How Should Vehicle Miles Traveled Displaced by E-Scooter Trips be Calculated?

Abstract

A dominant practice in assessing mode shift for scooter sharing is to survey scooter riders on what transportation mode they otherwise would have taken. Trip share displacing car modes and average scooter trip distance are common metrics used to support calculations like vehicle miles traveled (VMT) and emissions reductions. This practice assumes a 1:1 relationship between trip and mile replacement, implying no relationship between mode replaced and scooter trip distance. In our study, we test this assumption. An in-app end of trip survey was answered by nearly 12,000 e-scooter riders using Spin e-scooter sharing in four U.S. cities: San Francisco, Tampa, Washington D.C., and Portland. Trip GPS breadcrumb distance was linked to survey responses, enabling scooter distance traveled to be calculated by mode displaced. We calculate scooter distance share and average trip distance conditional on mode displaced for scooter trips that displaced personal car, taxi, and transportation network company trips as alternative metrics to trip share and average trip distance, respectively. To account for the effects of built environment, we cluster scooter trips on the U.S. Environmental Protection Agency National Walkability Index for trip start location. We find that across cities and walkability index clusters, the conventional trip share and average trip distance metrics consistently overestimate walk-mode shift and underestimate car VMT displacement by e-scooter use. To ensure an accurate representation of mode shift, we recommend that cities, operators, and researchers report average trip distance conditional on mode displaced or scooter distance share. The mode shift observed across the four cities in this study, for example, revealed that about one-third of the distance travelled by scooters displaced car VMT.