Tuesday, April 24, 2007

The National Inquirer, now fortified with Patrick Cockburn

So last week's mobile-phones-kill-bees screamer front page was bad enough. They ignored all the countervailing evidence and picked out a tiny uncontrolled study carried out in someone's spare time that neither mentioned the condition they were interested in, nor even attempted to measure how much RF energy they were using.

This Sunday, they were at it again, with another electrosensitivity pseudoscience screamer. This time it was WLAN that was going to kill everyone (never mind that, even if you believe that "pulsing" has a mystical influence more important than the amount of energy involved, WLAN works very differently from any cellular technology), based on following evidence.

1) A classics master at Stowe School, who complained of headaches before entering his classroom. This he attributed to the recent deployment of a wireless LAN, which was removed. No follow-up has been carried out to my knowledge to determine if he feels any better.

2) Some random bloke who had bees in his loft, which exterminators failed to remove, but which left after he installed a WLAN router.

The problem here is that if you do something, and something changes, your head is wired up by evolution to assume that it was because of your action. Cognitive psychologists call it the fundamental attribution error, and there's a lovely story about one of its discoverers, Daniel Kahnemann. Kahnemann was asked by the Israeli Air Force to lecture to their flying instructors on what his research showed about learning processes. Kahnemann prepared a lecture based on some results that seemed to show that positive reinforcement - being nice - was a more effective teaching technique than negative reinforcement - chewing-out anyone who gets it wrong.

When he gave the talk, though, one of the grizzled instructor pilots instantly responded to say that he knew without a doubt that no-one learns anything unless you SHOUT at them. So far, so stereotyped, but then, up pops another. No, he says, Professor Kahnemann is damn right. And so on. The problem was that statistically, if one of their students had a bad day yesterday, he was likely to have a better one today - regression to the mean. So, if the instructor had yelled at him, he was likely to perceive an improvement. And he was just as likely to perceive that, had he been supportive instead, because that's how human beings work.

This is why you need things like big statistical samples, null hypotheses, tests, follow-up and the rest. On the same page, the Indy mentioned a school where - wow! - after a campaign by parents, O2 and Orange had agreed to move a shared cell-site. This was given as evidence that mobile phones *are* dangerous - it might of course be that people like a quiet life when this costs little - but worse followed. The paper issued a string of figures "from the campaign" that seemed to show that a lot of people there had headaches, skin inflammations, or red eyes.

What was missing? Well, how many people among the population report headaches? Close to 100 per cent sounds about right. Nor is there any postevent data to find out if it had any effect. You must be joking.

But there was worse. The Indy's environment editor, Geoffrey Lean, again repeated the deeply stupid and dishonest claim that a recent Finnish study showed that one was "40 per cent more likely" to have a brain tumour on the side of the head you used your phone. But it didn't. In fact, the study - available here - showed that there was no greater risk of a brain tumour whatsoever. People who *did* have a brain tumour were 40 per cent more likely to say they used their phone on that side of their head.

Now, if this was a real result due to the phone, something really weird must have been happening. Mobile phone use must have been transferring brain tumours from one side to the other! This is obviously silly. More likely, those monkey brain logic bugs struck again. Confirmation bias means we seek out information that fits with our worldview. Could you really give an accurate estimate from memory of which side of your head you used a mobile phone over a period of ten years?

Also, Lean again ignored a string of copper-bottomed, peer-reviewed, randomised-controlled trials he didn't like. There's the Danish study of 420,000 people over 25 years I mentioned in the first link above. There's also this one in the British Medical Journal that shows that people who claim to come out in hives when they meet a phone have the same symptoms whether they are exposed to GSM signals, or whether they are just told they are. The Indy? Nix. They also managed to quote Sweden's Misleader of the Year 2004.

And finally, just to pile on the psuedo-scientific bullshit: Lean and the Indy even boast about the web traffic the last lot of cheap-ass crapola brought in, quoting three bloggers - but not one who disagrees.

Finally, someone will probably point out that I work for Mobile Communications International magazine. Well, it's true. Aren't I just seeking out information that suits me? Perhaps. But the good thing about science is that it's a machine designed to correct for bias, and I've got the data.


google said...

I used to expect better, but then I realised that I wasn't learning from experience. So in recent weeks I have been repeating to myself several times per day "man is not a rational animal". I recommend it to help avoid disappointment.

Matt Wardman said...

Indy link?

Or am I being stupid?

Quite suitablty the captcha on this comment is "apawl".

Rickey Henderson said...

testing to see if this works...

Dick said...

Good stuff, Alex. I used to get headaches after classics lessons. The poor sod at Stowe is probably just experiencing the effects of years of exposure to a dreary dead language.

james higham said...

Alex, one wonders about the findings of many of these studies, who's funding them and who stands to benefit by the 'findings'. I'm a bit cynical about it all.

kostenloser Counter