Exploring behavior shifts and sample selectivity issues among speeding single-vehicle crash-injury severities before-and-after the stay-at-home order.

Abstract

This study systematically explores the cause of the increase in single-vehicle speeding crash injury severities in California during and after the stay-at-home order. 27,696 speeding crashes on both highways and non-highways before-and-after the order are selected from the California Highway Patrol system. Specific countermeasures and implications of heterogeneity in means and variances are analyzed based on marginal effects. Out-of-sample simulations are employed to address two fundamental causes of the rise in injury severities: a shift in driver behaviors and the overrepresentation of riskier drivers. Results indicate that a shift towards more aggressive driving behaviors is the main reason for the increments of injury severities on highways after the order. The overrepresentation of riskier drivers is identified as the main cause during the order (both roadways) and on non-highways after the order. Since the predicted proportions on non-highway models before and during the order are closer compared to highways, this further suggests that local drivers are more inclined to violate the restriction and travel within neighborhoods during the order, which could contribute to the selectivity of riskier drivers. The findings of behavior shifts and sample selectivity issues provide valuable insights for future stay-at-home order practice, restriction improvement, and complementary policy development.