that discusses differences between libertarians (the construct group he seems to be in favor of), liberals, and conservatives on IQ. Needless to say, all differences are less than +/- 15 IQ points away, which you may remember is the margin of error for IQ scores, and are in fact only 2-5 point differences, which is completely meaningless. Just like all of the other studies, these rely on self-reports, IQ test, non-meaningful differences in scores, and constructs.
The point of this article, however, is not to shame Mr. Bailey in any way. I believe he presented this information with all good intention; well as good as looking to prove your own view right can be. In the age of easy access to information we must be more skeptical than ever of research, particularly those that confirm our own views.
So, are there intelligence differences between people of differing political views? Based on the research presented in Ronald Bailey’s article, no there isn’t. Which means the question I asked earlier was implicit in his article (as well as others like it), are people with certain political views may only hold them because they don’t know any better. No – or at least not any more than people of other political views.
Personally, I think this is a good thing to find out. Maybe it will lead people to ask why people really do hold views both similar and different from themselves. Then, maybe, just maybe, we could have real conversation when it comes to politics- even when we hold different views.
 Skinner, B. F. (1957). Verbal behavior. New York, NY: Appleton-Century-Crofts.