The reliability and scheduling delay of travel time attributes have been considered as important factors in traveler’s decision making. Numerous studies have attempted to incorporate travel time reliability and scheduling delay early/late attributes into traveler’s choice models since the last decade. However, there is still a wide-ranging debate on empirical valuations, and substantial differences of estimation values are shown among studies. Our aim in this study is to investigate several unresolved issues in the empirical valuation of reliability and scheduling delay delay/late and estimate these effects by means of a multivariate statistical technique: meat-analysis. The main finding is that including all reliability and scheduling delay early/late attributes in choice model would lead to lower estimated values for these attributes. We also find that the stated preference data produce substantial lower values for the ratio between scheduling delay early/late and travel time coefficients and the possible explanation may be the misperception error together with the risk aversion attitude of travelers. Key words: travel time reliability, scheduling delay early, scheduling delay late, meta-analysis.
