| Co-Investigator(Kenkyū-buntansha) |
ZHANG Junyi Hiroshima Univ, Fac of Eng, Research Assistant, 工学部, 助手 (20284169)
FUJIWARA Akimasa Hiroshima Univ, Graduate School for International Development and Cooperation, A, 大学院・国際協力研究科, 助教授 (50181409)
大東 延幸 広島大学, 大学院・国際協力研究科, 助手 (60274130)
|
|---|
| Budget Amount *help |
¥2,000,000 (Direct Cost: ¥2,000,000)
Fiscal Year 1996: ¥500,000 (Direct Cost: ¥500,000)
Fiscal Year 1995: ¥1,500,000 (Direct Cost: ¥1,500,000)
|
|---|
| Research Abstract |
We have shown that Mass Point approach is effective to consider the effects of omitted variables in RP (revealed preference) panel data. However, SP panel data which we are interested in here, is accompanied with another troublesome bias, namely non-commitment bias. With respect to this bias, several methods, for instance, the one using RP data and the one using Transfer Price, have been suggested for correcting it. Because non-commitment bias is referred to as one part of the reporting bias in this paper, we verify the effectiveness of Mass Point approach to correct the reporting bias by comparing with the previous correcting methods.
To achieve the purpose described above, we use panel data including SP panel data of five waves (1987,1988,1990,1993 and 1994) and RP data (1994), which were obtained in Hiroshima Urban Area. This SP panel data is collected to investigate how much the residents prefer the New Transit System, opened in August 1994, as compared with the existing private car and bus, and what extent they actually use the new mode after its opening. Since some respondents drop out every time when the panel survey is repeated, refreshment samples are collected at that time. Consequently, different respondents have different participating patterns in the panel (i.e., participating waves are different in number across individuals). This means that we have to incorporate complete panel data and refreshment data simultaneously in the same model. Mass Point approach can easily deal with this kind of data sets. Finally, we estimate SP mode choice models (both in static and dynamic) with and without using Mass Point approach. Based on the estimated results, we evaluate Mass Point approach empirically in terms of the goodness-of-fit of models and its temporal transferability to RP data.
|
