An Analysis of Transit Bus Driver Distraction Using Multinomial Logistic Regression Models

This paper explores the problem of distracted driving at a regional bus transit agency to identify the sources of distraction and provide an understanding of factors responsible for driver distraction. A risk range system was developed to classify the distracting activities into four risk zones. The high risk zone distracting activities were analyzed using multinomial logistic regression models to determine the impact of various factors on the multiple categorical levels of driver distraction. The models demonstrated that the highest source of driver distractions was due to passenger-related activities and the level of distraction was influenced by driver demographics, driving hours, and location. The model results were validated by simulation over an entire range of random input variables. The model and results could assist in mitigating distraction and improving transit performance