It was an incredible number. To think that texting while driving increased the risk of a car crash by a factor of 23! But when the Virginia Tech Transportation Institute issued its bombshell study in 2009, that’s what it seemed to be saying.
After The New York Times published a story about the study, the scary statistic at its center was everywhere, and soon “23” was making regular appearances not just in news articles, but in a variety of informational materials designed to discourage people from texting and driving. The FCC used it on its fact page about texting and driving. AT&T used it as part of its “It Can Wait” campaign, including it in a Web video, an infographic, and a site designed to simulate the effects of texting while driving. The National Highway Traffic Safety Administration, part of the Department of Transportation, used it as the lead statistic on the front of Distraction.gov, the US government’s official website about the dangers of distracted driving.
Then something happened. In April 2013, the Virginia Tech Transportation Institute published a new study about texting and driving. This one was different from the 2009 one in a few important ways. First, the researchers had richer data—not just video of their test subjects driving, but also phone records that showed exactly when they were texting or talking. Second, they used a different population of drivers. Where the 2009 study looked at people operating commercial trucks, the new one looked at people driving regular cars.
The study contained some terrifying findings—including that the total amount of time the average person spends not looking at the road when sending a text is almost 25 seconds. But there was another noteworthy result, too. According to the data, texting while driving didn’t increase the likelihood of a crash by a factor of 23, but rather increased the likelihood of a “safety critical event”—meaning a crash, a near crash, or a “crash-relevant conflict” (e.g., a driver slamming the brakes to avoid a crash)—by a factor of just...two.
Twice as dangerous is still extremely dangerous, obviously. But it’s not 23 times as dangerous. The authors of the two studies had some theories about what might account for the massive discrepancy. Maybe it was that cellphones had changed so much in the intervening years. Maybe people had learned to be more careful about when they texted—choosing stoplights more and highway on-ramps less. Or maybe 18-wheelers were just a lot harder to drive than normal cars.
Regardless, VTTI’s new data created something of a conundrum for everyone who had been engaged in the effort to convince people not to text and drive—including the US Department of Transportation, which funded the study. Their solution, at least at first, was mostly to ignore it, and stick with the older, more alarming statistics: Though DOT issued a press release about the new study, six months later anyone visiting Distraction.gov was still being told drivers who were texting were “23x more likely to crash.” A similar warning continued to appear on StopTextsStopWrecks.org, another government-funded site, while AT&T executives involved in the “It Can Wait” continued to cite the figure.
In early October, while researching a story about the psychology behind texting and driving, I found the two VTTI studies and noticed that the more recent one didn’t seem to be getting much attention. When I contacted NHTSA on Oct. 21 and asked why the now-outdated finding from the truck study was still being trumpeted on Distraction.gov, a spokesman declined to comment. But a few days later, the website suddenly changed, with the “23” in “23x” transformed into a “2.”
In a written statement, the NHTSA spokesman said the update had been in the pipeline before the Globe’s inquiry, but did not explain why it had taken half a year to make the change. About a week later, after I pointed out that “23” was still being used on NHTSA’s other site, StopTextsStopWrecks.org, that too was quickly changed.
AT&T has started updating its materials as well, editing the Web video that once proclaimed a crash was 23 times more likely if a driver was texting to say that it was simply “much more likely.” But the FCC page still cites the old number. It also appears on websites like textinganddrivingsafety.com, textfreedriving.org, and focusdriven.org, and it continues to get invoked in news reports.
The reasons why “23” has persisted are obvious enough. It’s a startling figure that perfectly underscores what public safety officials are trying to tell you: Don’t text and drive. That admonition is well founded—the new data attest to that—and getting people to accept it could save lives. But the difficulty that advocates have had in accommodating the new figure points to an uncomfortable reality about the use of statistics as tools of persuasion: Even when it comes to promoting a self-evidently worthy cause, the facts don’t always stay on message.