Public Transit Rider Origin–Destination Survey Methods and Technologies (2019)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

1 The purpose of this synthesis study is to understand how the current state of practice in transit rider surveys, notably origin–destination (OD) surveys, is evolving due to emerg- ing technologies and supplemental data sources. This study builds on the work of TCRP Synthesis 63: On-Board and Intercept Transit Survey Techniques (Schaller, 2005). The rider survey (also referred to as an intercept survey or an on-board survey), is a transit system survey method in which participants are recruited in public, typically on board a transit vehicle or at a transit stop, on a randomized basis. Transit providers rely on rider surveys to collect a range of data about riders and their trip-making behavior. One of the most complex types of rider surveys is the OD study. OD studies attempt to capture the trip patterns and characteristics of transit riders. Great effort is placed in ensuring that the results from these studies are statistically valid, because data from OD surveys can play several critical functions: providing inputs for regional travel demand models, meeting federal reporting requirements (notably around Title VI of the Civil Rights Act—[FTA, 2012]), supporting transit service planning, and assessing the impacts of transit investments. The expressed purpose of any given survey, however, influences survey methods that are used and the costs associated with them. Over the last decade, new technologies—such as mobile devices, high-speed cellular Internet, and automated data collection tools—have changed how organizations conduct surveys. Moreover, the emergence of big data generated by fare cards, mobile phones, and the GPS system provides new alternatives to traditional rider surveys. This report follows up on prior TCRP research on the topic to document the changing state of survey practice. The synthesis was developed to provide an overview of current survey methods and technologies to transit professionals, and assumes no prior knowledge of rider surveys. The information presented is derived from a broad literature review, a nationwide survey of 67 organizations responsible for transit surveys, and in-depth examples of innovative survey and OD study practices. One of the major findings is the growing prevalence of handheld tablet devices as a survey input mode. Half of the respondents to the synthesis survey have used tablets to conduct rider surveys. The main motivating factor for using tablets was improved data quality; tablets can eliminate transcription errors, enable the real-time monitoring of surveyors, and allow for the automatic validation of responses. Other survey methods include paper-based surveys (either self-administered or inter- view-administered), online surveys with in-person recruitment, and telephone surveys with in-person recruitment. S U M M A R Y Public Transit Rider Origin– Destination Survey Methods and Technologies

2 Public Transit Rider Origin–Destination Survey Methods and Technologies Another major finding is the impact that passive data—automatically collected data originally intended for other purposes—is having on survey practices. Survey organiza- tions increasingly rely on automatic vehicle location (AVL) systems, automatic passenger counters (APCs), and fare-card data to develop survey sampling plans and survey expansion factors. Some transit providers utilize predictive models, such as fare-card data, to infer boarding and alighting patterns without a passenger survey. One transit provider respon- dent has decided to use third-party cell phone, location-based service (LBS) and GPS data to replace on-board OD surveys. The use of passive data, notably third-party LBS data, is still in its infancy. One of the limits of passive data is that while they can provide survey administrators with a large sample of information, the data only report on limited characteristics (e.g., boarding/ alighting location). Passive data are often not suitable to collect demographic, language proficiency, first and last mile information, or customer satisfaction data. New data sources and improvements in computing power and data processing techniques may yield signifi- cant innovations in the field of passive data. This synthesis found a lack of standards for survey wording and deployment. The types of questions asked in OD surveys are similar across transit providers, yet there is little coordination among providers on question wording and the specifics of survey deploy- ment. The commonalities between surveys that do exist appear due to the use of the same survey contractor, and not a concerted effort to develop common standards for OD surveys. When creating a new survey instrument, organizations often create questions from scratch or rely on their previous survey instrument. Survey methods vary widely across the industry, including the preferred sampling plan, survey mode, and expansion methodology. Even some key measures used to monitor surveys, such as response rate, are inconsistently tracked or defined in different ways by the survey participants. This lack of standardized design and question wording makes evaluating the effectiveness of various survey practices a challenge. Regardless of the survey method used, the same fundamentals of good survey design hold true. Surveyors must field a suitably large sample to yield statistically valid results. The necessary sample size depends on the type of data collected and level of precision desired; an OD survey on a route will need a larger sample for valid stop-specific data than simply route-level data. Sample bias is another area to be carefully addressed in survey design and implementation. Certain groups of riders are less likely to participate in surveys, and special attention paid to them could ensure that the survey sample is representative of overall transit ridership. Commonly underrepresented groups include populations with limited English proficiency (LEP); riders on short trips; persons with cognitive, visual, or auditory disabilities; and persons with limited literacy. This synthesis identifies several areas for additional research. Without a controlled study environment, it is challenging to develop replicable strategies for improving survey practice. Some specific areas for further research that might help transit systems include: • Standardization of survey questions, including an assessment of optimal wording; • Establishing a consistent set of metrics to be reported on for study efforts, including a standard method of calculating the response rate, count of usable surveys, and survey cost; • Controlled study of various survey methods to determine the most effective survey mode and sampling strategies; • Research on more sophisticated expansion methodology in a controlled environment where results can be validated against other data sources;

Summary 3 • Research on the emerging and theoretical uses of passive and big data to support or supplant transit OD surveys; • Inquiry into the skills and training that transit providers and contractors need to better take advantage of new tools and techniques that rely on big data; • Research into the potential of crowdsourced data (used extensively in public engagement) for OD studies; • Research on the feasibility of collecting OD data voluntarily through a rider’s personal device (opt in), including the response bias associated with this method; • Assessment of incentives to increase response rates; • Impacts of technology use—such as tablets and computer-aided telephone interview (CATI)—on the survey response rates of underrepresented populations; • Understanding effective strategies to increase the participation rates of individuals with LEP; and • Guidance on the appropriate sample size for OD surveys based on an organization’s desired level of precision.