E-scooter adoption behavior model

Contents

E-scooter adoption behavior model#

To generate the E-Scooter usage level in POLARIS, a Shared E-Scooter adoption model is developed using a suvey conducted in Chicago. Due to the ordinal nature of the Shared E-Scooter usage, the model is developed using a version of ordered probit model. Particularly, it adopts a zero-inflated version of ordered probit model since the dataset consists of excessive non-users of Shared E-Scooters (Harris and Zhao, 2007). After data analysis, it was found that E-Scooter non-users account for the largest proportion of our dataset (i.e., nearly 12% of whole dataset and 91% of potential users). Traditional ordered probability models have deficiency in handling this preponderance of zero counts in the response variable. Alternatively, zero-inflated models, developed by Lambert (1992), are primarily developed to handle excessive zero counts. Based on the data collection, the model considers zero counts of Shared E-Scooter usage for individuals (1) who responded not to use Shared E-Scooters since they consider it unsafe and will not change their behavior even if improved features and pricing options are provided, and (2) found it difficult to adopt but are open to change their behavior if convenient features are offered. The zero-inflated ordered probit (ZIOP) model has two stages. At first, it estimates whether a person tends to use E-Scooters. Following this, the model estimates his/her frequency of E-Scooter usage that includes zero usage as well. In the first stage of the model, a binary choice of potential Shared E-Scooter users (yes/no) is estimated using a zero-inflated formulation, and in the second stage, the frequency of SES usage is estimated with an ordered probit formulation.

References#

  • Khan, N. A., Gurumurthy, K. M., Davatgari, A., Auld, J. A., Mohammadian, A. (2023). Exploring the Shared E-Scooter Adoption Behavior: A Case Study of Chicago. Conditionally accepted in Transportation Research Record.

  • Harris, M. N., and X. Zhao. (2007). A Zero-Inflated Ordered Probit Model, with an Application to Modelling Tobacco Consumption. Journal of Econometrics, 141(2), 1073–1099. https://doi.org/10.1016/J.JECONOM.2007.01.002. Lambert, D. (1992). Zero-Inflated Poisson Regression, with an Application to Defects in Manufacturing. Technometrics, 34(1), 1–14. https://doi.org/10.2307/1269547.

Contents