The new "helmet study" is a joke
Another quick hit:
So Elizabeth Christian has gone berserk defending her husband's new proposal for a study of cyclists who end up at the hospital with injuries (correlating to helmet use). This is exactly how the original Thompson/Rivera study went wrong. Short summary:
- Voluntary helmet-wearers and non-wearers are quite different groups, as it turns out. The helmeted cyclists were more likely to be yuppie recreational riders (like Ms. Christian's husband) while the un-helmeted cyclists were more likely to be poor and/or just trying to get around (in which case a helmet is enough of a pain in the ass that most rational people leave it at home).
- Later analyses of the Seattle study showed that in addition to behavioral and locational differences, helmet-wearers were also far more likely to go to the hospital for a given injury than non-wearers (probably due to the above socioeconomic differences).
- This means that the doctor in the emergency room is only going to see a non-helmeted cyclist when the injury was very serious; but he in fact sees the helmeted cyclist for minor injuries.
- Surprise! Helmet use seems to correlate with less severe injuries!
- As it turned out, though, you were also able to use the same data from this study to 'prove' that wearing a bicycle helmet reduced your likelihood of getting a leg injury by a similarly high percentage. Again, the guys with broken legs went to the hospital no matter what; but the non-helmeted guys with cuts and bruises just went home and sprayed Bactine while the helmet-wearers were more likely to go to the hospital; and the helmet-wearers were more likely to be leisurely riding through a park and suffer their falls in the grass rather than be hit by a motor vehicle on the roadway.
This is a clear study error. The "control" group in this case-control study is not similar enough to the "case" group to make these conclusions. Statistics 101; and don't believe the typical bullshit response about lies, liars, and statistics - this example is pretty damn clear-cut. The study was flawed; and this new study will be equally flawed.
Of course, the Chronicle didn't bother going into this level of detail, despite the fact that I'm sitting right here, and am no stranger to those guys. It's as if they're not even interested in trying anything more strenuous than reporting on press releases these days...
More on the Thompson/Rivera study from a slightly different angle.