Saturday, May 30, 2015

Height and intelligence

The GSS provides another data point supporting the moderate correlation between height and intelligence. In 2014, the survey queried respondents on their personal heights for the first time. The following table shows mean IQ (converted from wordsum scores with an assumed national average of 98 and standard deviation of 15) by height. To avoid racial and sexual confounding, only white men are considered (n = 306):

6'1" or taller104.9
5'6" or shorter98.0

I'm not sorry I had you, son. I was always
proud that you weren't a short man.
Both traits are presumably marginally influenced by a large number of different genes. I'd guess genes that correlate with intelligence and that correlate with height also both tend to correlate with other broader measures of general health.

GSS variables used: HEIGHT(57-66)(67-69)(70-72)(73-79), RACECEN1(1), SEX(1)


Anonymous said...

I see tons of ethnic confounding. Somehow I doubt that short French men are dumber than tall Serb men. Serbs have the lowest average IQs of all the European peoples, and other South Slavic peoples aren't much different.

Anonymous said...

I don't necessarily disagree that such a correlation exists, but besides ethnicity, might not age be a factor here too? If fluid intelligence declines with age and younger cohorts are generally taller (not to mention actual shrinkage with age) that could be influencing these results.


Anonymous said...

Using rachisp, american born and women, who are basically less likely to inflate their heights than men: white women 5 foot 5, black women 5'4.5", Hispanics 5'3.5".

Anonymous said...

I would assume people exposed to heavy metals and toxic chemicals would be shorter and less intelligent and generally less vigorous than people not exposed to these toxins.

sykes.1 said...

I ran across this correlation when I was an undergraduate in the 1960's.

Audacious Epigone said...


Yes, it looks like an interracial/interethnic phenomenon.


The wordsum is a quick 10-question vocabulary test. It correlates with adult IQ at .71 so it's not a perfect measure, but it's a good one, especially for something so quick and easy to administer. It also measures 'crystalized' intelligence more than 'fluid' intelligence, and consequently average scores top at around age 48.

Dan said...

Very good! This is actually a substantial correlation. This leads one to believe that a substantial part of the Flynn effect could be real. People are bigger and taller than their forebears, and have proportionally bigger brains.

Steve Sailer said...

Interesting correlation.

These are non-Hispanic whites?

There was probably a nutrition-driven correlation between class and height before the Baby Boomers, but most non-Hispanic white Baby Boomers were all pretty well fed. If you look just at NH white males born from 1946 onward (probably have a sample size > 200), do you still see the correlation?

Audacious Epigone said...


Yes, this is restricted to non-Hispanic white males, with a minor caveat. The GSS takes the convenient approach to the question on race by allowing respondents to choose only one of the following: White, black, six different Asian nationalities, Hispanic, 2+ races, and other, so there are probably some number of guys who look like Jorge Ramos (but speak English as a first language) who are from Central or South American backgrounds who self-identify as white rather than Hispanic. The flight has been away from white though, so I doubt that's too much of a confounding issue here, and anyway they're probably mostly to all European by ancestry anyway.

The trend remains almost identical because only 15 of the 306 on the height question were born before 1946 (2014 is the only year height has been asked about, so those born before WWII ended were 68 at the youngest at the time of the survey).

M said...

Re: White ethnic height differences in the US, you can split up the variable by ethnicity.
Problem is some groups have low sample size within a given sex, and uneven ratios of sampling both sexes, which makes comparisons hard.

So what I did was take the male and female samples for all groups, work out how many SDs they were from the US population mean, then use this to predict an overall height.

All samples -

Weighted N greater than 10 -

I probably wouldn't trust the all samples set that much, as some of the weighted N are still very low and they tend to be outliers for that reason (e.g. Hungarians).

The ethnic groups with >10 do suggest a standard North Central Euro American (British, German, Polish) male average of around 5'11", then a Scandinavian American (Norwegian, Swede) average of 6', and a Southern European American (Italian, various Spanish groups) average of around 5'9" (about an SD difference in height, if it were IQ, it'd be 15 points).

These are all estimated heights, so there might be some tendency for Scandinavians for instance to go "I'm Swedish so must be tall" and blur the lines a little.

Midwestern Americans (those Scando descendents) also tend to overreport their height, which messes up their obesity scores, making them look less obese than the South when they're not - and ( compares state obesity ranks when height and weight are measured vs asked). Midwesterners think they're taller than they are.

But that's probably not the whole effect.

(Note, I didn't adjust based on % N by sex when creating sex normalised averages, so that could make this a little stronger).

M said...

Height by degree and by father's highest degree, scores, White men.

Audacious Epigone said...

M's links more easily accessible for mobile device users out there:

Link 1
N = 10+
Father's degree

The sample sizes for several ethnic groups are really small. Hopefully the GSS will run with this question for a few more iterations of the survey.

JayMan said...

The GSS is a poor place to investigate this correlation. Sample sizes are too small, and measurement error on both measurements is too great here. Heck, my experience with JayMan Jr. has shown that even clinical height measurements can be substantially unreliable, to say nothing of self-report.

Anonymous said...

Albert Einstein was 5'8"

Audacious Epigone said...


It's still pretty suggestive. It would be highly unlikely for the results to just be noise given the way they shake out.

Anonymous said...

Averages, please.

''THE SMARTER'' in this kind of psychometric research is or tend to be statistically contextual. ''Average-smarter'' (compared with the group of ''average-less-smarter'') and not ''THE smarter''. I think to the human cognitive (positive) excepcionalities we will have a greater diversity of biotypes.


David Pittelli said...

I wouldn't assume it's mostly genetic. Environmental factories such as nutrition, and the vagaries of individual health, could affect skeletal and brain growth commensurately.