In August 2010, 19-year-old Daniel Schatz ran his pickup into the back of a tractor-trailer truck on Interstate 44 near Gray Summit, Mo. The collision set off a chain-reaction accident with two school buses, killing Schatz and a 15-year-old bus passenger, Jessica Brinker. Investigators later discovered that Schatz had been exchanging text messages on his cell phone while he drove, sending six and receiving five in the 11 minutes prior to the crash.[1]

The wreck underscores the danger of driving while distracted. Some people would argue it also shows the need for laws prohibiting cell phone use while driving. Last month, the National Highway Transportation Safety Board cited the Missouri crash in a recommendation that states ban all driver use of cell phones, even if they’re hands-free.[2]

But at the time of the accident, it already was illegal for Schatz to be texting. In August 2009, Missouri implemented a ban on under-21s from texting while driving.[3] The law did not prevent the tragedy. Worse, empirical evidence suggests the law increased the chances of it happening—more on that below. This underscores an important question for public policy: Do we want laws that have good effects, or laws that just appear to be good?

There’s no doubt that distracted drivers are a safety hazard. Studies of people using cell phones while in driving simulators have repeatedly shown an increased risk. One such study by University of Utah researchers found that both handheld and hands-free phone use increase the risk of accident to roughly the same level as drunk driving.[4]

However, those simulator results do not translate neatly to the real world. Virginia Tech’s Transportation Institute has conducted several “naturalistic” studies that use cameras to watch the behavior of real drivers in the moments before an accident. Those studies repeatedly found that cell phone use did correlate with increased risk of accident, but most of the increased risk occurred when drivers were hand-dialing or texting, not when talking.[5] More interesting, a 2007 statistical analysis by two then–graduate students at the University of California, Berkeley found no statistical correlation between periods of increased cell phone use and vehicle accident rates.[6]

More troubling is the results of a study by the Highway Loss Data Institute, a research group funded by the auto insurance industry, which has strong incentive to find ways to lower accident rates. The study examined collision insurance claim rates in four states that adopted bans on driver cell phone use, along with the rates of neighboring states without bans. The institute found that, in three of those states, the presence of a ban increased the rate of collision insurance claims.[7] Advanced statistical techniques failed to find evidence that the ban had a beneficial effect in any of the four states, either for all drivers or for drivers under age 25.

Why would a cell phone ban increase the risk of collision? Two reasons come to mind. First, cell phones can help to prevent some accidents. The running-late driver can call ahead rather than speed to his destination. The lost driver can call for directions rather than drive frantically. And drivers can receive calls alerting them to roadway hazards. Those benefits may be lost under a cell phone ban. Second, drivers in states with bans may remain determined to use their devices—but would try to hide that use, increasing their level of distraction and risk of accident.

Regardless of the reason, the empirical evidence indicates that, at best, cell phone bans do not improve roadway safety and, at worst, they produce more accidents. Policymakers who want to help the public should oppose the bans and work to repeal ones already in place. But how to explain such opposition when, on the surface, a cell phone ban seems like such a good idea?

It’s much easier to back laws that appear to do good, even if those laws really put the public at greater risk. That explains NHTSA’s recommendation, and why more and more jurisdictions are adopting cell phone bans.