How did COVID-19 impact driving behaviors and crash Severity? A multigroup structural equation modeling

Risky driving behaviors such as speeding and failing to signal have been witnessed more frequently during the COVID-19 pandemic, resulting in higher rates of severe crashes. This study aims to investigate how the COVID-19 pandemic impacts the likelihood of severe crashes via changing driving behaviors. Multigroup structural equation modeling (SEM) is used to capture the complex interrelationships between crash injury severity, the context of COVID-19, driving behaviors, and other risk factors for two different groups, i.e., highways and non-highways. The SEM constructs two latent variables, namely aggressiveness and inattentiveness, which are indicated by risk driving behaviors such as speeding, drunk driving, and distraction. One great advantage of SEM is that the measurement of latent variables and interrelationship modeling can be achieved simultaneously in one statistical estimation procedure. Group differences between highways and non-highways are tested using different equality constraints and multigroup SEM with equal regressions can deliver the augmented performance. The smaller severity threshold for the highway group indicates that it is more likely that a crash could involve severe injuries on highways as compared to those on non-highways. Results suggest that aggressiveness and inattentiveness of drivers increased significantly after the outbreak of COVID-19, leading to a higher likelihood of severe crashes. Failing to account for the indirect effect of COVID-19 via changing driving behaviors, the conventional probit model suggests an insignificant impact of COVID-19 on crash severity. Findings of this study provide insights into the effect of changing driving behaviors on safety during disruptive events like COVID-19.