Methodological proposal for stated preference scenarios regarding an exploratory evaluation of ride-hailing implications on transit: A Brazilian context analysis

Abstract

This study aimed to design Stated Preference (SP) scenarios for evaluating ride-hailing implications on transit. Furthermore, it proposes to identify the main factors that influence ride-hailing choice in Brazil. Based on these factors, attributes can be chosen to formulate SP scenarios to evaluate the propensity of transit users to use these new services. The results indicate that transit users in Brazil choose ride-hailing considering three main aspects: comfort, speed, and travel cost. The SP attributes were selected from these factors. Afterward, the attribute levels were identified by using data from simulated trips in ride-hailing applications. Based on these data, the application of the CART (Classification and Regression Trees) algorithm was proposed to obtain attribute levels and build scenarios. Based on the SP experiment implementation, we performed exploratory analysis by applying the CHAID (Chi-Squared Automatic Detection) algorithm and a simplified confirmatory analysis through binary logistic regression to identify the factors influencing transit substitution with ride-hailing. The results suggest that transit users are influenced by factors such as the cost of travel modes, poor assessment of transit service quality and exemption/discount for transit passengers. This article contributes to the literature by proposing a methodology based on Decision Tree algorithms regarding: (1) attribute choices and associated levels and (2) analysis of variables that influence transit substitution with ride-hailing.