Bad Science

I'm an enormous fan of Bed Goldacre and his Bad Science activities: the blog, the column in The Guardian and of course the book so by way of a tribute here's two things that annoyed me in this month's CSMA Club Magazine¹.

The first article concerns electronic stability control (ESC). It goes under lots of other names and TLAs too, depending on the manufacturer but they're all playing the same game: an ECU watching the steering wheel and what the car is actually doing and can then brake individual wheels to help the driver accomplish what they're trying to do. The journalist in question tried it in a Honda Civic on a skid pan and loved it.

Up to that point nothing wrong with the article at all but then the statistics are called into play to "prove" why ESC should be compulsory on new cars. Apparently the DoT say that "cars fitted with ESC are 25% less likely to be involved in a fatal accident than those without".

Now at first glance that sounds like a good reason but there's two problems with the statistic. Firstly are they comparing like with like age wise: older cars won't have ESC but then they're less likely to have various other safety features like air bags.

Secondly the article states that ESC is currently fitted to about 50% of new cars and it tends to be smaller, cheaper cars which don't have it. Now my strong suspicion is that you're already more likely to suffer a fatal accident in a small car than a big one (more momentum, bigger crumple zones, etc, etc).

The bottom line is that one has to doubt the 25% can be attributed to the fitting of ESC: correlation does not imply causation.

The second article is more up Ben's street. This is in the "Well Being" section which serves up masses of dubious junk science every month. And here's a belter.

The article is headed "Healthy Nuts"(!) and I'll quote it in its entirety as it's short:

Just a handful of pistachios a day may help reduce the risk of lung and other cancers, a new study has found. The nuts are rich in gamma-tocopherol, a form of vitamin E, and in a six week trial participants who ate around 120 nuts a day showed a significant take up of the vitamin.

Dr Laura Bell of Cancer Research says "Research is gradually helping to unpick the complex relationship between lifestyle factors, such as diet, and cancer risk. However the biggest risk factor for lung cancer is still smoking."

It really is hard to know where to start really. For a start there's the concept of eating 120 pistachios a day. Every day. Now I like the odd pistachios but would you really want to eat that many every day?

As for gamma-tocopherol it is indeed a form of vitamin E and there have been various studies trying to link vitamin E to reduced cancer risk for various cancers, including lung, and at least one large study² unexpectedly found an increased risk of lung cancer from taking alpha tocopherol (a related form of vitamin E). But the article doesn't say anything about that. Indeed it doesn't say anything stronger than "may help reduce the risk" and that's unsurprising really given the state of research.

So that leaves poor Dr Bell. I'd love to see what she said before that sentence - certainly what she is quoted as saying is hardly a ringing endorsement for a diet heavy in pistachios.

Which is unsurprising really.

1. For the benefit of those of you who've not come across it the Civil Service Motoring Association is a weird backwater which you can join if you've ever worked for the civil service and offers a very good deal of car breakdown cover which is why we're members. But it also offers lots of other benefits too, which it offers through its monthly club magazine, which also has lots of articles reviewing new cars and advertorial on holidays, investment, health etc. It's clearly aimed at an older demographic: lots of adverts for cruises and vitamin pills.
2. More on this here on the Cancer Research UK web site.