Wednesday, June 30, 2010

Correlation, causation, and the most offensive article I've ever seen in The Guardian


On the way home today I picked up the copy of the Guardian that was left behind by the person sat next to me.

Now, normally I'd be proud to admit to being the sort of lily-livered lefty who reads the Guardian, but then I wouldn't be banking on page 13.

The Guardian is the home of Bad Science, an excellent column by Ben Goldacre.  Yet page 13 contains an article that is a textbook example of bad science.  I'd also suggest that in the headline - Countries' high rates of disease linked to lower IQs, claim scientists - that the editor should could arguably have insisted on putting the word "scientists" in inverted commas and preceding it with the word "racist".

It starts innocently, albeit banally, enough with a no-shit-Sherlock opening gambit:
People who live in countries where disease is rife may have lower IQs because they have to divert energy away from brain development to fight infections, scientists in the US claim.
Well, yes, quite.  It should be no surprise that small children who are ill or suffering from malnutrition do not have the same resources to devote to brain development.  But this is hardly news.

Nevertheless, there has been research, apparently, that delivers the astonishing results that countries with high incidence of disease also tend to have low national IQ averages.

This should not be surprising to anyone.  It's an obvious correlation.  Countries that are not very good at providing healthcare tend also to not be very good at providing lots of other things, like education.

This element is a correlation.  The two are related, but not they are directly influencing each other - they are simply being simultaneously influenced by a third independent variable.

There is also some amount of causation.  These elements mean that the state of someone's health may have a direct knock-on effect on their own or someone else's IQ score.

If a small child is sick a lot, their brain development may be affected.  If an older child is sick a lot, they will miss days off school.  If an adult is sick a lot, a class will go without a teacher, or a child will stay home to look after their parent.

Similarly, there is an obvious causal link the other way - poor education increases ignorance in the prevention of disease. These two variables are inter-related.

As the article states:
The average intelligence of a nation is likely to be governed by a complex web of interwoven factors.
We all know that IQ tests are very good at testing whether people are any good at IQ tests.  Which is decent enough if you're comparing people from similar cultures and backgrounds.  They're an awful measure of intelligence the moment you want to compare people on opposite sides of the world.  White middle class Westerners will win hands down every time.  That's because we designed the damn things.

And even if IQ tests were any good across cultures, you cannot separate out the correlation from the causation for this research and suggest a predominately causal link simply by looking at the health numbers and comparing them with the IQ numbers.

Yet Dr Randy Thornhill of the University of New Mexico claims to have done precisely that, and more:
Thornhill believes that nations who have lived with diseases with long periods may have adapted, by developing better immune systems at the expense of brain functions.
Oh look, what a surprise.  After all that it turns out to be looks very much like a "Brown People Are Intellectually Inferior" study, with an extra helping of "Don't Shoot The Messenger, It's Evolution" on the side.

I'm not quite sure how they account for the fact that until very recently (in evolutionary terms), child sickness and mortality was a very real and serious issue in Western countries, if genetics are responsible for the difference.

[Edit: In fairness I could be off beam with accusations of racism (or xenophobia) here, and my understanding of what is meant by "developing better immune systems at the expense of brain functions", but that was my interpretation when I first read the article.

Either way, the science looks to me to be way off.  The improvement (by evolution or otherwise) of one body function does not automatically necessitate the impairment of another.  And his use of statistics is dubious at best.  A very sophisticated study would be required to draw the conclusions he's come to, yet all he seems to have done is take existing data and compared numbers.]

But then, we should expect little more from "Dr" Thornhill:
Thornhill made the news in 2000, when he coauthored a provocative book called A Natural History of Rape in which he argues that sexual coercion emerged as an evolutionary adaptation.
Ah yes, we can add mysoginist to Thornhill's list of crimes.  According to Kenan Malik's review at the time, the Natural History of Rape is little more than a litany of excuses for rape (or "sexual coercion" as the Guardian cowardly insists on referring to it).  It's another version of the tired old "I can't help it, it's my biology", and "well, you shouldn't dress so provocatively then", wrapped in a cloak of academia.

The very first page of results I get for him on Google include such intellectual masterpieces as It's All Good Nudes For Fertile Strippers, Female appearance: facial and bodily attractiveness as shape, Female Infidelity and Paternal Uncertainty, The Logic of Female Orgasm, Sex-mad girls 'like animals' and, bizarrely, Why do old couples look alike?

What a lovely chap.

What really makes me angry about this is not that these sorts of people exist, is that they're given any credence at all by a supposedly respectable (and liberal) national newspaper.


* picture pinched from jewcy.com

3 comments:

  1. There's an interesting discussion on the Bad Science regarding the article http://www.badscience.net/forum/viewtopic.php?f=6&t=16990

    ReplyDelete
  2. Very interesting indeed, though it's focusing on the accusations of racism rather than the bad science.

    I think you can very quickly end up with an argument about whether or not something is racist rather than whether the research was conducted well - which is why on reflecting I made a few edits.

    I'm absolutely certain that the variables involved have a far more complicated relationship than a simple one-way causal one, and unless the research turns out to have been badly represented by the paper, then I'd be dismissing it very quickly.

    I'd particularly like to know what the research really says about the "developing better immune systems at the expense of brain function" issue. The wording just sits very uncomfortably with me indeed, both scientifically and ethically.

    It does sound, at the very least, suspiciously like Dr Thornhill may have decided on a conclusion based on his own preconceptions or prejudices (racist or otherwise) and then used available data to support this.

    If the research has been misrepresented by the article then it's probably very dull and obvious research.

    ReplyDelete
  3. ***We all know that IQ tests are very good at testing whether people are any good at IQ tests. Which is decent enough if you're comparing people from similar cultures and backgrounds. They're an awful measure of intelligence the moment you want to compare people on opposite sides of the world. White middle class Westerners will win hands down every time.***

    That isn't correct. East Asian's average above european populations.

    I'd recommend you check out Garrett Jones & Joel Schneider's paper below which discusses the growing body of cross-national research:

    "Considerable effort has gone into producing nonverbal IQ tests that can be used in any culture. These “culture-fair” or “culture-reduced” IQ tests have been shown to predict important life-outcomes with validity coefficients comparable to traditional IQ tests designed for specific populations."

    http://mason.gmu.edu/~gjonesb/Immigrant%20IQ

    ***And even if IQ tests were any good across cultures, you cannot separate out the correlation from the causation for this research and suggest a predominately causal link simply by looking at the health numbers and comparing them with the IQ numbers.***

    True. What is interesting though is that average iq does seem to predict any number of socio-economic outcomes. In terms of health, Linda Gottfredson & Ian Deary have done a lot of research on this.

    Gottfredson, L. S., & Deary, I. J. (2004). Intelligence predicts health and longevity, but why? Current Directions in Psychological Science, 13(1), 1-4.

    http://www.udel.edu/educ/gottfredson/reprints/2004currentdirections.pdf

    ReplyDelete

Hello. Feel free to leave your comments and related links here. (Be nice.)