Use of Subway Smart Card Transactions for the Discovery and Partial Correction of Travel Survey Bias

Abstract

Although the theoretical sources of bias in travel surveys have been documented, data that describe an entire population of travelers rarely permit the reliable detection and measurement of bias. The existence of large databases of smart card transactions in public transit systems presents an opportunity to do so. In this paper, a typical average weekday of travel demand data from the Montreal, Canada, household travel survey is confronted with a single, specific day of smart card transactions. The object of comparison is the Montreal subway system, which is involved in 10% of all daily trips within the metropolitan area. The results of the initial analysis indicate that although the survey accurately reproduces daily subway ridership, it overestimates subway boardings by 24% during peak periods. This overestimation can be corrected by adjusting the weights of home-based trips to match entry volumes at subway stations during the morning peak period. The results of the reweighting procedure suggested that francophone households that use transit had a greater propensity to respond to the survey compared with other households. Furthermore, even after reweighting, the travel survey underestimated off-peak demand by roughly 21%. The underestimation was likely attributable to underreporting of non–home-based trips by respondent households and nonresponse of specific population groups.