Living Dangerously: Driver Distraction at High Speed

Publisher Website
Google Scholar
Abstract
Recent research indicates that cell phone use can distract drivers from safe vehicle operation. However, estimates of the prevalence of cell phone use while driving have been limited to daytime hours and low-speed roadways. This paper describes the results of a study to estimate rates of cell phone use and other distractions by examining approximately 40,000 high-quality digital photographs of vehicles and drivers on the New Jersey Turnpike. The photographs, which originally were collected as part of a separate study, were taken both during the day and during the night and at different locations across the span of the Turnpike. A radar gun linked to the camera recorded the speeds of vehicles as they passed. This provided us with the speeds of every vehicle photographed, and allowed us to determine population counts of vehicles. A panel of three trained coders examined each photograph and recorded the presence of cell phone use by the drivers or any other distracting behavior. Demographic information on the driver was obtained during previous examinations of the photographs for an unrelated study. A rating was considered reliable when two out of the three coders agreed. Population estimates (and confidence intervals) of cell phone use and other distractions were estimated by weighting the cases by the inverse probability of vehicle selection. Logistic regression was used to predict cell phone use from demographic and situational factors. The results indicated that the most frequent distraction was cell phone use: 1.5% of the drivers on the Turnpike were using cell phones compared to the 3 to 4% use rates reported in the National Occupant Protection Use Survey (NOPUS) surveys conducted during the daytime on lower speed roadways. The Turnpike survey indicated that cell phones were used less on weekends and at night, and when the driver was exceeding the speed limit or had a passenger in the car.

This publication has 5 references indexed in Scilit: