This study aims to model driver acceleration or deceleration rates on a complex two-lane rural highway when approaching or departing horizontal curves under nighttime driving conditions. The data used in this study were from a field experiment conducted in Pennsylvania. Research participant speeds were continuously tracked along the experimental roadway. Separate models were estimated for drivers approaching and departing horizontal curves using Ordinary Least Squares (OLS) regression. Because of the existence of heteroskedasticity in the two OLS models, efficiency was gained using a heteroskedasticity-consistent standard error estimator. The results indicate that the explanatory variables associated with acceleration or deceleration rates when approaching or departing horizontal curves in the present experiment included many geometric variables, such as curve direction, the length of approach and departure tangents, length of horizontal curve, and a vertical curve index (average rate of vertical curvature). Also, the influence of successive curves was considered and included in the models presented.
Abstract