Do bicycle helmet laws work? The best answer is that it depends typically on the context of the question. Basic physics, experience, and common sense illustrates that helmets can save the lives of those who are about to sustain an imminent head injury. However, this is emphatically not the correct question to ask when exploring whether to implement such policies. In fact, discerning the correct question can be immensely complicated.
When considering a law that requires adult cyclists to wear helmets, the problem can become difficult to conceptualize, measure, and analyze for a number of reasons.
Recently, a Canadian study published in the British Medical Journal analyzed over 66,000 hospital admissions for cycling-related injuries in Canada since the early 1990s. This study measured changes in head injury rates both before and after helmet law interventions in each province and revealed that helmet laws have no effect.
How could this be?
In an editorial published shortly after the British Medical Journal study, Ben Goldacre speculated on the complexity of this issue, suggesting that people who choose to wear helmets are fundamentally different from those who choose to abstain in the first place. This selection bias could mean that those who wear helmets in the absence of helmet laws are less likely to sustain an injury regardless of whether or not they are wearing protection. Furthermore, Goldacre argues that it is important to consider the effect that helmets have on future behavior. Cyclists may be less cautious while wearing head protection. Not to mention that drivers give less road clearance to cyclists wearing helmets. Both of these effects create a moral hazard that controlled experiments cannot typically capture.
Helmet laws may also reduce the level of cycling, an otherwise healthy activity. This is an unintended consequence due to a dislike for helmets or the financial costs associated with wearing one, however, it is nonetheless difficult to measure in any meaningful sense.
These studies can also have discrepancies. This particular study, for instance, uses emergency room admissions as a measure of injuries, which does not capture information about cyclists who do not get into accidents. Instead, it compares cycling head injuries to other cycling injuries, suggesting that the data is conditional on an accident actually occurring. As Goldacre writes, “analyses … therefore assume that wearing a helmet does not change overall accident risk,” which, of course, it does.
To summarize, even simple questions produce extraordinarily complicated circumstances when expanded beyond a single case study with rigid assumptions. While it is likely perceptive for medical associations and governments to keep cycling helmet recommendations, if not laws, in place, it appears that this stance may not necessarily be evidence-based. Personally, I will continue wearing my helmet when I cycle.
These reservations also offer general lessons for public policy research. Namely, just because a given policy appears beneficial at the individual level does not mean necessarily that it will produce similar results at the aggregate level. Taxation, for instance, is a useful example. Holding the behavior of an individual constant, raising the tax rate will result in this individual paying more taxes. At the macroeconomic level, however, this is not always the case (as is demonstrated by the Laffer Curve). Sometimes, higher taxes persuade businesses to produce less. In this case, tax revenues are lower.
Here, the lesson is that public policy has immediate and long run consequences. In essence, policymakers set expectations about the future, upon which people base their behavior. This behavior, however, sometimes creates unintended, and, often, immeasurable consequences. It is likely, therefore, that this theme holds across a broad range of public policy and lawmakers should exercise caution when attempting to change behavior.
Mike Craig is a 2013-2014 Atlantic Institute for Market Studies’ Student Fellow. The views expressed are the opinion of the author and not necessarily the Institute