Follow Datanami:
June 1, 2021

Heads Up: Phone Data Shows Road Characteristics Linked to Distracted Driving

The National Highway Traffic Safety Administration (NHTSA) reports that distracted driving killed over 3,100 people in 2019, nearly a 10% increase over 2018. As habitual smartphone use becomes increasingly ubiquitous, distracted driving poses a greater and greater threat, even as cars themselves become smarter and safer. Recently, researchers at Texas A&M University leveraged a massive phone dataset to understand the factors that influence distracted driving.

“While I am driving, I always notice many drivers who are on their phones talking, texting or scrolling,” said Xiaoqiang “Jack” Kong, one of the authors of the paper and a doctoral student in Texas A&M’s Department of Civil and Environmental Engineering, in an interview with Texas A&M’s Alyson Chapman. “There are many times the cars in front of my car didn’t move after traffic lights turn green. It seems to happen to me every day. As a transportation Ph.D. student, I started to wonder how exactly this behavior could impact traffic safety.” 

Kong and his colleagues – Subasish Das, Hongmin “Tracy” Zhou, and Yunglong Zhang – approached the question from a data-driven angle. One problem: distracted drivers are, understandably, not liable to self-report their distraction. So the team worked with an (unidentified) private company that provides a smartphone application that tracks users’ driving behavior, obtaining a pseudonymized dataset of that behavior that the researchers could then match with recorded driving events on Texas roads.

Using factor analysis (a form of unsupervised machine learning), the researchers were then able to identify a set of external factors that influenced the rate of distracted driving incidents. Wide shoulders, wide medians, high speed limits, and high lane counts all contributed to distracted crashes, as did the absence of traffic lights when merging onto an interstate. In essence: if it made drivers feel more comfortable and safe, they were more likely to use their phones.

Based on the study, Kong said, there are many possible avenues for reducing distracted driving crashes.

“More visible signs and law enforcement should be placed at these urban roads with full access control and wide shoulder and medians if these urban roadways already have higher distracted crash occurrences comparing with other urban roadways,” Kong said. “Additionally, the roadways with high-speed variations also being identified as high distracted crash locations could be the roads that need more attention from transportation agencies. The countermeasures could improve traffic conditions or more strict law enforcement.”

The data isn’t flawless, of course: any app-based data will likely reflect a younger, tech-savvier population, for one. Still, it’s a promising first step, and additional data could paint an even clearer picture of the causal relationships that cause distracted driving accidents. “With more data, researchers may associate this phone use behavior with drivers’ social demographics. In this way, we may understand this behavior more at individual levels,” Kong said. 

Datanami