Growing up, I was forced to wear a bike helmet by my parents and the law. In Pennsylvania, kids under the age of twelve must wear helmets. The reasoning behind wearing a helmet is to reduce risk of head injury in the event of an accident, but like many people, however, I have a distaste for helmets because it ruins my hair. I rarely ride my bike around campus
, and I probably will not commute via
bike to my future job, but I do see the increasing trend in metropolitan areas. Because of this trend, I wonder if regulators will implement mandatory helmet laws and if those laws would improve or reduce safety.
Last year a leading neurosurgeon, Dr. Henry Marsh, made controversial comments with regard to bicycle helmets. He stated that people using helmets are wasting their time, and that the helmets are “too flimsy” to do any help. Dr. Marsh uses anecdotal evidence to back his claim by referring to patients of his whose helmets did not protect them. He goes on to mention the fact that he has been riding for about 40 years and never has only been knocked off his bike once and without major injury.
Dr. Marsh’s comments are anecdotal observations and do not have any observational or experimental studies in support of his observations. However, because of his medical prestige, his opinion is valued more by the public and newspaper publications. This is all too common in the media. A report is made about an opinion of a doctor or a scholar, and the general public reads it at face value and believes it.
Although Dr. Marsh’s were not convincing to me, there was a study conducted by Ian Walker, a professor at the University of Bath, which found that bicyclists wearing helmets encountered riskier automotive maneuvers in their vicinity than non-helmet wearing bicyclists. Walker mounted several ultrasonic sensors on himself and his bike while he was riding. Over the course of two months, Walker rode his bike with and without his helmet. During that time span, 2,355 vehicles overtook him, and he found that on average, vehicles passed him about 3.5 inches closer when he was wearing a helmet. Walker hypothesizes that drivers are more cautious around people not wearing helmets. Walker’s study seems well down. It has a large sample size. He does, however, ride on multiple different street types (one-way, two way, those with bike lanes, etc.) which might interfere with testing the alternative hypothesis that wearing bike helmets alone cause cars to pass closer. Walker also measured curb distance he was riding, the type of car passing (i.e. truck, bus, car, etc.), the color of the car, the time of day, etc. All these measurements may lead to a Texas sharpshooter problem. A correlation is bound to arise with all of these measurements.
Upon re-analysis of this study however, Professor Jake Olivier, a statistician, disputes Walker’s findings. Olivier first points out that the average passing of a vehicle was over a
meter in length. Therefore, Olivier re-analyzed the data according to “the recommended one metre rule into close (less than 1 m) and far (greater than or equal to 1 m) [passing] distances.” After crunching the numbers Olivier found that there is no significant difference to passing buffer distance given to Walker when he was or was not wearing a helmet. This re-analyses gives important insight to this theory. It shows that the deviation between average passing space by c
ars when Walker was and was not wearing a helmet is not significantly different.
Furthermore, a meta-analysis of bike injuries published in Accident Analysis and Prevention found that helmets significantly reduce risk of injury if an accident occurs. The analysis found risk “reduction estimates of at least 45% for head injury, 33% for brain injury, 27% for facial injury and 29% for fatal injury.” The meta-analysis was conducted the best it could be. There is now experimentally designed studies with regard to injury, and many of the studies included evaluated children’s injuries while my main inquiry was with regard to adults. A file drawer problem is suggested because most of the studies found helmets had a positive effect of reducing risk of injury. However, of the 63 studies, 7 did not find a positive effe
ct, which suggest that a file drawer problem is not present because 9% of the studies failed to reject the null hypothesis.
Overall however, I am concerned about the data in the individual studies. Is it possible that risk taking is a confounding variable? People who take risks do not wear helmets and also take risks while riding. Or, do people who wear helmets have more confidence, herefore, take more risks. As of now, the evidence on the topic is not very convincing either way. I would be hard to conduct any blind experiment because the cyclist and other auto traffic know if he/she is wearing a helmet. I guess the best study would be observational of bike crashes, measuring many third variables and limiting the study to only adult riders in metropolitan areas.