‘Are Men Smarter Than Women?‘ The Verdict

By Sheril Kirshenbaum | May 28, 2009 12:04 pm

On Tuesday I composed this letter to my blog BFF Isis, forwarding her a troubling inquiry that hit my inbox on Memorial Day:

Can I ask, from your perspective, what you think of this study suggesting that men are smarter than women? 

First, thanks to readers for so many terrific responses (I especially enjoyed this comment from Zen Faulkes). And today Sb’s resident Goddess has provided her take on Rushton’s so-called ‘study‘.  Here’s an excerpt:

You see, here is the problem with any study that attempts to investigate the innate intellectual differences between men and women — like some of the papers I reviewed this week, any study that could possibly be done would lack the appropriate experimental control.  Comparing men and women, even of you controlled for things like, age, cranial circumference, years of education, waist-to-hip ratio, etc. is still like comparing apples and oranges.  This is because boys and girls are treated differently as they develop.  Comparing these two groups in the context of today’s society would be like comparing the response of two strains of rat, who had been housed in different conditions, fed different diets, given different amounts of exercise and unique stimulation, to a particular mediator and then concluding that differences in response are related only to the strain difference.  It’s lunacy I’ll tell you.
Isiss Stamp.jpg

Figure 1: Isis’s editorial decision for John Philippe Rushton’s manuscript

Read Isis’ full and most excellent letter and make sure to continue on over to ScienceWomen for more…

CATEGORIZED UNDER: Culture, Media and Science
MORE ABOUT: email, gender, intelligence

Comments (24)

  1. Comment crossposted from Isis’ blog:

    Your critique is focused on the inference that the measured difference is innate — you say that lots of things about the ways that boys and girls are raised could result in a difference in intelligence test scores. But that concedes that the observed difference in test scores is real, and just disputes where the difference comes from. I think you’re giving him too much.

    A pretty substantial body of previous research has found no difference in measured general intelligence between men and women. If I was reviewing this paper, I’d start by asking whether the finding is even real — not disputing its interpretation.

  2. Blogger

    >>This is because boys and girls are treated differently as they develop.<<

    Is that wrong or right?

  3. Erasmussimo

    There is no indication that this paper has been published in any refereed journal. If it has not been so published, why should anybody take it serious regardless of its content? I could write a paper demonstrating that little green men are abducting people for hideous experiments. So what? Until it’s published in a refereed journal, it’s just one person’s opinion and devoid of any credibility.

  4. By that token, is any such comparison a controlled experiment? And Erasmussimo, I sympathize; I have long tried to convince others of the little green men story. Such is the burden of genius as you well know.

  5. A random passing physicist

    There’s no reason to take it seriously even *after* it is published in a refereed journal. That’s just argument from authority, and a poor authority at that. *Lots* of peer-reviewed papers have turned out to be wrong. The only relevant question is whether the methods and evidence stand up.

    In this case, they don’t. The differences in upbringing is a valid problem, but one that feminists tend to overemphasise. Attempts to bring up children neutrally have not been entirely successful. There are a few innate differences.

    The main problem with the hypothesis, though, is that the question is meaningless. IQ is not defined objectively, but by putting together a bunch of questions, measuring how well the population can answer them, and then trimming the questions and scoring system so you get a pre-defined statistical distribution. The metric is calibrated and weighted so that men and women score equally. The sexes have equal IQs by definition – the best one could hope to find is that the particular IQ test you are using is miscalibrated.

    One could then try to ask the question about ‘general intelligence’, but that doesn’t get one much further because intelligence isn’t one-dimensional. Some people are good at one thing, others are good at another. There is no ‘general intelligence’, only lots of different ‘specific intelligences’.

    But you can answer the question in a certain sense by saying that *both* sexes are innately more intelligent than the other, if you base your measure on the sort of problems people are naturally required to solve. The innate differences between the sexes are evolutionary adaptations to different environments. If you ask which sex is better at being a woman, women are. And men are obviously better at being men. By the only metric that matters, the evolutionary one, each sex is better than the other at what it does. Each has to be measured according to its own definition.

    As for any specific individual on any specific sort of problem, like that implied by the original questioner, the only questions you need to ask is do *you* have it, and do *you* want to use it? Sophie Germain was a better mathematician than the majority of men could ever hope to be, and not even physical force could stop her doing it. My absolute favourite mathematical physicist of the 20th century was Amalie Emmy Noether, and not because she happened to be a woman. She was a genius probably greater than Einstein, the work she did was simply mindblowing. There are plenty of other examples too. Sex is clearly no disqualification, but equally clearly, it isn’t simply a matter of choice.

    Forget about whether whether ‘girls do science’ – do you *like* science? do you want to spend your life doing it? Are you actually *any good* at it?

    If the answers are ‘yes’, you’re surely not going to let what a bunch of *men* think stop you, are you? And if they’re not, you’re not letting the side down by doing something else.

    If few girls thereby choose science, making decisions that are right for *them*, so be it.

    The same goes for boys too, or for any other subject.

  6. Passing physicist, I agree with the sentiment of your last few paragraphs — this debate should have no bearing on a girl’s decision to pursue science. That is the most important point.

    But… you’ve got a few things wrong about intelligence and IQ tests. In standard research practice, test scores are not trimmed or weighted in order to change the shape of the distribution. Nor are scores normed and weighted to make men and women score the same.

    The theory of general intelligence is based on the robust empirical finding that tests of different mental abilities (verbal, spatial, etc.) are positively correlated. The most widely accepted interpretation of this intercorrelation is that there is a single dimension of mental ability (general intelligence) that is necessary but not sufficient for performance in all task domains (“not sufficient” because most tasks also require domain-specific ability or experience). There are serious alternatives to this theory, but they still have to account for the positive correlation. Your simple statement that “there is no ‘general intelligence’, only… ‘specific intelligences'” does not do so.

    I’d recommend taking a look at this consensus statement. It was put together by a task force of researchers who are frequently on different sides of various debates about intelligence:

  7. honesty

    It seems intelligence knows no gender. In fact, to define the word “smart” is very subjective.

    Are you smart because you can answer every question on Jeopardy!?

    Are you smart because you know every little detail and fact to a certain area of study?

    It also depends on the circumstance. Read thru http://raysweb.net/poems/articles/tannen.html which is written by Deborah Tannen. She talks about different aspects of a relationship where both the man and the women are “right” and “wrong” because they have different ways of handling the decision making process.

    There are plenty of doctors who can tell you the purpose of every molecule in your body yet they can’t quite figure out their new computer. Meanwhile a 12 year old can take apart and rebuild a computer with their eyes closed yet not even know our bodies are made up of molecules.

    So really, the answer is: “Who cares?”

    Maybe if we stopped caring about which gender is “better” or “smarter” than the other, then maybe, just maybe, we can have some REAL equality via the “not caring” model.

    Just a little info about myself:
    I hate everyone equally.


    You’re bloody straight they’re smarter!

  9. A random passing physicist


    Thanks very much for that!

    There are some excellent quotes I can take from your document. What do you think…?

    “By convention, overall intelligence tests are usually converted to a scale in which the mean is 100 and the standard deviation 15.”

    “Most standard tests of intelligence have been constructed so that there are no overall score differences between females and males.”

    “While some tasks show no sex differences, there are others where small differences appear and a few where they are large and consistent.”


    Yes, I agree with your point that the different varieties of intelligence are correlated to some degree. For any distribution you can nearly always apply principal components analysis and extract such a ‘common’ measure, and it isn’t terribly surprising that the first principal component here is comparatively strong and positive. It simply says that there are factors that contribute strongly to many different sorts of intelligence at once (besides sex). But there’s a lot of variation left in all the other components; enough so that that’s where this particular question is left undecided.

  10. RPP –

    Regarding your first quote: Scaling a test doesn’t change the shape of the distribution or force it to be normal. Take any variable x from any dataset you have lying around (for best results pick something really skewed and funky looking). Plot the distribution. Now create the following new variable:

    IQx = 100 + ( 15*(x – mean(x)) / sd(x) )

    Compute the mean and SD of IQx. Gee whiz, it’s 100 and 15. Now plot the distribution of IQx. Hold it up next to the plot of x. Look familiar?

    Regarding the second quote: notice that I said “in standard research practice.” In applied settings you can norm a test however you like. Take a more obvious example: In applied settings most intelligence tests are age-normed. But if you were a researcher studying whether math ability changes from age 5 to age 10, and you based your comparison on age-normed math test scores, you would reach the improbable conclusion that 5-year-olds are just as good at math as 10-year-olds. You would also be roundly laughed at. Any researcher studying gender differences is not going to compare gender-normed scores. When I stated that previous research has found no sex differences in overall intelligence, the researchers I was talking about were not idiots.

    Regarding your third quote: What’s your point? General intelligence theory states that performance on a given task reflects general intelligence (shared with other tasks) plus specific ability (shared only with that particular kind of tasks). The consensus statement says that there are some sex differences on some tasks. That is entirely compatible with the idea that there is such a thing as general intelligence, that there are no sex differences in g, and there are some sex differences in specific abilities. (The origins of those specific differences could be from experience, they don’t have to be innate.)

    Now, a quote of yours: “… the first principal component here is comparatively strong and positive. It simply says that there are *factors* that contribute strongly to many different sorts of intelligence at once.” (emphasis added) No, it doesn’t say that. A model with one principal component posits one common factor. If you want to dispute the theory, you have to produce a better-fitting model in which a single factor does not affect all different tasks.

  11. A random passing physicist


    I agree that scaling on its own doesn’t change the distribution. However, I’m pretty sure they do more than simply scaling when designing these tests. Smoothness is also a desirable property. I don’t think I mentioned Normality at all.

    I’ll try a simple example. Say I have three different types of intelligence, or IQ test, or whatever, and these are affected linearly by three different factors: sex, wealth, and other genetic factors as follows: s*(1,8,5)+w*(25,45,15) + g*(10,4,2). If you’ve got a fair spread of wealth, the results are going to be strongly correlated, and the common factor is going to weight the types fairly close to (25,45,15). Measuring the component of the vector in this direction is going to give you your common factor, and it’s going to look like sex is correlated with it (because of the strong second component), but it isn’t so well correlated with the other innate factor. What if this is largely the innate “general” intelligence? And if wealth isn’t so widely spread, you’ll get a different result. Sex and general intelligence may both be buried among the other factors.

    All you’re measuring is the effects of the strongest contributors (that you haven’t controlled for). There’s nothing to say that this is in any sense the *right* measure of intelligence, or that it is due to some innate general intelligence function. It might be, but correlation on its own doesn’t say so. Given the number and strength of ‘nurture’ contributors I tend to doubt it, but I don’t claim to be an expert.

    You need some functional reason for thinking the principal component measures general intelligence, you can’t deduce that from the statistics alone.

  12. WhatMeWorry

    It’s pretty obvious to me, that the smartest responses to this post have been written by men.

  13. RPP –

    You seem to think I think some things that I don’t think, at least based on how you’re arguing. In your example, you are trying to show how we might observe sex diffs in test scores that we would mistake for sex diffs in general intelligence. I’m saying, we haven’t made that mistake. I’m vehemently disagreeing with Rushton. (See the top of the thread.)

    Heck, I’m not even necessarily claiming that there is such a thing as general intelligence. I’m just saying that if you’re going to dispute the theory and claim that there are *only* specific intelligences (your original point), you have to explain why different mental tasks are correlated.

    You constructed an example in which the causal influence of wealth dwarfs genetics and everything else. Problem 1 is that that doesn’t square with a big pile of data that shows a substantial non-sex-linked genetic contribution to test scores. Problem 2 is that you are treating wealth and genetics as different things, and you haven’t accounted for the relationship between them (the income of the home you grew up in is probably correlated with your parents’ intelligence, which is affected by genes that you share with them). Problem 3 is that you seem to think that if wealth strongly affects all 3 tests, that disputes general intelligence theory. It doesn’t. “General intelligence” does not mean “genetically determined” (you seem to be conflating those concepts) — there can be a general intelligence factor that is substantially affected by nurture. So if I saw the (highly unrealistic) pattern of data you’re proposing, where 1 easily measurable variable (wealth) has a huge impact on 3 different tests, my next question would be: is wealth affecting these 3 different tests in 3 different ways, or is it is affecting a single underlying mechanism that they have in common? So I’d start generating hypotheses and testing them experimentally. (And you’re right that we cannot deduce that answer from the pattern of correlations among tests. Intelligence researchers do a lot more than factor analyses.)

  14. Scott

    Even if it were true that men are smarter than women or women smarter than men, it shouldn’t change how we treat people. Everyone should have the chance to use whatever talents they have to the best of their ability.

  15. Dark Tent

    Sanjay says:

    “The theory of general intelligence is based on the robust empirical finding that tests of different mental abilities (verbal, spatial, etc.) are positively correlated. The most widely accepted interpretation of this intercorrelation is that there is a single dimension of mental ability (general intelligence) that is necessary but not sufficient for performance in all task domains

    The fact that an interpretation is “widely accepted” (whatever that means) does not mean it is right. Existence o f positive correlations between different IQ tests is not proof of the existence of a “general intelligence”, but it is at least possible that some of those who “accept” g as “real” are doing so because they are actually under the misapprehension that the correlations do constitute such proof.

    My purpose here is not to get into a debate about the reality of g, but merely to point out that the argument that something is “widely’ accepted is really irrelevant, particularly when it is not clear that g is not widely accepted precisely because of the correlations between tests.

    According to statistician Cosma Shalizi, general intelligence “g” is a Statistical Myth

    “the case for g rests on a statistical technique, factor analysis, which works solely on correlations between tests. Factor analysis is handy for summarizing data, but can’t tell us where the correlations came from; it always says that there is a general factor whenever there are only positive correlations. [emphasis added by me] The appearance of g is a trivial reflection of that correlation structure. A clear example, known since 1916, shows that factor analysis can give the appearance of a general factor when there are actually many thousands of completely independent and equally strong causes at work. Heritability doesn’t distinguish these alternatives either. Exploratory factor analysis being no good at discovering causal structure, it provides no support for the reality of g. ”


    And actually, even the summary statement you linked to above admits that correlation does not prove the reality of g.

    One way to represent this
    structure is in terms of a hierarchical arrangement with
    a general intelligence factor at the apex and various more
    specialized abilities arrayed below it. Such a summary
    merely acknowledges that performance levels on different
    tests are correlated; it is consistent with, but does not
    prove, the hypothesis that a common factor such as g
    underlies those correlations. Different specialized abilities
    might also be correlated for other reasons, such as the
    effects of education. Thus while the g-based factor hierarchy
    is the most widely accepted current view of the
    structure of abilities, some theorists regard it as misleading
    (Ceci, 1990).

  16. “Are Men Smarter Than Women?”

    I would say that the fact that the answer isn’t obvious is evidence enough that the answer is no.

  17. MadScientist


    The answer is obvious: there is no way to make a valid general comparison so the question itself is absolutely meaningless. How do you answer these types of equally meaningless questions which focus on actual individuals:

    1. Who was smarter, Albert Einstein or Richard Feynman?
    2. Who was smarter, Plato or Aristotle?
    3. Who was smarter, Leonhard Euler or Max Planck?

    Anyone who would believe that some sort of intelligence test could actually determine conclusively that Albert Einstein was more (or less, or equally) intelligent than Richard Feynman is indeed a fool. I never took these IQ tests seriously; even as a teenager I could tell that they were pretty damned stupid so I deliberately tried to get all answers wrong. The only use I can see for any IQ test (at least if answered honestly) is to say “person X is far poorer (or better) than the general population at activity Y” but anyone who believes that all categories tested can simply be lumped together or that small differences in scores are of any significance doesn’t have a clue what they’re doing.

  18. chezjake

    While we are discrediting Rushton’s thinking on gender differences, let’s not forget that this man has been twice disciplined by his own university for unethical treatment of both his students and research subjects and has also been widely criticized for years as a racist. He’s the director of the Pioneer Fund, which is listed by the Southern Poverty Law Center as a hate group.

    I don’t see any reason for taking anything Rushton has to say seriously.

  19. anonymous

    Actually, the question is meaningless. IQ tests are meaningless. My fifteen year old scores “very superior” across the board but he is unlikely to pass tenth grade this year. Where is the CQ (creativity quotient) test? Intelligence in no way predicts creativity and creativity is necessary for discovery/achievement in every field. I also don’t think motivation is measurable. Plenty of people with average intelligence but high motivation are high achievers and plenty of geniuses never achieve a thing.

    Remember the book published several years ago Failing at Fairness (Myra Sadker)? Her analysis explains why test scores might look so different (gendered), at least in this country.

    One has to ask, “What are IQ tests good for?” If the answer has anything to do with creating opportunities/allocating resources to the kids with the most “potential,” then we are making a big mistake. I say we put the “dumb” kids in the “gifted and talented” tracks at school and see what happens.

  20. WEUer

    That was an utterly ridiculous study. I don’t think it can even be classed as “scientific”.

    “Oh, let’s just get some random 17-18 year olds to do this random IQ test and compare the results. Oh, there are more female surveyees? Oh well! Hm… it seems I’ve forgotten to take into account all those environmental factors that influence a child’s mental development. Oh well; I’m sure nobody will notice! I can’t really compare people of different body size, background, sleep patterns, etc, and, according to myself and no-one else, race? Meh; that’s just a small thing.”

    And that study he carried out previously that “proved” intelligence was influenced by race? That was… (struggles to find the words)!!!! Professor Rushton has obviously no understanding nor interest in Economics. He definitely has also never read _Freakonomics_ by Steven Levitt and Stephen Dubner.

  21. Superstringy Indian

    Whoa folks.Calm down.
    I thought men and women were pretty much equal.Now,I am not so sure.
    For anybody doubting IQ tests’ validity:
    I live in India.My age is 13,and I have an IQ of 135.How many 13-year olds read the Origin of Species,The blind watchmaker,or Das Capital?
    My older brother has an IQ of 142.He came 60th in the IIT-JEE,among 3 lakh people.That’s a 1 in 5000 rank.
    My younger sister has an IQ of 120.She does quite well in her grades.
    My younger brother has an IQ of 101.He is average in his intelligence,as far as I can tell.
    On socialization:
    It is true that boys and girls are treated differently.It is also true that this differential treatment is solidly grounded in biology:See David Reimer.In other words,a la Dawkins,the sexes come upon different memes purely because of genetics.Although it appears as socialization,it really is not:it is a genetic byproduct.As humans,we use fast-scale memetic evolution and hence the arrangement.As an example take Mozart:would Leopold have given him the same training if Mozart was a total asshole in music?No,of course not,and this is not socialization,just exploiting strengths.
    A reply to Sanjay:
    A pretty substantial body of previous research has found no difference in measured general intelligence between men and women. If I was reviewing this paper, I’d start by asking whether the finding is even real — not disputing its interpretation.
    Sorry Sanjay,but the Raven’s progressive matrices,which happen to be the best measure of g factor we have show a consistent male advantage.In early years it is masked because of earlier female maturity and this is what Richard Lynn suggests led to the well known similar IQs.

  22. Zach


    Yeah, and thanks for contributing.

    @SuperStringy Indian

    For anybody that takes IQ tests seriously:
    This 13 year old has an IQ of 135, reads Das Kaptial, The Blind Watchmaker, and The Origin of Species, yet does not know enough to put spaces after commas or periods.

    Yeah, I’m pretty much sold.


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

About Sheril Kirshenbaum

Sheril Kirshenbaum is a research scientist with the Webber Energy Group at the University of Texas at Austin's Center for International Energy and Environmental Policy where she works on projects to enhance public understanding of energy issues as they relate to food, oceans, and culture. She is involved in conservation initiatives across levels of government, working to improve communication between scientists, policymakers, and the public. Sheril is the author of The Science of Kissing, which explores one of humanity's fondest pastimes. She also co-authored Unscientific America: How Scientific Illiteracy Threatens Our Future with Chris Mooney, chosen by Library Journal as one of the Best Sci-Tech Books of 2009 and named by President Obama's science advisor John Holdren as his top recommended read. Sheril contributes to popular publications including Newsweek, The Washington Post, Discover Magazine, and The Nation, frequently covering topics that bridge science and society from climate change to genetically modified foods. Her writing is featured in the anthology The Best American Science Writing 2010. In 2006 Sheril served as a legislative Knauss science fellow on Capitol Hill with Senator Bill Nelson (D-FL) where she was involved in energy, climate, and ocean policy. She also has experience working on pop radio and her work has been published in Science, Fisheries Bulletin, Oecologia, and Issues in Science and Technology. In 2007, she helped to found Science Debate; an initiative encouraging candidates to debate science research and innovation issues on the campaign trail. Previously, Sheril was a research associate at Duke University's Nicholas School of the Environment and has served as a Fellow with the Center for Biodiversity and Conservation at the American Museum of Natural History and as a Howard Hughes Research Fellow. She has contributed reports to The Nature Conservancy and provided assistance on international protected area projects. Sheril serves as a science advisor to NPR's Science Friday and its nonprofit partner, Science Friday Initiative. She also serves on the program committee for the annual meeting of the American Association for the Advancement of Science (AAAS). She speaks regularly around the country to audiences at universities, federal agencies, and museums and has been a guest on such programs as The Today Show and The Daily Rundown on MSNBC. Sheril is a graduate of Tufts University and holds two masters of science degrees in marine biology and marine policy from the University of Maine. She co-hosts The Intersection on Discover blogs with Chris Mooney and has contributed to DeSmogBlog, Talking Science, Wired Science and Seed. She was born in Suffern, New York and is also a musician. Sheril lives in Austin, Texas with her husband David Lowry.Interested in booking Sheril Kirshenbaum to speak at your next event? Contact Hachette Speakers Bureau 866.376.6591 info@hachettespeakersbureau.comFor more information, visit her website or email Sheril at srkirshenbaum@yahoo.com.


See More

Collapse bottom bar