I heard on the radio today that children who were breastfed as babies have a 90% chance of having an improved IQ. Or possibly of having a higher intelligence – I can’t remember the wording.
My immediate response to this was to feel hard done by yet again. It’s all very well these scientists coming up with all these benefits that encourage current mothers to breastfeed, but what about those poor folk like me who were bottle fed? Do they think how it makes us feel? Not at all. Cold, unthinking lot, scientists.
But then a little pet hate crept in and I wondered just what the BBC had said. I checked out their website, and there was a piece entitled Gene ‘links breastfeeding to IQ’ hence my title to this post. So far, so good. In fact certainly better than the bit of radio news I heard that left me thinking that it was a random distribution where 90% came out better, rather than having a genetic cause (though they may have gone on to cover the genetic aspect, as I turned the radio off at this point).
However, and it’s a big ‘however’, the online version of the news goes on to commit a cardinal sin. The teaser paragraph under the headline reads ‘A single gene influences whether breastfeeding improves a child’s intelligence…’ it says. No it doesn’t.
How many times does this have to be said? IQ is a measure of how good you are at passing IQ tests. It was designed for dubious reasons, and has been thoroughly trashed as a measure of intelligence. A single number, based on a very selective set of tests, simply can’t indicate a level of something as complex as intelligence. Yes, it’s correlated to aspects of academic achievement – but then a fair amount of academic achievement is really just a measure of how good you are at passing tests too.
For a harder-to-read (in part because of the font – sans serif fonts work better on screens, Ed), but more satisfying description of the story see Not Exactly Rocket Science
Brian, you should not sully your high intelligence with such trash. In much the same way that Frank Zappa said that rock journalism was people who couldn’t write interviewing people who couldn’t talk for the benefit of people who couldn’t read, much science journalism of this sort consists of biomedical scientists (most of whom knoweth not a statistic from an hole in the ground) peddling correlational studies to credulous journalists (whose grasp of statistics is even less secure) to members of the public (whose understanding of statistics is notoriously bad) – and concluding that correlation and causation are more or less the same thing, especially in complete darkness in a coal cellar during a power cut. I offer as my example the famous German Stork Theory in that birth rates in Germany correlate with the abundance of pairs of breeding stork.
Thanks for the link Brian.
I just want to point out that all the journalists who wrote silly pieces misinterpreting this research had absolutely no excuse. When I asked the press office in charge for the paper, they not only sent the PDF across but a press briefing detailing exactly what the study showed and didn’t show. All the right info was there in easy-to-read language under obvious headings. Neither the press release nor the press briefing made any scientific slip-ups and neither of them mentioned words like ‘intelligence’, ‘smart’ or ‘brainier’.
This was an absolutely textbook example of researchers and press offices collaborating well and getting it right. There’s no excuse for ballsing it up at the final hurdle.
Oh and yes, I was thinking about changing the font.
Henry – I seem to remember when studying stats as part of my OR course that there was also quite a good correlation between imports of bananas into the UK and pregnancy rates in the years immediately following the Second World War. The imagery is striking, but the causality is limited.
For an excellent book on abuses of statistics and other numbers by journalists and the like, see The Tiger that Isn’t by Michael Blastland and Andrew Dilnot.