Examining Individualistic Driving Volatility Changes During and One Year into the COVID-19 Pandemic Using Paneled Connected Vehicle Trajectories

Abstract

The coronavirus disease 2019 (COVID-19) pandemic led to a substantial reduction in activity-travel resulting from a shift in the primary purpose of travel from work/study to shopping and essential trips. Further, several researchers noted changes to roadway safety during the early phases of the pandemic, such as a decrease in traffic crashes but an increase in traffic fatalities with respect to vehicle miles traveled. However, the existing literature is limited to holistic trends, with little inference to microscopic driving styles. This research utilizes a paneled approach to compare the freeway behavior of select passenger car drivers with respect to changes in driving volatility stemming from the lockdown period and one year into the COVID-19 pandemic. The methodological design employs a rich paneled dataset obtained from the connected vehicles pilot study in Tampa, Florida. Crash data visualization is also performed to identify crash hotspots within the study area. Volatility measures such as the standard deviation, coefficient of variation, and time-varying stochastic volatility (TVSV) are then generated and fused with traffic and weather information. Hierarchical modeling is then performed using a panel approach to account for unobserved heterogeneity within multiple observations per individual driver. The results show that the series of jerk-related driving volatility measures and the TVSV of speed increase one year into the pandemic, suggesting a reduction in overall roadway safety along the study segment. Based on the observed individual volatility changes, policy recommendations such as enhanced traffic enforcement guidelines, data-driven safety campaigns, and systematic implementation of lockdown protocols were identified.