Predicting Buckled and Unbuckled Trips of Occasional Seat Belt Users in SHRP2 Data

Abstract

Seat belts are the most effective injury prevention tool available in motor vehicles. However, researchers have a limited understanding of the factors underlying occasional seat belt use. Strategic Highway Research Program 2 (SHRP2) Naturalistic Driving Study data provide researchers a way to develop a detailed understanding of occasional belt use behavior. The present study examined two predictors (trip distance and average trip speed) of buckled and unbuckled trips from occasional seat belt users using the SHRP2 data. The present study utilized both descriptive analysis (to confirm previous finding) and predictive analysis (to assess predictive power of the models) using trip distance and average trip speed. The results from a logistic regression model showed that unbuckled trips were associated with short trips and lower speed trips compared to buckled trips. Two Random Decision Forest models were applied to predict buckled/unbuckled trips of the occasional seat belt users, and the results showed that model performance was within a fair performance range. The models differentiated between buckled and unbuckled trips at an accuracy of 61-64%. Results suggest that the occasional seat belt users may decide their seat belt use prior to trips based on the distance/duration and average. The preliminary findings of the study shed light on strategies that can be used when designing countermeasures to promote seat belt use.