How dangerous are cell phones when driving?

Maybe not as much as you think.  Here is the abstract from a new paper (AEA gate) by Saurabh Bhargava and Vikram S. Pathania:

We investigate the causal link between driver cell phone use and crash rates by exploiting a natural experiment induced by the 9pm price discontinuity that characterizes a majority of recent cellular plans. We first document a 7.2 percent jump in driver call likelihood at the 9pm threshold. Using a prior period as a comparison, we next document no corresponding change in the relative crash rate. Our estimates imply an upper bound in the crash risk odds ratio of 3.0, which rejects the 4.3 asserted by Redelmeier and Tibshirani (1997). Additional panel analyses of cell phone ownership and cellular bans confirm our result.

Here is another way to put the puzzle:

Cell phone ownership (i.e., cellular subscribers/population) has grown sharply since 1988, average use per subscriber has risen from 140 to 740 minutes a month since 1993, and surveys indicate that as many as 81 percent of cellular owners use their phones while driving—yet aggregate crash rates have fallen substantially over this period.

I don’t want you to start or continue this practice, as numerous other studies do find significant risk.  Still, maybe this matter is not quite as settled as many people think.

Here is a non-gated version of the paper (pdf), and the official ungated version is here.


Didn't read paper or full post, doesn't seem like a good natural experiment though at first glance.

Traffic is different between 8 pm and 9 pm than it is from 9 pm to 10 pm. Since there's likely less traffic between 8 and 9 pm, I'd expect fewer crashes both on aggregate and average (people do often crash into another, which is easier to do when roads are crowded).

The fact that this decrease isn't observed, could be evidence that cell phones do indeed lead to more crashes.

They do compare with pre-1998 data (no cellphones); see Fig 3. That should be a control.

a few other things have changed since 1998, improvements in motor vehicles, braking, steering, etc could all swamp the effects of driver distractions. I suspect texting or talking on a cell phone has simply crowded out other distracting side shows in the car like fiddling with the radio, applying makeup, and yelling at the kids in the backseat. I doubt it detracted much from the attention already given to driving the car and certainly not enough to overcome all the positive safety trends. would be interesting to know if cause of accident (but not total accidents) shifts around at 9pm if that data is collected accurately.

still people are poor multi-taskers and few are better drivers when on the phone. we are the danger, not our toys.

Not only is the traffic different, but because it is dark, the calculus of risk awareness is different during this period as well. It's dark, this signals you have to be attentive; you might be more attentive to the additional cell risk of talking than you would be during other periods. There are also fewer cars to interact with, so even if you are more attentive than during other periods, the type of risk is different for an error--you run off the road, but you don't hit a vehicle and there is no report because you get yourself out of the ditch. Also, the composition of drivers is different and not representative of the mix during the morning rush hour.

"doesn’t seem like a good natural experiment"

There is no such thing as a bad natural experiment.

You choose to get in your car at 9pm because cell phone rates drop at 9pm?

How many people get in their car at 8:45 for a 30 minute commute and think, "I need to wait 15 minutes to call the wife and kids and tell the kids to go to bed at 9 and tell them I'll say goodnight when they are in bed soon". It wouldn't matter because they would have been calling him asking if he'd left yet.

I'm very aware of the reduced rate incentives being one of the earliest customers to drive up AT&Ts profits when AT&T was required by the Feds to reduce its profits and it tried to comply by reducing rates at 7PM and 11PM. I went from a few dollars a month on long distance to a steady $10 back when $10 was real money - 8 hours at minimum wage. After the surge in profits from mandated lower rates, the 11PM rates were really discounted and I and a friend talked once a month for an hour to hour and a half, and stopped writing letters, at a cost max of $1 for more content.

But my point is I planned the calls to maximize the utility of the lower rate. I did not wait until 11PM just to make a 1 minute call to tell my parents I got home safely and thanks for everything, If I got home at 6:45, I called, and did not wait until 7pm.

Cops driving police patrol cars should be the worst case group for "distracted driving".

But nobody studies that sub-group for distracted-driving accident rates.

Modern police cars are loaded with interactive radios, computers, radar/laser, etc. ... and yes -- cell phones.

How is it that cops can juggle all that technology/communications gear, while simultaneously driving and visually scanning for law-breakers ??

Of course, cops are somehow legally exempt from all the driving/cell-phone laws.

Not to mention, have you watched these guys drive? They break every traffic rule.

Oh yeah, they are highly trained...

Great points Donlon. Never thought about these aspects of police patrol car activities. Great exception to the rule.

When the police are actively using communications devices under conditions of stress....they are also using their sirens to keep people away. Not comparable.

Last week I had a cop flip his siren and lights on to get past a few cars. Then he turned them off.

I've seen that sort of thing many times, multiple pl;aces I have lived.

Andrew, I'd like to know where you live that this was notable to you. Living in Vancouver and Las Vegas, I've seen this behaviour repeatedly.

Didn't read your full comment.

How about the possibility that the sort of air-heads who get absorbed in phone conversations while driving are the sort of air-heads who are capable of finding plenty of other opportunities to crash?

Or, the government runs a terrible quality system and has to continually over-market plausible reasons to blame operator error.

The quality system for autos? Crash fatality rates have been going down for decades. Whether that is companies engineering better cars on their own, more careful driving or government oversight I'm not sure. You want more government intervention here?

The roads, Jan.

btw, there are other axes aside from "more."

Better roads (many problem areas on roads have been fixed), safer vehicles, less tolerance for DUI. Those three probably account for most of the reduction. If we are talking fatality rates, then better trauma care also plays a role.

This paper doesn't seem to control for confounding at all. They acknowledge in their discussion three different possible hidden variables that would skew the results, but so far as I can tell they didn't even try to test for them in their regression. The statistics here aren't rigorous enough to be persuasive.

My worry is they are looking at response variables (crash rates) on which the direct impact of cellphone use is swamped out: NHTSA stats show that of all crashes ~15% are distraction related. Of those only ~10% are cellphone related.

It's like concluding off-piste skiing is not dangerous by looking at total accidental death rates (from all causes) and treating winter as a discontinuity design.

Cellphone-distracted-driving can indeed be incredibly dangerous to the driver yet not have any discernible impact on total accident rates in a study like this, just because of the large number of other factors that can cause accidents (and the general rare nature of accidents in the first place).

Very nice. OTOH, what we care about is the overall situation. With GPS, the ability to coordinate, and yes, wise use at stop lights to distract you from monotony provided by cell phones improve the driving experience.

You wouldn't think, from the hype, that cell phone distraction could be so easily swamped out. So, in a world of scarce attention and enforcement resources, what if we did more post hoc enforcement?

To chime in, alcohol accounts for only a small percentage of auto fatalities--less than 20%, I believe--yet no one disagrees that driving drunk is a bad idea that increases your chance of an accident. It is quite possible for cell phones to be dangerous while making only a small contribution to accident rates because the number of potential risk factors in driving is so high.

Actually, people do disagree, particularly with the exaggerated blood-alcohol level limits.

That's not disagreement. That's just haggling over the price.

I hope their study-city didn't have a railway line....

Oh okay. If this paper is flawed, then we'll just fall back on the mountains of literature justifying the government overreaction to cell phone use.

This is a symptom of a disease - the one course of action unavailable to a politician is 'do nothing.'

Reason doesn't matter in politics. As much as I agree with you, you are wasting oxygen...or electrons.

Nonsense, using a cell phone radiates magical rays that destroy engines and induce madness in drivers. This is why you can't use them on airplanes either. Will no one rid us of this scourge?

The anecdotal evidence is that when driving it often seems clear when someone ahead of you is on the phone. They are just not paying attention, or driving in a way that is scary for many of us.

Sampling bias? Confirmation bias?

Sure, perhaps so. But it's only a problem when it's a problem. The guy who can drive 65 mph and stay in his lane while texting is apparently not a problem. So as I say later, we must decide whether we should pull over that guy and issue a warning or citation or wait until we have evidence of impairment actually contributing to an accident. It's a difficult balance between prevention, enforcement, and culpability. The problem is that the cop/jury does not directly observe 99% of the available information.

A nicked wiper blade might have been the proximate cause of the accident, not the cell phone. But as a prosecutor I can't take a nicked wiper blade to a jury and I can't ignore or rule out the cell phone. And I couldn't believe the story of a defendant who blamed a nicked wiper blade instead of the cell phone when I have phone records. Even if I believed the story of the nicked wiper blade, I could amend the charge to say "negligently failed to replace a damaged wiper blade," but the jury and judge would laugh at that. The fact of the matter is that when there is a wrongful death, I will use the best evidence I have to make a case. If I don't have good evidence, I will accept a plea deal. Then I will get drunk and console myself that justice was done; this is why I left law.

Sad side effects of an adversarial law system? The extent to which the US system is using Plea bargains is distressing though.

I like the design, but I agree with the sentiment that it might not generalize to other times of day (e.g. more congested times with a higher density of distracted drivers; accidents may be nonlinear in the density of distracted drivers).

Perhaps people simply get better at driving and talking on the phone with experience. When it was a new experience for everyone to talk and drive, people were worse drivers, but now most who talk and drive are experienced at it. So a study 15 years ago would be expected to have different results than a more recent one.

You're right. The best drivers are the elderly.

The interesting thing here is the nexus between theory and reality.and policy. I think that there is no dispute that cell phones distract drivers. The public policy question is whether it is enough of a distraction to merit a law and subsequent enforcement. And if so, are we overlooking other behaviors behind the wheel that are as much if not more a source of distraction, e.g. eating while driving, tired driving, aged driving, cell phone with headset?

I prosecuted a lot of vehicular manslaughter cases, and often we searched for some evidence of driver impairment or coincident infractions such as speeding. Cell phones were practically non-existent back then. In some respects I think it was unfair to the driver who had a couple of drinks who was at fault - the drinking might not have been the proximate cause or contributing factor of the accident, but it was a powerful tool of persuasion for a jury/plea deal. Two all-beef patties, special sauce, lettuce, cheese, pickles, onions and sesame seed buns splattered on the windshield, steering wheel, drivers lap, and pedals should have been just as damning, but it wasn't. I lost the first case of that I tried, and I wouldn't have tried it again.

Hey all, I'm still waiting for Steve Landsburg (author extraordinaire of the "armchair economist") to show how banning cell phones while driving might actually increase the rate of accidents!

This reminds me of that "file-sharing doesn't decrease music sales paper."

They take an extremely low-powered test and try to reject a null of zero that no one believed to start with.

Sorry, not buying this one. In fact, I think the opposite is likely the case.

My guess is that future research will find cell phone use in cars is responsible for even more accidents than we think.

You can't compare rates between 1988 & now & think cell phones are the only factor. Did they stop to think road safety has increased since then? The only way to properly (scientifically) study this is to just isolate the cellphone variable.

This doesn't really surprise me, as airliners have had text messaging systems since the 1970s without problems. However, pilots are trained to keep their scan going. By the same token, it is important to do so while driving and only type a character every few seconds after reacting to changing driving conditions.

Anything that distracts a driver is dangerous, especially cell phones.

Comments for this post are closed