Thursday, July 28, 2016

There are more geniuses among men than among women, and more idiots too

Deary and colleagues (2007) conducted an interesting study on differences in intelligence scores among men and women. In the context of this blog, this study highlights yet one more counterintuitive and intriguing aspect of Darwinian evolution, adding to points previously made in other posts (see here, and here). Evolution may look simple at first glance, but that is a bit of a mirage. In my opinion, to really understand it one has to understand the mathematics underlying it, a lot of which comes from the field of population genetics.

What makes the study by Deary and colleagues (2007) particularly interesting is that its participants were opposite-sex siblings. This helped control for the influence of environmental factors. The downside is that the effect sizes might have been decreased, because of the high gene correlation among siblings, so we could expect larger differences between unrelated groups of men women. The differences, as you will see, are not in overall scores, but in score dispersion.

Let us get straight to the point made by the study. On average, men and women seem to score equally well on intelligence tests. The main difference is that there is more variation in the scores achieved by men than by women, which leads to an interesting effect: there are more geniuses and more idiots among men than among women.

This does NOT mean that a man’s genius is of a higher order; just that there is a tendency for more men to be geniuses (and idiots) than women in any random population sample. The women who are geniuses can be super geniuses, like two-time Nobel Prize winner Marie Curie, the first PERSON to receive such an honor. Albert Einstein is said that have greatly admired her intelligence.

As an illustration of this score dispersion effect, Deary and colleagues (2007) note that: “… for example, in terms of indices of scientific achievement, men were awarded 545 out of the 557 Nobel prizes awarded for science.” On the “idiot” end of the scale: there are a lot more men than women in prison, and one common denominator of prison inmates is that they tend to score very low on intelligence tests. (This is not to say that all criminals have low intelligence; perhaps mostly the ones that get caught do.)

Having said that, it is important to acknowledge that there are multiple types of intelligence, and even multi-indicator intelligence coefficients are usually poor approximations of an overall measure of intelligence (if there is one). This does not invalidate the main point of this post, which is related to score variability.

The table below (from: Deary and colleagues, 2007; click on it to enlarge; full reference at the end of this post) shows scores obtained by men and women (1,292 pairs of opposite-sex siblings) in various subtests of the Armed Services Vocational Aptitude Battery (ASVAB) test.


Note that nearly all of the differences between means (i.e., averages) are significant, but the direction of the differences (captured by the signs of the Cohen’s d coefficients, which are measures of effect size) varies a lot. That is, on several subtests (e.g., “Arithmetic”) men score higher, but in others (e.g., “Numerical operations”) women score higher. It all comes down to men and women scoring equally well overall.

Now look at the columns showing the standard deviations (“SD”) for men and women. In all subtests but two (“Coding speed” and “Numerical operations”) the standard deviation is higher for men; in many cases significantly higher (e.g., 44 percent higher for “Mechanical comprehension”). The standard deviations are about the same for “Coding speed” and “Numerical operations”. What this means is that variability in scores is nearly always higher, often significantly higher, among men than among women. I prepared the schematic figure below to illustrate the effect that this has on the numbers of individuals at the extremes.


The figure above shows two (badly drawn) quasi-normal distributions of scores. (This post shows a better illustration of a normal distribution.) The red curve refers to a distribution with a lower standard deviation than the blue curve; the latter is flatter. Each point on a curve reflects the number of individuals obtaining a particular score, which would be indicated on the horizontal axis. The number of individuals with that score is on the vertical axis. As you can see, the numbers of individuals scoring very high and low (geniuses and idiots, if the scores reflected intelligence) are greater for the blue curve, which is the curve with the higher standard deviation (higher dispersion of scores). The farther one goes to the left or right (the extremes), the bigger this difference becomes.

What does this have to do with evolution?

Well, there are a few possibilities, two of which appear to be particularly compelling. Maybe this effect is due to a combination of these two.

One is that ancestral women, like women today, selected mating partners based on a wide range of traits. Ancestral men on the other hand, like modern men, focused on a much smaller set of traits (Buss, 1995). The end result is more variation in traits, generally speaking, among men than among women. This refers to traits in general, not only intelligence. For example, there seems to be more variation in height among men than among women.

The other possible explanation is that, in our ancestral past, staying out of the extremes of intelligence was associated with higher survival success in both sexes. It seems that the incidence of certain types of mental disease (e.g., schizophrenia) is quite high among geniuses. This leads to more deaths due to related issues – suicide, depression leading to the metabolic syndrome, etc. And this is today, where geniuses can find many opportunities to “shine” in our complex urban societies. In our ancestral past the cognitive demands would have been much lower, and so would the practical value of being a genius.

If staying out of the extremes has indeed enhanced survival success in our evolutionary past, then it is reasonable to expect more women to fit that pattern than men. As with almost any “thing” that enhances survival success, women (especially pre-menopausal) naturally have more of that “thing” than men (e.g., HDL cholesterol).

The reason is that women are more important for the survival of any population than men; today and 1 million years ago. A population of 99 women and 1 man can potentially generate 99 children every few years. Here inbreeding in subsequent generations will be a problem, but that is better than extinction. A population with 99 women and 99 men (or even 1,000 men) will not generate significantly more children.

Reference:

Buss, D.M. (2003). The evolution of desire: Strategies of human mating. New York, NY: Basic Books.

Deary, I.J., Irwing, P., Der, G., & Bates, T.C. (2007). Brother–sister differences in the g factor in intelligence: Analysis of full, opposite-sex siblings from the NLSY1979. Intelligence, 35(5), 451-456.

30 comments:

Jenny said...

Sorry, but the definition of intelligence used here were skewed in ways that provide advantages to males for reasons that have nothing to do with how the brains of the people involved functioned.

During most of the century when Nobel prizes were awarded, the science establishment marginalized women no matter what their skills, and often males appropriated their work, as in the case of Rosalind Franklin.

And the armed services define intelligence in ways that promote military success.

Men and women tend to think in different ways, though studies of women in math and computers have found the women who build careers in these fields have much better math skills than the men--probably because they have to, to break in.

Let's not look to evolution to explain this kind of result but to social pressures that explain why, though females always outperform males in school through adolescence, they are steered away from achievement or kept from receiving its rewards as they get older.

Anarcho-Mercantilist said...

Typical IQ tests involve pattern matching and pattern recognition of shapes. So IQ tests are skewed towards spatial intelligence. In fact, men have much better spatial intelligence than women. So it does explain why there are male "geniuses" than women.

I have some theories why there are more male idiots than women idiots:
- Men have only one X chromosome. So a mutation in that chromosome results in a permanent defect. (Women have two X chromosomes, so each one protects the other).
- Men are more risk-taking. So they are more prone to do activities which result in physical trauma (such as brain damage).
- Men don't really care about scoring high on IQ tests as women do.

And, also, intelligence doesn't equal success. Even though the bell curves of between the sexes overlap, why are there much more successful men than women? Why are almost all inventors and composers men? I think it is because men are more hardwired to be status-seeking than women. Geniuses invent stuff to raise their status.

Pål Jåbekk said...

Although intelligence testes may be flawed in various ways they definitely measure something and I must say data showing sex differences are really interesting. Nature vs nurture is an ever going discussion but I really would have liked to see the intelligence scores correlated with sex hormone levels.

Pål Jåbekk said...

Woops! An extra e to the word tests and what do we get? A new and recently discovered form of gonads. Sexual humor not intended.

Anarcho-Mercantilist said...

I have replied to your article here.

Ned Kock said...
This comment has been removed by the author.
Ned Kock said...

Jenny:

The dispersion effect is real, and so are the consequences. That does not say anything about possible discrimination, which I do not doubt happens. It's rather sad. Having said that, Marie Curie (born "Maria" in Poland, and raised there, by the way) received two Nobels in the early 1900s.

anarcho:

Good points about the XX and XY chromosomes, and the protective effect of XX from deleterious heterozygous mutations.

Risk-taking behavior among men is one of the my favorite examples of Zahavian (or costly) traits in humans (see post below).

http://healthcorrelator.blogspot.com/2010/01/evolution-of-costly-traits-challenge-to.html

Pål:

Yes, sex hormones may mediate the score dispersion effect. This is a very understudied area; maybe too politically charged to attract funding.

Anarcho-Mercantilist said...

Jenny,

It's interesting that Richard Feynman has an IQ of only 126, and than James Watson has an IQ of only 115. I guess that their ambition is a better indicator of success rather than intelligence.

There's a study that testosterone increases status-seeking behavior. This study may suggest that males succeed not because of their intelligence, but because of their ambition.

There may be many women who are geniuses, but these women don't have the motivation to use their intelligence to become successful. Genius may be a product of ambition, rather than intelligence.

It's said to realize that the testosterone levels of both sexes are falling, and society in general are becoming less and less ambitious.

Ned Kock said...

I received an interesting comment by email on this post, and invited the commenter to post it here.

The dispersion phenomenon discussed in the post causes another interesting outcome: there are more women than men with "non-extreme" above-average intelligence. Say, within 1 standard deviation above the mean.

Ned Kock said...

anarcho:

The testosterone-related drive hypothesis makes some sense, but it is contradicted by the widely reported stats. showing that female students usually have higher university GPAs than male students.

Perhaps these stats. have something to do with what I said in my comment above.

Anonymous said...

It has to do with grey cells(thought proccesing). Men have more. Real novelty involves original ideas that can come only from multiple(unsafe) logical conclusions, a task that men are better at..
Women are good at dealing with data, but they are as not as good as men when it comes to theorizing on an abstruct level..

cheap viagra said...

Idiots are everywhere you look around, there are more idiots than geniuses for sure.

@Danon_40 said...

A lot of commenters take proxy explanations (number of grey cells, testosteone etc.) for ultimate ones. Others are just turning this into a silly feminist game. Why is the IQ spread in men higher? It means that both male geniuses and morons had more offspring than female counterparts. We may understand why being very intelligent can help, but being a moron? Maybe part of the response is that courage is negatively correlated with intelligence. So the best hunters and fighters had a lot of offspring too... and statistically speaking, many of them were not very smart.

Mr. Jensen said...

To the years old top comment.
Women and men perform different in school because they ARE different.
And the varity of boys varies MORE in intelligence as the chart currently shows.
This (hypothetical) make an average school education, based on the average student more fit for girls since they have a larger percent of averages.
Not to mention the explosive energy most boys have, make them very unfit for normal education.
This made me the absolute lowest scoring in my class when i was a kid, and i was considered to have way below average IQ maybe even brain damage.
And all this because of the stupidity and lack of understanding from teachers on how boys and girls are fundamentally different.
Luckily i later changed to a school, where they were able to run a education where kids that did not fit into the norm were able to learn the way that was best suited for them.

We can also talk different study habbits. Basically all girls i know within my field, studied hard got their degree - decent grades and off they were.
Me and afew other lazy(but smart) guys, 90-95% absence, never read a thing all semester, then 4 days before, read up on everything, make all the assignments given during the semester, and ready up for exams, did that with every single course on both my bachelor's.
True my grades were more spread out, but i got plenty of topgrades, to cause an outrage.

Danon_40
I think i saw a study someplace that being a moron, apparently makes you a more happy person, where high intelligence people tend to have way higher chance of depression and overall lower happines than the average population.
This is solely based on guessing, but a depressed genius is way less reliable than a happy moron. And many women craves stability, genes tells them to have a safe nest and all that. The happy moron were always there, even though he were an idiot at everything.

Anonymous said...

The following comment in my view is stupid.

"Let's not look to evolution to explain this kind of result but to social pressures that explain why, though females always outperform males in school through adolescence, they are steered away from achievement or kept from receiving its rewards as they get older."

Instead of deciding how feminism is best supported by the data, or explains the data, or is exemplified by the data, why don't we suggest different theories and then TEST THE THEORIES? I honestly am sick to death of people telling me that feminism is offended, that the data is misunderstood, because feminism has thereby been wounded. If you have something to add, add it. But I would like to hear what the author suggests, and I am interested in whatever effort you have made to come up with a increase in overall scholarship. Poisoning the well with preconceived and blindly defended opinions is ... well ... the sine qua non of stupid arrogance.

Anna Schafer said...
This comment has been removed by a blog administrator.
Elizabeth J. Neal said...
This comment has been removed by a blog administrator.
Ned Kock said...

This post is a revised version of a previous post. The original comments are preserved here. More comments welcome, but no spam please!

Ned Kock said...

Two (old) spam comments above deleted.

Tony K e4e said...

This is what forced Larry Summers to resign from Harvard. I don't know if it was this study in particular though.

Ned Kock said...

I think that Larry Summers’s statements gave the impression that he believed women were less likely to succeed in some careers than men, which is not at all what this blog post says. Btw, do you have a link to a transcript or video of what he actually said?

Anonymous said...

The research in this area seems contradictory. See e.g. Sex differences in intelligence: A multi-measure approach using nationally representative samples from Romania (just published and with 15,000 participants, although I haven't seen the full text):

"The authors investigate sex differences in mean scores and variance, in general and specific intelligence, based on raw data from the standardization samples of six cognitive ability measures adapted and normed in Romania. The results show that fewer than 10% of the possible comparisons exhibit sex differences; these differences have a random pattern and are not replicated across measures. The authors conclude that these Romanian data show no support for the sex differences in either mean values or variance of scores which were reported by other studies."

Ned Kock said...

Hi Anon. Do you have access to the actual numbers? If yes, could you post them here?

Anonymous said...

Sorry Ned, I don't.

Medcomments said...
This comment has been removed by a blog administrator.
RainbowExplorer said...

Of all the hundreds of articles I've read on the nature of intelligence and the impacts of nature vs nurture, this blog posting is certainly one of the most narrow and superficial. It clearly betrays the author's ignorance regarding the range and uses of both "intelligence tests" and "aptitude tests". No doubt, this is a direct result of his lack of social science education and expertise.

First, the blog begins by failing to define what the terms "intelligence", "aptitude", and "genius" mean, particularly avoiding any discussion of the known, inherent cultural biases in-built within any currently utilized formal IQ and aptitude testing processes. Secondly, it skims right over the topic of high intelligence rates within sibling cohorts, without thoughtful insight or reflection, let alone an extensive list of high quality research studies to back it up. Having been raised within a sibling cohort of five, three of which are considered intellectually gifted, I'm keenly aware that birth order, parental gender preferences, family socio-economic status, health and nutrition status, pregnancy difficulties, child spacing, and variations in high quality educational opportunities make significant differences in whether any individual is identified as "highly intelligent" or not, either by family members or the greater community.

Immediately, the article rushes on to discuss a sibling pairs' single year's ASVAB test results, as though suggesting those scores have been static, between the sibling pairs, throughout all of time. Any competent social scientist or certified educator could easily point out that a single day's test results cannot ever be assumed to accurately reflect one's ability level, due to countless confounders such as illness, malnutrition, injury, lack of adequate sleep, major family or personal emotional distress, time of day the test is administered, etc. Recurrent testing between the sibling pairs, over months or years, utilizing a true "intelligence test", rather than an "aptitude test", would be the only scientifically sound approach toward addressing the true nature of the sibling pairs' ongoing intelligence strengths and weaknesses. So many breathtaking assumptions, like these, stagger me in their obvious over-reach and blanket ignorance.

The fact that the author opted to focus merely on the ASVAB test results, an "aptitude test" - not an "intelligence test", of a single reported study to illustrate his personal opinions is problematic, on multiple levels. First and foremost, the ASVAB test is not uniformly given to members of all genders, either within or outside of the USA. Thus, its norms are not standardized across the genders. It is a military-focused "aptitude test", which is both optional and exclusionary. A non-randomized selection of secondary schools offer it, some by force and others optionally. Potential military recruits who wish to take the test may be denied by military personnel, regardless of their gender, interests, abilities, or any other qualifications.

As males remain the only members of the USA citizenry required to enroll in the selective service process, remain the dominant participants in all military branches, while few female military outreach efforts are in place, it is reasonable to suggest that the ASVAB aptitude test is administered to a disproportionate number of males vs females. These facts distort the norms of the ASVAB test, itself. In addition, the ASVAB test-takers are non-representative, in that they have self-selected to potentially enter military service, at some point. A more unbiased, randomized, and gender-equalized "intelligence test" process, along with decades of consistent gender-related scores, would make this author's arguments more creditable. As they stand now, they are not much more than thinly veiled, biased speculations.

Ned Kock said...

Hi Rainbow. In what way do you see the blog’s post as having “thinly veiled, biased speculations”? Btw, let you comment be clear evidence that I used moderation to weed out spam, not critical comments.

Anonymous said...

It's very popular for journalists and public intellectuals to downplay or criticize IQ and its implications. As a result, the public is wildly misinformed about the issue, and this is apparent in the comments here. How sad. There is simply nothing that's been formulated that has the type of validity with robust results as IQ testing has. It's by far the most refined and useful tool, both in terms of explanation and prediction, in the entire field of the social sciences.

tomR said...

"The other possible explanation is that, in our ancestral past, staying out of the extremes of intelligence was associated with higher survival success in both sexes." - but then the mental development wouldn't progress over time, we wolud have no civilization etc., getting stuck a the same level forever.

Isn't it more viable that the lineages of men on the left side of the IQ distribution were dying off ? With both women prefering to do the hypergamy with those on the right side, as well as those men on the right side being able to enforce a polygamy for themselves. Notice this was not a big hypergamy: around 2 women for each men.

http://tierneylab.blogs.nytimes.com/2007/09/05/the-missing-men-in-your-family-tree/

"The only definite thing was that twice as many previously living women as men have descendants alive today."

Some periods were an exception, eg. at the advent of the agriculture, with first amasing of welth possible the ratio temporarily went to 17 women leaving offspring for 1 men.

https://psmag.com/8-000-years-ago-17-women-reproduced-for-every-one-man-6d41445ae73d#.mnikmgxss

For more recent times, like in the Middle Ages, Gregory Clark did a research on survivability relation to the income of the father, based on data from the wills in England. It looked something like this: each woman had 5 children, for the rich people 4 survived, for the poorest only below replacement, something like 1,7 survived. The result over time was that 90% of the population a the beginning of the industrial revolution was descendant from just 10% of the richest of the Middle Ages, via a process of downward mobility (books: Farewell to Alms, The Son Also Rises).

https://www.youtube.com/watch?v=CEGhTtdTwfE

As far as this myth of men being less important - this is one of the least convincing aruments "A population of 99 women and 1 man can potentially generate 99 children every few years. Here inbreeding in subsequent generations will be a problem, but that is better than extinction. A population with 99 women and 99 men (or even 1,000 men) will not generate significantly more children.", because:

1) A popupation with 1 man are going to be conquered or wiped out by a populations with 99 men. Notice in the past wars were more violent, frequently leading the wipeout of the entire populations, and frequently NOT involving taking local as concubines. Eg. Original Mongol plan for conquest of china was to depopulate it, to make space for horse pastures - this was only changed after the allied tribe of Kitay tribe convinced them to tax the conquered instead.

http://londonprogressivejournal.com/article/view/1621/a-brief-guide-to-early-chinese-history-the-mongol-conquest-of-china-and-its-consequences

In Northern Europe there was a population replacement too:

https://westhunt.wordpress.com/2014/09/19/before-slavery/
http://polishgenes.blogspot.com/2014/09/corded-ware-people-more-versatile-and.html

2) A population with one man just wouldn't make it when it comes to the getting the resources necessary to survive.

3) The number of kids a woman can give birth to is much above just keeping the population alive. Top country by fertility has 7.6 kids per woman. Some women had like 20 children.




Anonymous said...

Oh well I have studied this phenomenon more closely I think. The variability hypothesis has been around since the beginning of the 20th century and there have been contradicting studies both supporting and contradicting it since then.
For instance well known psychologist Lewis Terman found the female standard deviation to be slightly higher in samples studied in 1917 thus concluding the hypothesis to be false. In 1920 he changed his mind after finding men many times overrepresented among gifted children only to change his mind again and attribute his findings to environmental factors. Poor Lewis just couldn't decide it seems.

As for modern studies there are several dealing with this now. Deary's isn't necessarily more accurate just because it deals with siblings. Seeing as those siblings took the ASVAB in the 60s there might have very well been different expectations of them which could have influenced these results.

Other than that there is Feingold's 1992 study
http://journals.sagepub.com/doi/abs/10.3102/00346543062001061
which found variability to be consistently higher among men in many domains in the US.

On the other hand his 1994 cross cultural analysis found that this pattern didn't hold up in other countries
http://link.springer.com/article/10.1007/BF01420741

There are the Hedges & Nowell studies analyzing data from TIMMS, Deary's assessment of the 1932 Scottish mental survey, but also contradicting studies such as one on a rather recent sample of 15.000 Romanians

https://www.researchgate.net/publication/305033910_Sex_differences_in_intelligence_A_multi-measure_approach_using_nationally_representative_samples_from_Romania

What often gets overlooked in all that though are the large representative samples from the SAT and ACT. In general the variability hypothesis seems to hold true for math and science reasoning where men consistently outnumber women in the top 0,5% by factors of about 2:1. (Standard deviation 10% higher)

Got some data from the SMPY to share which again shows boys overrepresented - this time by 4:1 - in the top 0,01% of participants.

https://www.psychologytoday.com/files/attachments/56143/sex-differences-in-the-right-tail-cognitive-abilities.pdf

Where the hypothesis seems to fail is verbal abilities which has women overrepresented at the top 0,01% by factors of about 2:1.
This can be attributed to a higher mean but similar standard deviation for girls in comparison to boys.

So no, I don't think it's fair to say men are more likely to be genius. Rather they seem more likely to be scientific geniuses whereas girls are more likely to excel in the verbal domain.
Consequently the variability hypothesis seems to be true for some cognitive abilities but not for all. Where general intelligence is concerned boys and girls - men and women - are most likely equal as for instance Jensen found in 1998. His studies comparing the performance of adult men and women on six g-loaded cognitive batteries were summarized as

"No evidence was found for sex differences in the mean level of g or in the variability of g. Males, on average, excel on some factors; females on others"

This finding is also supported by the Raven's progressive matrices widely known as one of the best measures of "Spearman's g" which if anything has women more likely to show greater variance although until recently women did score slightly worse on average.

http://emilkirkegaard.dk/en/wp-content/uploads/Sex-differences-in-means-and-variability-on-the-progressive-matrices-in-university-students-A-meta-analysis.pdf