Combining Bayesian hierarchical modeling with matched pair sampling for exploring gender differences in driver injury severity

Abstract

Gender equality is fundamental for achieving inclusive growth of the global community. To promote gender equality, challenges in designing inclusive and gender-sensitive public policies must be launched. For example, understanding gender differences in traffic safety is important for aiming to an equitable transportation system. Over the years, significant improvements have been made in vehicle safety technology, road design and crash test standards; however, systematic gender differences in injury severity persist. Literature shows contrasting findings concerning the effect of gender on crash injury severity. It seems that conflicting evidence is mainly due to confounding factors and unobserved heterogeneity in crashes. To examine gender differences in driver injury severity, this study proposes an innovative approach based on combination of matched pair sampling with Bayesian hierarchical modeling, analyzing 2022 police-reported motor vehicle crash data. Specifically, naturally occurring pairs of male and female drivers in two-vehicle crashes are analyzed. Results suggests that females face a higher risk of injury severity, but this gender difference seems less obvious when alcohol/drugs are involved. These findings highlight the need to develop safety policies oriented to minimize the gap, e.g. revising vehicle safety standards to better protect females, developing gender-sensitive safety interventions, and promoting effective driver training programs.