THE STUDY OF EXPERIMENTAL DESIGN FOR TRANSFER BEHAVIOR IN TRANSIT COMMUTING TRIPS
Transfer is a behavior that generates waiting time and lessens passengers? comfort in transit trips. It usually delays trips and affect level of service to the whole transit network. In this study, a 26 factorial experiment is designed to analyze 6 factors? effects on passengers? transfer behavior in transit commuting trips. The six factors are independent and selected as follows: 1st: O-D Length; 2nd: Curvature of routes; 3rd: Overlap ratio of routes to main roads; 4th: Friday or not; 5th: Morning or Afternoon; 6th: UW-Students or not. By both Lenth and Dong Method in a 26 factorial experimental design, one main effect together with one interaction effects significantly influences commuters? transfer behavior in transit commuting trips in 90% confidence interval (CI). The main effect is ?Overlap ratio of main roads? and the interaction effect is ?Curvature of routes & Overlap ratio of main roads?. One interaction effect of ?Curvature of routes & UW-Students or not? is significant only by Dong method in 90% CI. When CI is adjusted to 80% or narrower, one other interaction effect of ?O-D Length & Overlap ratio of main roads & UW-Students or not? becomes significant by Dong method. Based on significant effects, models are created to predict the probability of transfer in transit commuting trips in Madison Wisconsin. Local transit operational management team can apply the model to forecast probability of transfer and then dispatch transits or plan routes schedules in a more scientific way.