Improving the Quality and Cost-effectiveness of Multimodal Travel Behavior Data Collection

 

Principal Investigator Sean J. Barbeau, PhD
Final Report (DOI) Download final report
TRID View TRID entry
Policy Brief Available Soon

Abstract

Multimodal transportation such as transit, bike, walk, transportation network companies (TNCs) (e.g., Uber, Lyft), car share, and bike share are vital to supporting livable communities. However, current data collection techniques for multimodal travel behavior, including apps built specifically for travel behavior surveys, have limitations (e.g., significant negative impact on battery life, user acquisition) which prevent a better understanding of significant real-world challenges (e.g., multimodal traveler choices, relationships between travel behavior and health). In this project, the research team developed and deployed a proof-of-concept system to collect multimodal travel behavior data on an ongoing basis directly from users of a popular open-source mobile app for multi-modal information, OneBusAway (OBA). To overcome battery life challenges, the research team used the Android Activity Transition Application Programming Interface (API), which leverages hardware advancements in modern mobile phones. An update to the OBA app was released to 676 beta testing users. Over 10 weeks, 74 users opted into the study without any incentive and contributed 65,582 trips. Key concerns for data collection when conserving battery life are the timeliness and accuracy of data. Location data was collected for 86% of all origins and destinations. Most delays in location acquisition when starting or ending an activity were under a few minutes (e.g., 90th percentile of delay at origins was 3.2 minutes and the 68th percentile was 14 seconds). The locations for trip origins and destinations were accurate approximately to a building-level or better – the 95th percentile of estimated accuracy was approximately 48 meters. The primary cause of some low activity classification confidence values seems to be uncertainty as to when the user is walking or standing still, although further evaluation is required. The software deployed in this project is a promising new tool with a tradeoff of reduced data density for the ability to collect data from many users for longitudinal studies with little to any incentives required.

Grant DTRT13-G-UTC56
USF # 79063-19
Funding Amount $147,000
Project Start Date 1/1/2018
Expected Date of Completion 12/31/2018
Sponsor Organization Center for Urban Transportation Research- National Center for Transit Research
Sponsor Organization Office of the Assistant Secretary for Research and Technology- University Transportation Centers Program -Department of Transportation