Abstract
The factors that influence transit ridership are explored at the level of individual transit stops for the local and regional bus transit system in the region of Arnhem–Nijmegen in the Netherlands. Direct transit ridership modeling was used to explore simultaneously the influence of spatial, population, and network characteristics on bus stop–level ridership (number of passengers boarding and alighting from transit vehicles at each particular transit stop). Cross-sectional multiple regression models were built for two periods: March 2012 and March 2013. Between these periods, the regional transit supply changed considerably because of the start of a new tender period. The outcomes of the cross-sectional multiple regression models were compared with fixed-effects panel data models, which related the changes in ridership between both periods to the changes in transit supply characteristics. The adjusted R2 of the two cross-sectional models are .772 and .762, respectively; this finding shows that the models perform well in explaining the variance in ridership. Most selected independent variables are highly significant, and their influence on ridership is largely in line with expectations. The cross-sectional and the panel data models show large similarities, but the values of most coefficients in the panel data model are only about half of the corresponding variables in the cross-sectional models. This finding is likely due to the potential adjustment time that travelers need to get used to the changes in transit services and to an overestimation of the importance of transit supply because of the endogeneity between supply and potential demand in the cross-sectional models.
Get full access to this article
View all access options for this article.