Satoshi Kanazawa is an evolutionary psychologist that currently blogs for Psychology Today in an attempt to give us “a look at the hard truths about human nature”.  And one of his recent hard-hitting stories is “Why Liberals Are More Intelligent Than Conservatives”. Here’s a snapshot of his evidence:

  • He defines Liberalism as altruism toward strangers, people who are not family or friends, by paying taxes to ensure their well-being. He refers to this altruism as “evolutionarily novel”.
  • People who identified as “very liberal” in early adulthood scored (an average of) 12 points higher on a childhood IQ test than those that identified as “very conservative”.
  • Look – he even has a graph:

He goes on to point out that conservatives accuse liberals of dominating the media and academia, and concludes they do – because they’re more intelligent.

Well, it’s hard to accept this blog posting at face value. The first criticism that comes to mind is Kanazawa’s assumption that liberal fiscal policy is based on altruism. Next, his complete disregard for the way changes in the size and structure of society have affected us all and the fact that it’s possible the liberal’s life is bettered by paying taxes to ensure the well being of lower income individuals via lower crime, etc. Third, the fact that an IQ test score is wholly based on comparative intelligence and is, by definition, an average. So any sufficiently large sample should test…average. But forget my criticisms – let’s consult someone who actually knows what he’s talking about.

In this article, Dr. Shawn Smith takes Kanazawa to task not for the data he used, but for the way he interpreted it.

Dr. Kanazawa used data obtained in an independent study of 20,745 people performed in three waves from adolescence to early adulthood. The study (look here,  here, and here) recorded the results of intelligence tests in adolescence and political affiliation in early adulthood. The first wave assessed intelligence using the Peabody Picture Vocabulary Test, and the last wave asked each subject to rate their political affiliation using the continuum in the above graph. It turned out that kids who scored higher on the intelligence test identified as liberals in early adulthood. Thus, a paper in a psychological journal and the blog post linked above.

Here’s the problem. Wait – here are the problems.

First, the IQ scores Kanazawa used are only an estimate. An estimate based on a verbal intelligence test that may have a decent correlation with the verbal portion of an intelligence test, but not with overall intelligence. There are two major components to most IQ tests: the verbal and the performance portions. So while the scores above may have a high correlation with each child’s verbal ability, they are hardly a picture of each child’s overall intelligence.

Second, and even more problematic, is that kids who score high on the verbal portion of intelligence tests tend to do better in school and continue to college, but may not be more intelligent overall. Equating stronger verbal ability with higher IQ scores skews the study toward the assumption that college students are more intelligent, when that is not necessarily true. (I’ll tell you why this is so problematic in a minute, if you haven’t already figured it out.) Although kids who score lower on the verbal portion than the performance portion of IQ tests tend to have inferior performance in school and not proceed to college, they are not necessarily less intelligent overall. The discrepancy between these two scores is of high interest to psychologists.

This chart shows, for example, that in the “normal” IQ score range of 90 – 109, 45.5% of subjects show a 9+ point discrepancy between their verbal score and their performance score.

Dr. Smith eloquently sums up this particular criticism, “By relying on a test that merely approximates VIQ [Verbal IQ], Kanazawa ignored a major component of intelligence and hopelessly tainted his data. Had he instead tested subjects’ ability to rebuild a carburetor, he would have arrived at a different (and equally flawed) pool of “more intelligent” people. Instead, his definition of “more intelligent” is biased toward those with relatively higher VIQs and away from people with relatively higher PIQs [Performance IQ].”

Last, because Kanazawa’s IQ score skewed the results toward college students as more intelligent, the political party those students selected would inevitably win the intelligence war. And college kids are almost never liberal. Oh wait, yeah they are. College is a time at which people are most likely to be liberal, and academia is an environment that has been objectively shown to be liberal. But to be fair, Dr. Smith consulted a Pew Research study showing that kids between the ages of 18 – 28 (the wave 3 age) identify pretty evenly as democrat or republican. As they get older, they tend to become more conservative. So, these kids were asked to identify their political affiliation at the time in their lives when they’re most likely to be liberal, especially college students (who were deemed to have higher overall intelligence based on a verbal test).

So, in short, the methodology and interpretation that led to Kanazawa’s conclusion that liberals are objectively more intelligent than conservatives is flawed. Dr. Smith even posits that psychologist have a persistent personal agenda to beat up conservatives, and he is determined to fight it. There’s so much good stuff in his article than I couldn’t summarize here, so I’m linking to it again because it’s really worth a read. It’s a real exercise in critical thinking and applied skepticism.

The point isn’t that liberals aren’t more intelligent than conservatives – maybe they are, maybe they aren’t – but this study hasn’t proven it.  The point is that if the evidence is bad, I have to ignore it – even when it tells me what I want to hear.

Stacey

Stacey

Previous post

Skepchick Quickies 9.16

Next post

Fear of Flying

30 Comments

  1. Profile photo of waltdakind
    September 16, 2010 at 10:22 am —

    As a liberal I declare this study right and your rebuttal wrong.
    For proof, refer to the chart which says I’m smarter. :)

  2. Profile photo of lydia-encyclopedia
    September 16, 2010 at 10:24 am —

    Didn’t Kanazawa use this exact same method to try and prove that Asians were the smartest and most “intuitively creative” people, and that Africans are in poverty not because of colonial abuse and exploitation, but due to their IQs being lower on average?
    I would quadruple-check his data if he just declared that the weather was nice today.

  3. Profile photo of Zapski
    September 16, 2010 at 10:55 am —

    @waltdakind: I second this argument, adding an ad populi dimension to this.

  4. Profile photo of waltdakind
    September 16, 2010 at 11:08 am —

    @Zapski: Plus, Kanazawa is a *doctor*! Just to throw in a little argumentum ad vercundiam.

  5. Profile photo of Lukas
    September 16, 2010 at 11:30 am —

    “Third, the fact that an IQ test score is wholly based on comparative intelligence and is, by definition, an average. So any sufficiently large sample should test…average.”

    Depends on what it is a large sample of. The average IQ of a sufficiently large sample of liberals must not necessarily be equal to the average IQ of the whole population.

    “So, these kids were asked to identify their political affiliation at the time in their lives when they’re most likely to be liberal”

    Wouldn’t that merely mean that there are more liberals at all intelligence levels, meaning that we could still draw valid comparisons between different intelligence levels?

    The only criticism that makes sense to me is that by using a test that approximates VIQ, only one aspect of intelligence is being measured, meaning that a possible explanation for the difference would be that liberals and conservatives tend to have different strengths.

  6. Profile photo of justinmckean
    September 16, 2010 at 11:32 am —

    Very nice demonstration of the critical thinking process.

    The most interesting item of information I’ve seen on the subject of the sources of liberal vs. conservative thought has been election maps. County level.

    You see red and blue state maps everywhere, but there are red and blue county maps. These show that the higher the population density of a given county, the more likely that county will be blue. The lower the population density the more likely the county will be red. This is true even in places like Texas where you’d expect the state as a whole to go red, but the more populous counties often go blue.

    For example:
    http://www.washingtonpost.com/wp-srv/politics/interactives/campaign08/election/uscounties.html

    My own state, Oklahoma, was the only state in the union to be entirely one color. Red. Like my face when I found that out. This has a lot to do with the fact that in a presidential election Oklahoma only allows voting for the Republican and Democratic candidate, no third parties or write-ins allowed.

    So, to put it another way, there seems to be at least a correlation between the size of an area’s population and the voting habits of the people there.

    Populations also tend to be more diverse in terms of everything from economics to ethnicity to gender roles and, well, everything else, when those populations are larger.

    This is my interpretation: the more likely one is to be exposed to someone who lives differently than they do, the more likely that one will be liberal.

    And the more likely it is that one will not be exposed to someone who lives differently than they do, the more likely one will be conservative.

    It’s a matter of education and life experience, not intelligence. This is why academics tend to be more liberal, too. In the academic world one tends to be exposed to far larger ranges of worldview than one would be outside that world. The broader education lends itself to that blue perspective.

    This is just a hypothesis based on statistical information and personal experience. I’ve done no actual studies on this issue, but I’d love to see them if they exist.

  7. Profile photo of Stacey
    September 16, 2010 at 11:54 am —

    @justinmckean:

    This is my interpretation: the more likely one is to be exposed to someone who lives differently than they do, the more likely that one will be liberal.

    And the more likely it is that one will not be exposed to someone who lives differently than they do, the more likely one will be conservative.

    It’s a matter of education and life experience, not intelligence

    That’s really interesting!

  8. Profile photo of infinitemonkey
    September 16, 2010 at 12:34 pm —

    @waltdakind: I can’t imagine any reason why this Dr. would want to come to a wrong conclusion, therefore, it MUST be right!

  9. Profile photo of James Fox
    September 16, 2010 at 12:38 pm —

    @Stacey: That’s often been my experience in meeting more rural folk. Also in a rural or perhaps a more agrarian setting, it could be that change is more often negative and consistent predictable patterns are desired and indicative of success. Definitely an interesting topic to speculate about.

  10. Profile photo of Stacey
    September 16, 2010 at 1:05 pm —

    @infinitemonkey: You guys are cracking me up. :)

  11. Profile photo of RodSkeptographer
    September 16, 2010 at 1:06 pm —

    You have said some things that are almost certainly false. E.g., “…may have a decent correlation with the verbal portion of an intelligence test, but not with overall intelligence.” If you have heard of the general factor of intelligence, you know that all intelligence tests correlate with each other to some degree. The question is how much. The Multiple Intelligences idea is appealing, but it is more popular than the data warrants. The general factor is large and the other intelligence factors are small and mostly overlap with it. So when you make the quoted statement, it is a bit of an extraordinary claim. You should have provided the correlation and cited a source.

    Plus you have to consider what happens when you use group statistics versus individual scores to make decisions. That a test may be inadequate as a tool for making detailed educational decisions for an individual child does not mean that it is worthless, especially for making group comparisons. Certainly, large numbers don’t fix bias problems, but that is another argument.

    To use an analogy, the argument about score discrepancies is like a global warming denialist bringing up huge daily temperature discrepancies to argue that you can’t say something more precise like that the earth has warmed say, 1.5 degrees in x years.

    It is very liberal of you to consider the arguments of the other side, but it helps to know about the subject matter to not swallow the bad arguments along with the possible good.

  12. Profile photo of Smashley
    September 16, 2010 at 1:32 pm —

    @justinmckean: This is really cool logic. I’ve never thought of it that way before. That really clearly explains why college kids are more often liberal.

    On another note, it looks like Atheists are still smarter! http://blog.okcupid.com/index.php/the-real-stuff-white-people-like/ (scroll down to the 2nd graph from the bottom, though this whole article [and the whole site] is fascinating!)

  13. Profile photo of Draconius
    September 16, 2010 at 1:33 pm —

    Hi there!

    I always like to warn people that I am a bleeding-heart tree-huggin’ peacenik hippie Liberal. But I still can’t take this study at face value. :(

    Someone once told me that the problem with Liberals is that we don’t have the courage of our own convictions. This is because our greatest weakness seems to be: “Seeing the other guy’s point of view”. This seems to inform the entire Liberal platform. We support welfare, because we’d hate it if WE were poor. We vote pro-choice, because we’d hate if WE were a pregnant teen with no other options. We support gun control, because although we personally wouldn’t own a gun, we’d hate to be shot with one. We extend the Olive Branch to other countries, because we fail to see them as a big scary monolithic group of monsters who think alike. (we reserve that opinion for corporations) This makes it difficult to argue with conservatives. We keep trying to see THEIR side of the story. We feel like, if we could just sit down and RATIONALIZE with conservatives, they’d see things from our side.

    So yes, on one hand, I could see how this could parse as “smarter” on certain tests and examinations, as @justinmckean has said, above. Liberals are usually exposed to a multitude of ideas, and therefore see other points of view. We can never seem to resolve an issue without THINKING at it from every which angle.

    But it really doesn’t make us SMARTER. At least, I don’t think it does.

    Or am I just saying that because I’d hate it if I saw a study saying that *I* was less intelligent, and I’m trying to see the Conservative point of view?

    Gah.

    I’m really bad at Maths, that counts for something, right? :(

    — Craig

  14. Profile photo of Stacey
    September 16, 2010 at 2:06 pm —

    @RodSkeptographer: So, just to clarify, you believe that Dr. Kanazawa’s data interpretation was sufficient to draw the conclusion that liberals are more intelligent than conservatives? Specifically, that scores on the Peabody Picture Vocabulary Test are adequate indicators of IQ because there’s no significant documented difference between verbal & performance IQ (i.e. the chart I provided is incorrect) , and although the PPVT appears to be an inadequate tool to assess an individual’s IQ, it is sufficient to assess the IQ of a group?

    To respond to your request for correlation between the PPVT and general IQ test, the former has a .91 correlation with the verbal portion of the latter and the link to that information can be found in the main article I linked to.

    Here’s the quote so you don’t have to find it:

    According to the publisher of the PPVT, the current version boasts a .91 correlation with the Verbal Intelligence Quotient (VIQ) of the WISC-III. Translation: a kid’s score on the quick and dirty PPVT will probably be close to the score that child would obtain on one portion of the Wechsler Intelligence Scale for Children (WISC).

    And the link, in case you want to find it.

  15. Profile photo of thad
    September 16, 2010 at 2:56 pm —

    Once again I feel inclined to chime in “IQ is rubbish!”, and point to Stephen J Gould’s “Mismeasure of Man” for a great book on the subject.

  16. Profile photo of Stacey
    September 16, 2010 at 3:24 pm —

    @thad: That’s a point worth considering too. IQ tests definitely have their limitations. I don’t think they’re worthless, but one should definitely be aware of the limitations when using them in a study.

  17. Profile photo of RodSkeptographer
    September 16, 2010 at 3:38 pm —

    Stacey, I did not make a statement with regard to Dr. Kanazawa’s data interpretation. I made statements about yours. The chart you provided is likely correct, just as global warming denial data on daily temperature fluctuations may be correct.

    I thank you for trying to provide the quotes so as to not waste my time, but what you quoted did not give the correlation at issue here: the correlation between the PPVT and the WISC full IQ. “Correlations of the Peabody Picture Vocabulary Test-III form A and B scores with scores of the WISC-III VIQ are 0.91 and 0.92; …Slightly lower correlations with the non-verbal scores were predicted; with the WISC-III PIQ 0.82 and 0.84; … Full scale IQ on the WISC-III was 0.90 and 0.90.” (http://www.kids-iq-tests.com/Peabody-Picture.html). I did not check to see if these were the same versions of the tests used, but typically revisions of IQ tests are quite conservative.

    I did not say, “PPVT appears to be an inadequate tool to assess an individual’s IQ”. I was making another point. And actually, looking at the above numbers, the PPVT is actually quite good.

  18. Profile photo of RodSkeptographer
    September 16, 2010 at 3:45 pm —

    Stephen J. Gould’s book “Mismeasure of Man” goes against the consensus of the experts in the field. Most journalism does also.

  19. Profile photo of RodSkeptographer
    September 16, 2010 at 3:46 pm —

    @thad: Stephen J. Gould’s book goes against the consensus of the experts in the field.

  20. Profile photo of waltdakind
    September 16, 2010 at 4:23 pm —

    The discrediting of this study is part of a vast right wing conspiracy.

  21. Profile photo of waltdakind
    September 16, 2010 at 4:23 pm —

    And for good measure, I’d like to add that any time one encounters a so-called liberal who isn’t smart, it’s because he or she isn’t a real liberal.

  22. Profile photo of T-Storm
    September 16, 2010 at 5:56 pm —

    I think part of the difference is that liberals tend to see a bigger picture while conservatives are more concerned with the home front.

    What I noticed from the county map referenced above that the vast majority of the counties are red, not blue.

  23. Profile photo of mrmisconception
    September 16, 2010 at 6:28 pm —

    As a liberal I must say while I was reading this I believed it to be one of the biggest piles of horseshit I have seen in a long time.

    Then again, maybe that proves that I AM more intelligent.

  24. Profile photo of infinitemonkey
    September 16, 2010 at 6:36 pm —

    I blogged about this on my blog, but I hope I did it justice.

    dullestriangles.wordpress.com

  25. Profile photo of mrmisconception
    September 16, 2010 at 6:40 pm —

    I think I located the problem with Mr. Kanazawa’s article. It’s right there in the first sentence of your story.

    Satoshi Kanazawa is an evolutionary psychologist that currently blogs for Psychology Today in an attempt to give us “a look at the hard truths about human nature”

    Do you see it?
    Look at the fifth and sixth words.
    Yeah, evolutionary psychologist.

    Grain of salt need, kosher preferred.

  26. Profile photo of Buzz Parsec
    September 16, 2010 at 6:53 pm —

    @mrmisconception: I want to give EP a fair shot, I really do…

  27. Profile photo of mrmisconception
    September 16, 2010 at 6:58 pm —

    @Buzz Parsec: Oh, I do too.
    But like theoretical physics anything goes, and it usually does.

  28. Profile photo of Stacey
    September 16, 2010 at 8:36 pm —

    @mrmisconception:

    As a liberal I must say while I was reading this I believed it to be one of the biggest piles of horseshit I have seen in a long time.

    Then again, maybe that proves that I AM more intelligent.

    Oh the irony!

    @mrmisconception: I agree with you about evolutionary psychology, but I do have to give Kanazawa credit for at least using data to back up his assertion instead of just offering speculation, even if I disagree with the way he interpreted that data.

    @RodSkeptographer: Thanks for the clarification – I really don’t want to get into a back & forth with you – clearly we’ll just have to agree to disagree – but I want to address a couple things:

    No, the PPVT III Forms A & B were not the tests administered. The test was given in the mid 1990s and that revision wasn’t issued until 1997.

    The PPVT is an 11-12 minute test in which vocabulary is tested with pictures. It is mainly used to test disabled kids who lack the capability to take a full-scale intelligence test.

    Here’s a list of things the PPVT tests:

    Vocabulary

    Here’s a list of things the WISC tests:

    Verbal Comprehension:
    Vocabulary
    Similarities
    Comprehension
    Information
    Word Reasoning

    Perceptual Reasoning:
    Block Design
    Picture Concepts
    Matrix Reasoning
    Picture Completion

    Working Memory:
    Digit Span
    Letter-Numbering Sequence
    Arithmetic

    Processing Speed:
    Coding
    Symbol Search
    Cancellation

    The PPVT is not comparable to the WISC. It’s not a similar test, and it’s not used for similar purposes. Any comparisons should be interpreted in the context of what each test is.

    Another problem – even full-scale IQ tests given at a young age aren’t reliable estimates of IQ scores later in life. Much less a 12 minute vocabulary test given prior to maturity.

    Another problem – even the .91 correlation to VIQ isn’t “quite good” enough to truly predict VIQ. It’s just a quick and dirty estimate. So, even at best, if this study had been administered several times throughout the longitudinal study instead of once in early childhood, it could possibly be a quick and dirty estimate of IQ. It was not, and it is not.

    So you have a double whammy – he’s using a vocabulary test to get at best a quick & dirty estimate IQ, and he’s estimating the IQ of a mature person based on one test administered in childhood.

    But even that point is ultimately moot. As Thad pointed out, the usefulness of IQ tests is often criticized. They do have worth, and that worth lies mainly in predicting academic achievement. So even if the PPVT had a 100% correlation to IQ, or Kanazawa had access to actual IQ scores, the data would be just as skewed toward college students who would, again, get to choose which political party is smarter. The fact that Kanazawa based his results on the PPVT makes his assertion especially egregious.

    It always amazes me when these studies come out (like the one that “proved” men are smarter than women using Ravens Progressive Matrices) pawning off these inadequate tests as equivalent to IQ. If Kanazawa truly wanted to correlate IQ and political affiliation, he should have taken a large, randomized sample of each party and given them IQ tests – at the very least.

    In regard to the criticism of the chart showing the difference between verbal and performance scores in full-scale IQ tests – you lost me. I totally get what you’re saying about global warming; I don’t understand how that applies to the chart. If using a large sample of verbal and performance IQ scores isn’t the best way to analyze the difference between verbal and performance IQ scores, I’m not sure what is. I thought I had articulated your thoughts when I said you were critical of the chart because it was based on a large sample (group), which may not hold true for every individual. If that’s it, I think that’s true of most/all scientific studies. A large sample is used to find out what’s true for people overall, but that specific result can’t be reliably predicted for every individual. In your opinion, what is a superior method of testing for this difference? You might have to explain that one to me – just straight up, no involvement of global warming necessary.

  29. Profile photo of Buzz Parsec
    September 16, 2010 at 9:50 pm —

    Since liberals are smarter than conservatives, any valid scientific study will show this. Since this study came to this conclusion, it must be valid. Therefore, liberals are smarter than conservatives.

    I know this is true, because I channeled the ghost of Carl Sagan and he told me so.

  30. Profile photo of TheSkepticalMale
    September 17, 2010 at 4:49 pm —

    In this particular political climate, whatever intelligence there is seems to get drown out by the rampant confirmation bias in the electorate. When more and more people are tuning into Fox from the right and MSNBC from the left for “news”, does it really matter which population is ever so slightly more intelligent?

    Instead of coming up with a third or fourth potential solution for – oh, let’s just say, rising health care costs – it seems like the political discourse has broken down to nothing more than a team sport – dogma and demagoguery. (“If the other side is for it, then I am against it!”) I am at a point where I wish they would make it easier on all of us and just have all the politicians wear red or blue jerseys, festooned with patches to show their biggest donors (you know, like NASCAR racers).

Add Comment Register



Leave a reply