Just in case something urgent pops up and you don’t have a chance to finish reading this whole piece, here’s the condensed version: Wear a helmet!

Maybe you’re a great biker.  Maybe you would never, ever crash your bike.  So what?  Do you know how many dudes from my old high school drive around drunk at all times of day?  Doesn’t matter whether or not you’re superfly on two wheels if somebody in a pickup swipes you.

Image by waferboard on Flickr.

Conk your head hard, you might die.  Even if you don’t die, a traumatic brain injury can leave your thoughts fuzzy and eyesight blurry for the rest of your life.  Some five million people in the U.S. are suffering from traumatic brain injuries, picked up from concussions during youth sports, exposure to brain-pummeling blasts of sound in the military, sudden impact during car or bike wrecks.  Now they have throbbing headaches, chronic fatigue, troubled sleep, depression …

Your brain makes you you.  And there’s no good way to fix it.  This is tragic, obviously.  If you’re feeling too chipper, go ahead and search Pubmed for the wacky psuedoscience people are trying in their flailing attempts to feel better — and that’s the stuff that’s peer reviewed.  It gets even worse if you search the whole internet.  There’re plenty of quack doctors out there willing to profit off false hope.

It’s not worth risking all that just to keep your hair from getting mussed.

Image by Ruth Ellison on Flickr.

And yet.  Even though everybody knows now that football causes brain injuries, some hundred million people watch the Super Bowl, and over a million high schoolers play.  Even though everybody knows you’re safer wearing a bike helmet, something like 50% of neurosurgeons don’t.

Those doctors know better than anyone else how horrible traumatic brain injuries can be.  But the practice of medicine encourages cognitive dissonance.  Spending all day every day around sick people can wear you down unless you maintain a clear separation between you, the healer, and them, the sick people.  And so, even though their professional organization distributes information saying that everyone should wear a helmet every time they bike, they think of themselves as special.  Separate.

They’re wrong, of course.  Medical doctors get knocked off their bikes just like everybody else.  Indeed, the more training and expense that’s been poured into your brain, the more you ought to value it.  Seems absurd me that anyone would risk easily-preventable brain damage after spending several hundred thousand dollars on an education.

Researchers who don’t wear bike helmets must realize that their cognitive dissonance is absurd.  Because every so often another flawed study is published suggesting that bike helmets make you less safe.  The only explanation I can think of for all these studies is this: some researchers don’t want to wear helmets, and then, because they’re oh-so brilliant, they design & publish crappy experiments to justify their decision.

Image by Greg Raisman on Flickr.

The first time I learned about these studies was shortly after beginning graduate school.  The New York Times publishes an annual “year in ideas” magazine, and one of the great “ideas” that year was that drivers might think they need to be less cautious when passing a biker with a helmet.

There were obvious problems with this study.  The biggest problem is that only a single biker was studied: the researcher who designed the experiment and who acknowledges that he doesn’t like wearing helmets and wanted to find a negative consequence of helmet-wearing.  This is as far from a double-blinded study as you can get.

Even then, it took bizarre data analysis for him to get the conclusion he wanted.  (Insert obligatory Ronald Coase quotation here: “If you torture the data long enough, it will confess.”)

Any time you obtain lots of data and analyze it lots of different ways, you can make it seem to support the conclusion you want.  You can read a very clear description of this — replete with lots of fun examples — in Gary Smith’s Standard Deviations.

Even though the researcher’s bias probably led him to bicycle more recklessly when he had a helmet on, his data don’t support his conclusion.  Consider this scatterplot prepared by other researchers who re-analyzed his data.



Drivers often don’t pass any more closely when he has a helmet on.  At a glance, the only conclusion I’d make from looking at this is that he was probably safer when he was farther from the curb — there’s less debris farther into the road, and drivers were giving him just as much room.  But, regarding helmets?  The data are a wash.

Of course, given scientific publishing’s & our popular media’s preference for exciting, counter-intuitive results, the correction — you should wear a helmet, after all! — got much less press than the original, incorrect conclusion.

Yet another “maybe you shouldn’t wear a helmet” study was just published.  This one purports to investigate “risk compensation,” the idea that wearing a helmet makes you feel safer, and that the feeling of safety might cause you to bike recklessly, with the net result being that you’re even less safe than before, sans headgear.

It’s a cute idea.  And it’s been studied extensively.  Bikers who usually wear helmets do seem to bike more slowly when they don’t have a helmet on.  That same study found that bikers unaccustomed to helmets didn’t change their behavior when wearing one.  And another study found that helmeted bicyclists typically engage in less risky maneuvers.  (That last study simply tracked people who did & did not choose to wear helmets, as opposed to randomizing the population.  It’s quite possible that people who choose helmets are less likely to take risks in general.)

All of which indicates that, yes, you should wear a helmet.  Don’t worry, kid, you’re not going to psych yourself out and get the itch to careen in front of cars as soon as you strap one on!

But some researchers don’t like these results.  They want to bike without helmets, and they want to feel brilliant while doing so.

Thus, the new study.

This one used a highly contrived scenario, where research subjects were awarded prizes based on how large they could inflate a computer-simulated balloon.  They had to wear headgear, ostensibly to track their eye motion, during the experiment.  For some participants, the eye-tracking cameras were attached to baseball caps, for other participants the cameras were attached to helmets.  Voila!  A blinded study!  These people are now wearing helmets without even realizing it!

The researchers found that helmeted subjects inflated their virtual balloons some 30% larger than baseball-capped subjects.  Voila!  The researchers demonstrated risk compensation!

Except … people don’t act riskier in the real world when they strap on helmets.  Yeah, it’d be somewhat curious if having a helmet on — which clearly won’t protect a digital balloon from disaster — caused people to behave more cavalierly.

Their results hardly show that, however.  They barely even published their results!


Each “virtual balloon” was designed to “pop” after a random number of inflations, between 1 and 128.  Participants “inflated” the balloon by clicking a mouse.  Which seems odd in and of itself — if they’re using balloon imagery, why wouldn’t the thing be more likely to pop when it was big than when it was small?  If you’re claiming that people’s sense of real-world protection from wearing a helmet affects their decisions, you should make your virtual design as close to the real-world as possible.  Otherwise you’re just introducing a strange cognitive disconnect between expectations and results.

What the researchers published was, for each of the 80 study participants, the average number of clicks for all trials in which the virtual balloon did not pop.  I looked through their paper, but I couldn’t find the number of trials excluded because the balloons popped.  So even if they found a significantly lower average number of clicks between helmeted & baseball-capped participants, this doesn’t mean anything.  You’d see this same result if the effect of a helmet was to make people less erratic, i.e. safer.

An erratic participant might choose a number of clicks with high variance — in one trial she might click 30 times, in the next trial 90 times.  A safety-conscious participant might always click the same number of times — always 60 clicks.  Well, if the second balloon for the erratic participant popped, the averages recorded would be 30 clicks for her, 60 clicks for the other subject.  The reckless participant, because of this strange data manipulation, looks more risk averse!

This very well might be what the researchers saw.  With a quick calculation, you can find the optimal strategy for number of clicks… it turns out to be 64 clicks.

Here’s that calculation.

Half the time you use this strategy, your balloon will pop, but your total prize will be as high as possible.  And, with the researcher’s strange way of calculating the average number of clicks, someone using this strategy would have an average of 64.  In each trial, they either clicked this number of times, or the balloon popped.

bike helmet graphHere’s the researchers’ graph of their data.  What you can see: zero study subjects wearing caps used the optimal strategy — everyone from that population had an average below 64.  No one from the helmet-wearing group had an average of exactly 64 either, but there’s a cluster of scores nearby.

My conclusion?  Wearing a bike helmet makes people smarter.  The helmet group was much closer to playing the game optimally.

(Um, sure, my conclusion is spurious too.  Most people in both groups played the game the same way.  All that these data show is a small number of players in the helmet group with higher average click numbers — again with the caveat that the method used to calculate the average is bizarre, and wrong.  You can’t really conclude anything from this study.  But you certainly can’t conclude that wearing a helmet influences people to take more risks.)