Phones Top Crash Risk Factors for Cars

Man driving and looking at smartphone
Photo: Frederic Cirou/Getty Images

Drivers constantly reaching for their phones may be the single factor most responsible for car crash increases in recent years. Such distracted driving behavior was caught repeatedly on video in the largest study of car crash risk ever conducted using real-world driving data.

The damning data came from a US $70-million study designed by the Virginia Tech Transportation Institute and funded by the U.S. Transportation Research Board. That research collected more than 55 million kilometers of real-world driving data from cars rigged with video cameras and other sensors. Such a big data study involving more than 3,500 drivers provided the first large-scale opportunity to study driver behaviors contributing to car crashes. For example, researchers found that driver distractions doubled the overall crash risk and occurred during 52 percent of observed driving time.

“The overall level of distraction suprised us quite a bit,” says Tom Dingus, director of the Virginia Tech Transportation Institute in Blacksburg, Va. “Over half the time, the drivers are doing something other than driving, such as messing with the radio, messing with their cell phones, looking around at something not related to driving, or interacting with passengers.”

Dingus and his colleagues analyzed the crash risk factors based on the second Strategic Highway Research Program Naturalistic Driving Study (SHRP 2 NDS), a database that went live in 2015. Driving data from that study came to a total of 2 petabytes (a petabyte is 1,000 terabytes) and was collected between 2011 and 2013 for the Virginia Tech analysis.

Such a huge collection of driving data has, for the first time, allowed Dingus and his colleagues to calculate the crash risks for “model driving” behaviors: when drivers were sober, alert, and paying attention to the road. The model driving scenario became the baseline for calculating the increased crash risks related to distracted driving, driving under the influence of alcohol or drugs, and other factors. Dingus was lead author on the resulting research detailed in the most recent issue of the journal Proceedings of the National Academy of Sciences (PNAS). “It’s really like a missing piece of the puzzle,” Dingus says.

Handheld Cellphone Use is Both Common and Dangerous
Odds Ratio Prevalance (percent)
Drugs/alcohol 35.9 0.08
Drowsiness/fatigue 3.4 1.57
Emotion 9.8 0.22
In-vehicle device 2.5 3.53
Cellphone handheld use 3.6 6.4
Interactions with adult/teen passenger 1.4 14.58
Reading/writing (including on tablets) 9.9 0.09
Eating 1.8 1.9
Drinking (non-alcohol) 1.8 1.22
Personal hygiene 1.4 1.69
Child in rear seat 0.5 0.8

Source: “Driver crash risk factors and prevalence evaluation using naturalistic driving data,” PNAS.

Dingus and his colleagues found that driver-related factors such as error, impairment, fatigue and distraction were present in almost 90 percent of all car crashes recorded during the huge study period. That represents a big shift from not so long ago when roadways were worse and cars may have suffered more catastrophic failures that led to car crashes.Previous understanding of car crash risk has relied on more indirect data sources with missing information. One older method was to look at crash investigations filed by police after the incident had already occurred. That method relied heavily upon the incomplete and sometimes incorrect memories of drivers or passengers involved in car crashes. A second method involves observing the driving behaviors of volunteers during driving simulations. But such driving simulations could not truly replicate the conditions of drivers doing their usual commutes in the real world. 

The U.S. fatal crash rate has been falling for a long time. But that may mostly reflect the rise of modern vehicle safety features such as automatic emergency braking and better crashworthiness, Dingus says. On the other hand, driver behavior may not have necessarily gotten any better. And in the case of phone use, certain risky driver behaviors have only become more common.

Holding the cell phone to talk or text while driving led to a 3.6 times greater crash risk and occurred during more than six percent of the observed driving time. Fiddling with the vehicle devices such as the radio or climate control led to a 2.5 times greater crash risk and occurred between 3 and 4 percent of observed driving time.

Certain non-technological factors had an even greater crash risk, but did not occur as much as phone use. Driving under the influence of drugs or alcohol increased the crash risk by about 36 times, but only happened during 0.08 percent of the observed driving time. Driving while angry, crying, or otherwise emotionally agitated increased the crash risk by 9.8 times, but happened just 0.22 percent of the time. Driving while tired or drowsy increased the crash risk by 3.4 times and occurred 1.57 percent of the time.

The data also allowed researchers to accurately quantify the risk of certain driving performance or judgment errors. For example, right-of-way driving errors led to a huge jump in crash risk—936 times higher than the baseline—but occurred just 0.01 percent of the time. By comparison, failure to respect a stop or yield sign increased the crash risk by 5.3 times and occurred about 1.04 percent of the time.

Getting a handle on specific crash risk factors can help vehicle designers decide on whether to install certain systems or technologies that help reduce distracted driving, Dingus says. Car manufacturers have already been moving in that direction with certain handsfree or voice-activated infotainment systems. Still, past studies have suggested that even those systems can prove distracting for drivers, even if such studies did not directly measure the related car crash risks involved.

“There are things we can do with integrated car systems so that people are reasonably productive and reasonably safe,” Dingus says. “This helps with that by highlighting what the highest risk factors are.”

The work of Dingus and his colleagues could also help shape laws and regulations related to driving. Law enforcement officers could perhaps focus their efforts on catching and punishing the driving errors that contribute the most to unsafe roads. Officials could also target public safety campaigns by focusing on the riskier behaviors.

Having more solid knowledge of how human driver behaviors contribute to car crashes may provide extra fuel for advocates of robotic self-driving cars. Mark Rosekind, administrator of the National Highway Traffic Safety Administration, previously suggested that widespread deployment of driverless cars could lead to a 94 percent drop in the number of fatal crashes caused by human error. In that vision of the future, the human factor in car crash risks could become a thing of the past.

Dingus sees the spread of automated smart car systems that fall short of fully driverless vehicles as almost a certainty. But he remains more cautious about when self-driving cars could fully take over the tricky task of driving under all road and weather conditions—not to mention reacting to distracted drivers sharing the roads. He estimates that fully driverless cars remain perhaps 25 to 30 years away.

“I’m not scared our data will be obsolete in the next year or two,” Dingus says.


Cars That Think

IEEE Spectrum’s blog about the sensors, software, and systems that are making cars smarter, more entertaining, and ultimately, autonomous.
Contact us:

Senior Editor
Philip E. Ross
New York City
Assistant Editor
Willie D. Jones
New York City
Senior Writer
Evan Ackerman
Berkeley, Calif.
Lucas Laursen

Newsletter Sign Up

Sign up for the Cars That Think newsletter and get biweekly updates, all delivered directly to your inbox.