The other day I was talking on the phone, handsfree, while driving a car. I was trying to avoid rush hour traffic, traveling on some unfamiliar side streets. Out of nowhere I hear a loud horn, followed by some curse words. I look around: I’m the only one here. I must be the motherfucking asshole.

The other driver had a point: I’d just run a stop sign. It wasn’t a close call – no screeching brakes were required – but it could have been ugly. I got off the phone and drove the rest of the way home in chastened silence.

This little episode got me thinking: how bad is it to talk on the phone when driving? Does it really increase the risk of an accident? These are psychological questions with vast practical implications, as more than 80 percent of drivers admit to cellphone use while behind the wheel. According to a 2007 survey by the National Highway Traffic Safety Administration (NHTSA), roughly 11 percent of drivers are using a cellphone on the road at any given moment.

At first glance, the answer seems obvious: driving requires focus. We should give our full attention to these 4000 pound machines hurtling down the road at high speed. Such a response is rooted in everyday experience – 45 percent of drivers say they’ve been hit or almost hit by another driver chatting on the phone – and numerous studies, both in the field and the lab. Perhaps the most cited work comes from David Strayer at the University of Utah. In one study, Strayer and colleagues observed more than 1700 drivers as they approached a four way stop sign on a residential street. The scientists noted whether or not the drivers were talking on the phone. They also noted whether or not they obeyed all traffic laws. The differences were not subtle: those drivers on the phone were roughly 10 times more likely to commit a stopping violation. (Handsfree devices provided "no safety advantage.") In other words, I’m not the only one.

A similar pattern emerged from realistic simulations of driving conducted in the Strayer lab. In one study, the scientists compared the performance of 40 drivers under three conditions: driving without distraction, talking on the phone while driving and driving with a blood alcohol concentration at the legal limit. Based on the detailed driver data, Strayer, Jason Watson and Frank Drews concluded that “the impairments associated with cell-phone drivers may be as great as those commonly observed with intoxicated drivers.”

We could stop here. And many do: Strayer’s research is frequently cited in discussions around cell phone driving laws. It seems to provide clear evidence that phone conversations significantly impair driver performance. The gadgets should probably be banned.

There is, however, an obvious rebuttal to this research, at least if you stop and think about it. If talking on a cell phone while driving is really as dangerous as being drunk, then why haven’t crash rates surged along with cell phone usage? Pull up to a red light in my city and half the drivers are talking to their dashboards, or maybe into a dangling white cord. (The other half are texting.) And yet, the roads are not littered with wrecked cars.

This conundrum led me to a recent study by Saurabh Bhargava at Carnegie Mellon and Vikram Pathania at the London School of Economics. The scientists begin their paper with a telling chart, which documents the seemingly inverse relationship between cell phone ownership and crash rates:

What the chart doesn’t show, of course, is that the rise of cell phones has also paralleled the rise of air bags, anti-lock brakes, traction control, automated braking systems and other car technologies that, at least in theory, have also made driving a less fraught activity. It’s entirely possible (likely, even) that the rate of car crashes would have plummeted even more if it weren’t for cell phones.

But Bhargava and Pathania are only getting warmed up. The bulk of their study is taken up with an attempt to establish a causal relationship between “cell phone use and crashes in the field.” To do this, they exploit a natural experiment triggered by an old quirk in the cost of a cell phone call. I’d forgotten about these phone plans, but many cellular companies used to divide weekdays into “peak” and “off peak” hours. For the most part, calls placed during the “off peak” hours did not count towards the allotted minutes included in the plan. Not surprisingly, this led to a 7.2 percent spike in calls placed by drivers just after the start of the “off peak” period, which was typically 9 PM.

Bhargava and Pathania then compared the crash rate from 8 to 9 PM – the hour before the surge in cell phone use – to the 9 to 10 PM hour. Did the rise in calling correlate with a rise in crashes? The answer: Not really. (They also ran control analyses on weekends, when the peak rules were not in effect, and on crash rates from 1995 to 1998, before phone plans had peak restrictions.) While this null finding does not, by any means, rule out a possible increase in crash risk caused by cell phone conversations, it does suggest that, as Bhargava and Parathi note, “cellular use is not analogous to drunk driving as some policymakers and academics have suggested.” In particular, it suggests that prior estimates of the hazards posed by cell phone conversations in the car were significantly too high.

The finding comes with numerous caveats. For starters, Bhargava and Pathania aren’t able to ensure that their cell phone data comes from drivers on the phone, and not chatty passengers. It’s also possible that the lack of increased crash risk is unique to either the region in question (outside the downtown area of a large city in California) or the time (between 9 and 10 PM.) Perhaps the drivers most vulnerable to the hazards of cell phone use are less likely to be driving at night, or maybe the lack of cars on the road at off peak hours makes distraction less dangerous. It’s also possible that the 9 to 10 PM sample is picking up an “unrepresentative” set of cell phone calls. Perhaps these calls are less urgent, or somehow require less attention than calls made earlier in the day.

To mitigate some of these concerns, the economists conducted two additional statistical analyses. First, they looked for correlations between local variation in cell phone ownership – some regions adopted mobile technology at a faster pace – and changes in crash rate. No correlations were found. Secondly, they looked to see if there had been any impact on the number of fatal crashes in three states (New York. Connecticut and New Jersey) and two large cities (Chicago and Washington D.C.) that had banned handheld cell phone usage. They found no short-term or long-term impact triggered by these changes in the law. More recent studies of other states have come to similar conclusions.

This is the part of the blog where I tell you the neat takeaway, how this new study displaces all of our old knowledge. But it’s not that simple, is it? While Bhargava and Pathania estimate a minimal increase in crash risk, Strayer’s experimental research estimates a 20 to 30 percent increase. Police annotations on accidents put the increased risk around 1 percent, while a 1997 analysis of individual crash records puts the increased risk closer to 33 percent. Meanwhile, a paper looking at calls made using the hands free OnStar system found no evidence of an increased crash risk (at least for crashes that involved the deployment of an airbag), and a thorough field study by the NHTSA came to similar conclusions, but an analysis by Jed Kolko found that increased cell phone ownership was associated with an 11 percent increase in fatal crashes in bad weather.

In short, the effect is either huge or negligible. Talking on the phone either makes you drive like a drunkard or has virtually no impact on your performance. Banning all cell phone use in the car will either save no one or it will save thousands of lives every year. In an email, Bhargava put it this way: "One way I think to reasonably reconcile the lab findings and field findings like ours is that cell phones could distract, as lab studies sensibly suggest, but on the road, people may compensate for such distraction by using their phones in particularly safe situations, or adjusting their driving when on the phone. It’s also possible that cell phone use simply substitutes for other risk-taking behaviors such as talking to a passenger or playing with the radio.” This substitution effect might be especially potent in the era of smartphones - it’s almost certainly safer to talk on the phone than to send a text while driving.

Social science is hard. If nothing else, the struggle to pin down the riskiness of talking on the phone in the car is a reminder of the difficulties or parsing causes in a world overstuffed with variables. It’s been eighteen years since the first prominent study of cell phone use and driving and yet we’re still not that close to a definitive answer.

As for me? I’m fairly convinced by these field studies that posit a lower risk ratio. Nevertheless, the elegant statistics carry less emotional weight than my own n of 1 experience. I have no idea if the conversation caused my traffic violation, or if it was triggered instead by one of the many other factors at play. But at the time it felt like I should just shut up and drive. Such feelings carry little validity – people are notoriously terrible at introspecting on their mental limitations – but it’s what comes to mind first when my phone rings and I’m on the road.

Suarabh Bhargava and Vikram Pathania. “Driving Under the (Cellular) Influence.” American Economic Journal: Economic Policy, Vol. 5, No. 3, p. 92-125, 2013.