Empirical Results of Effects of Various Causal Factors on Car-Following Behavior

Experimentation and analysis methodologies for data intended for the development of microscopic car-following models are presented. The data methodology centers on the notion that it is better to sample a varied group of real drivers when the subjects are unaware that an experiment is being conducted. This contrasts with the methods used to calibrate most conventional models. The analysis focuses on determining possible causal factors, beyond the kinematic variables most often used, that could play a role in real traffic. These factors include various human characteristics (e.g., gender and distractions related to in-vehicle conditions like telephoning and vehicle occupancy); traffic and road characteristics (e.g., type of vehicle, congestion level, and location of driving lane); and environmental characteristics (e.g., time of day and weather conditions). The paper shows the magnitudes of the relationships that are discovered and offers suggestions as to how these results might contribute to the development of better models.