If you ride a bike, you likely have a story about a close call with a car. According to the National Highway Traffic Safety Administration, 840 US cyclists died in 2016 in collisions with motor vehicles (pdf). In the end of its report, the NHTSA offers a list of safety tips; first among them is to wear a helmet. “A helmet is the single most effective way to prevent head injury resulting from a bicycle crash,” it says.
Both common sense and decades’ worth of studies strongly suggest that helmets keep our bucket o’ brains intact in serious impacts. However, over the last decade, cyclists and the scientists studying them have been debating the veracity of one rumored negative consequence of wearing a helmet: drivers seeing a helmeted biker behave more dangerously.
The touchstone for the debate is a study published in 2007 by psychologist Ian Walker. Walker rode 320 km (about 200 miles) (paywall) on his bike in two English towns, wearing street clothes to appear like a casual commuter. He varied certain details across his rides. To test the theory that riding confidently in a lane (rather than on the shoulder) makes it more likely that cars will respect bikers’ space, Walker would change up how closely he rode to the shoulder of the road. And for some of his rides, Walker donned a long, blonde wig to make him appear to be a woman from behind, to see whether drivers behave differently based on the perceived gender of a cyclist.
In his travels, Walker was passed by over 2,000 vehicles, and sensors mounted to his bike measured how closely those vehicles passed. Those distances ranged from over 11 feet to “less than 0.” Walker was hit by two different trucks while collecting data for his study. (He was wearing a helmet both times.)
Overall, Walker found that drivers of large, commercial vehicles like delivery trucks give cyclists less room than drivers of private vehicles. When Walker wore a helmet, vehicles got 8.5 cm (3.3 inches) closer while passing him, on average, than when he was helmetless. Cars also gave him more room while he was wearing the wig. And contrary to popular beliefs among cyclists, riding towards the center of a lane wasn’t safer; the further he rode into the lane, the closer drivers got as they passed him.
Walker’s study made headlines when it came out, and ever since, it’s been trotted out in heated discussions about helmet use. While Walker himself said his results were not an endorsement to bike sans helmet, it’s easy for cyclists to draw that conclusion anyway, or to at least justify to themselves that biking without a helmet might not be so bad. For instance, in a recent essay, Peter Flax, former editor-in-chief of Bicycling magazine, said he no longer wears a helmet, and cites Walker’s study, which Flax says “substantiated my anecdotal observations.”
“Laypersons have used Walker’s findings to justify the removal of mandatory helmet laws,” write biostatisticians Jake Olivier and Scott R. Walter in a 2013 paper reanalyzing Walker’s results. Their analysis confirmed that when Walker rode further into the driving lane, vehicles passed closer, and that drivers of large trucks left cyclists less room.
However, they were critical of Walker’s sampling method, and suggest an alternate analysis of Walker’s data on helmet wearing. In Walker’s data, vehicles’ average passing distance was more than a meter (around 3 feet), the typical recommended passing distance in the US and UK, and so Olivier and Walter looked only at cases of “close passing,” in which vehicles left less than a meter when passing Walker. In those cases, helmet wearing doesn’t make a difference in driver behavior, and, they conclude that helmet-wearing must not encourage more dangerous driving.
And to continue the back-and-forth, Walker and colleague Dorothy Robinson wrote a rebuttal to Olivier and Walter in an upcoming issue of the journal Accident Analysis and Prevention, claiming the one-meter cut-off Olivier and Walter chose for their analysis is arbitrary: passing at a meter is not necessarily safe, and if cyclists’ helmets caused any change in driver behavior, even if it was a difference of 2.5 meters versus two meters, that might still be noteworthy. He also refutes claims they made about the statistical power of his results, and levies the same claim about their analysis.
When I talked to cyclist friends of mine about the debate, their first question was: why are these researchers going back and forth on a dataset collected over a decade ago?
Walker’s dataset is one of the few that comes from a controlled study. Most research on cyclists and safety is based on incidents from a database, since that info is already available, rather than putting participants in danger to collect new data. But not all dangerous behavior results in an accident that would be logged in that a city or state accident database, and other details of the incident, like the positions of the vehicle and bike, or whether the cyclist was wearing a helmet, aren’t available. It’s incredibly valuable to collect data on what happens when you manipulate specific variables, as Walker did, like the distance cyclists ride from the curb, and the institutional review boards that approve studies are likely to find ethical and logistical issues with doing anything that puts participants in harm’s way.
In any case, Walker is ready to put this debate to bed. “I’m really hoping that paper is the last of any back-and-forth. It spells out pretty much everything I’d want to say on this subject,” he says, pointing to the paper’s conclusion: that we should be focusing on safer streets, not helmets. “Helmets are not an effective solution to being hit by motor vehicles—the only sensible solution is to prevent collisions in the first place.”