Contents lists available at ScienceDirect Transport Policy journal homepage: www.elsevier.com/locate/tranpol Comparing data quality and cost from three modes of on-board transit surveys Asha Weinsten Agrawal ⁎ , Stephen Granger-Bevan, Gregory L. Newmark, Hilary Nixon Mineta Transportation Institute, USA ARTICLE INFO Keywords: Survey methods Public transit planning Customer research Performance measurement ABSTRACT Many transit agencies invest substantial resources in surveying their passengers to generate data used for planning, marketing, and equity analyses. Within the industry, there is considerable interest in replacing traditional paper-based self-complete surveys with new approaches that might lower costs or generate better quality data. However, very limited research has been done to identify the relative performance of different transit passenger survey modes. This paper begins to fill that gap. The research investigates the relative data quality for three different bus passenger survey methods distributed or administered on the transit vehicle: self-complete paper surveys, interviewer-assisted tablet- based surveys, and self-complete online surveys. The research used an experimental design, with the same survey questionnaire distributed via the three survey modes. All factors about the survey and distribution process were kept identical to the extent feasible, so that the only variation would be the survey mode itself. The findings by survey mode are compared in terms of the overall response and completion rates, the completion rate for individual questions, respondent demographics, and labor costs per complete. The study results suggest that many agencies may still find the old-fashioned, low-tech paper survey to be the best option for bus passenger surveys. The paper mode required less labor per complete, and for many of the metrics discussed it generated data that was as good as—or better than—the tablet survey. In addition, the findings suggest that online survey invitations distributed on the transit vehicle are not a good option because they were labor intensive and had very low response rates. 1. Introduction This research investigates the relative data quality and labor costs for three different modes of surveying bus passengers using survey methods distributed or administered on the transit vehicle: self- complete paper surveys, interviewer-assisted tablet-based surveys, and self-complete online surveys. Many transit agencies invest substantial financial and time re- sources into surveying their customers, with costs easily running from $500,000 to a $1,000,000 for a large agency. For example, a 2006 survey of passengers on Chicago’s Metra commuter rail system cost more than $600,000. Expensive survey efforts are justified on the grounds that the data collected are fundamental inputs for a wide range of purposes that include travel modeling, system-wide or route-level planning, setting fare policy, and communicating with existing custo- mers (Schaller, 2005; Members and Friends of the Transportation Research Board’s Travel Survey Methods Committee, 2015). Since the fall of 2012, the US Department of Transportation’s Federal Transit Administration (FTA) has directed larger transit agencies to conduct passenger surveys every five years and ensure participation from minority and low-income riders who have histori- cally under-participated in such efforts (U.S. Department of Transportation, Federal Transit Administration, 2012). This FTA directive will require many U.S. transit agencies to survey more frequently than they have in the recent past. Thus, agencies have newly-strengthened interest in identifying survey methods that will reduce costs while still gathering high-quality data. When planning a survey, transit agencies must choose among a variety of possible survey modes, with little guidance available to help them assess cost and quality tradeoffs. The growing availability of affordable information and communications technologies has led a number of agencies to experiment with new survey modes in hopes of either improving data quality or reducing costs. In recent years, agencies have tried a variety of survey modes, including: 1. On-board distribution of self-complete post-cards that collect phone http://dx.doi.org/10.1016/j.tranpol.2016.06.010 Received 27 January 2016; Received in revised form 29 April 2016; Accepted 1 June 2016 ⁎ Corresponding author. E-mail addresses: Asha.weinstein.agrawal@sjsu.edu (A.W. Agrawal), stephen.grangerbevan.sjsu@gmail.com (S. Granger-Bevan), gnewmark@ksu.edu (G.L. Newmark), hilary.nixon@sjsu.edu (H. Nixon). Transport Policy (xxxx) xxxx–xxxx 0967-070X/ © 2016 Elsevier Ltd. All rights reserved. Please cite this article as: Agrawal, A.W., Transport Policy (2016), http://dx.doi.org/10.1016/j.tranpol.2016.06.010