Unconstrained and partially constrained temporal modelling of pedestrian injury severities

Abstract

This study aims to explore the phenomenon of pedestrian crash severity by investigating how pedestrian injury levels have evolved in incidents occurring prior to (2019), during (2020), and after (2021) the COVID-19 lockdowns. Using Louisiana crash data, distinct annual models for pedestrian injury severity (categorised as severe (fatal and severe), minor (moderate and minor), and no injury) were developed using a random parameters logit approach, accounting for potential heterogeneity in means and variances of random parameters. Likelihood ratio tests were employed to assess the overall stability of model estimates across the studied years, and a comparison was made between partially constrained and unconstrained temporal modelling approaches. The results reveal statistically significant differences in injury severity before, during, and after the COVID-19 pandemic.