An experiment was conducted to assess the distraction potential of secondary tasks performed using in-vehicle systems (radio tuning, destination entry) and portable phones (10-digit dialling, selecting contacts, text messaging) while driving. One hundred participants, ages 25-64, completed a single session in which they drove a low-fidelity (PC-based) simulator while performing the secondary tasks. The phone tasks were performed with two smart phones, one with a touch screen interface (iPhone) and one with a hard button interface (Blackberry). The Dynamic Following and Detection (DFD) driving protocol, which combines car-following with target detection, in which drivers responded to simple visual targets presented in the simulated roadway display, was used. Each combination of primary (driving) and secondary task was performed during a single 3-minute drive. Driving performance metrics included: lane position variability, car-following delay, target-detection accuracy and target-detection response time. Text messaging was associated with the highest level of distraction potential. Ten-digit dialling was the second most distracting task; radio tuning had the lowest level. Although destination entry was no more demanding than radio tuning when task duration effects were eliminated with DFD metrics, it exposes drivers to more risk than radio tuning and phone tasks due to its considerably longer duration. Modest differences between phones were observed, including higher levels of driving performance degradation associated with the touch screen relative to the hard button phone for several measures. Additional analyses demonstrated that the way in which task duration is considered in the definition of metrics influenced the outcomes of statistical tests using the metrics. The results are discussed in the context of the development of guidelines for assessment of the distraction potential of tasks performed with in-vehicle information systems and portable devices. Additional analyses were conducted to compare the DFD and Alliance and decision criteria in a simulated compliance scenario. With the large sample size (N = 100), both protocols supported the conclusion that neither text messaging nor 10-digit dialling is suitable for combining with driving; however, when a smaller (N = 40) sample was used, the protocols led to different conclusions. Additional analyses found that for, using just the vehicle performance metrics (not the eye glance metrics), samples of 20 participants did not provide sufficient statistical power to differentiate among secondary tasks. Driver age had significant effects on both primary and secondary task performance; younger drivers completed more secondary task trials on a given drive, with relatively less primary task interference than older drivers. Tests conducted using samples with wide age ranges (25-64) required larger samples to compensate for reduced homogeneity relative to samples with narrow age ranges. Half of the participants were given specific monetary incentives, while half received an equivalent amount in an unspecified completion bonus. Incentives had some effects, primarily among older participants, but no consistent overall effects on primary or secondary task performance, or on the emphasis given by drivers to the primary task. (Author/publisher)
Abstract