Abstract
Opposing to a number of its advantages, panel data in transport also has some disadvantages such as an attrition bias caused by the non-random dropping out of participants in the panel survey and a response bias caused by the fatigue of repetitive participation in the panel for long term. The existing studies concerning the correction of the at trition bias in model building have mostly tackled with two-period panel data. However, the problem becomes more complicated and serious in modeling based on more than two-period panel data. This study aims to propose a new correcting method of the attrition and response biases in three-period panel data by using Mass Point approach. As a case study, the proposed method was applied to correct the biases in mode choice models based on SP panel data in Hiroshima. It was found that this method has higher power than the conventional methods with regards to the correction of the biased oarameters and the estimation of modal shares.
