Investigating changes in longitudinal associations between declining bus ridership, bus service, and neighborhood characteristics

Abstract

Amid the recent transit ridership decline, gaining an understanding of the factors affecting ridership becomes crucial for transit agencies to utilize limited resources effectively. I use generalized linear multilevel negative binomial models to investigate the longitudinal relationship and changes in the associations between neighborhood-level bus ridership and a series of socio-economic and bus service factors in Philadelphia between 2014 and 2018. Data come from passenger boarding at bus stops in Philadelphia. Results show that the associations between bus ridership, population and the number of jobs, and the percent of zero-car households are positive, but weakened over time. The associations between ridership and bus service supply are inelastic. The findings have implications on transit agencies’ resource allocation and service adjustments as they recover from the ridership and revenue losses during the COVID-19 pandemic while facing competition from new travel options such as Uber and Lyft.