National Academies Press: OpenBook
« Previous: Chapter 3 - Current State of Practice
Page 41
Suggested Citation:"Chapter 4 - Case Examples." National Academies of Sciences, Engineering, and Medicine. 2019. Public Transit Rider Origin–Destination Survey Methods and Technologies. Washington, DC: The National Academies Press. doi: 10.17226/25428.
×
Page 41
Page 42
Suggested Citation:"Chapter 4 - Case Examples." National Academies of Sciences, Engineering, and Medicine. 2019. Public Transit Rider Origin–Destination Survey Methods and Technologies. Washington, DC: The National Academies Press. doi: 10.17226/25428.
×
Page 42
Page 43
Suggested Citation:"Chapter 4 - Case Examples." National Academies of Sciences, Engineering, and Medicine. 2019. Public Transit Rider Origin–Destination Survey Methods and Technologies. Washington, DC: The National Academies Press. doi: 10.17226/25428.
×
Page 43
Page 44
Suggested Citation:"Chapter 4 - Case Examples." National Academies of Sciences, Engineering, and Medicine. 2019. Public Transit Rider Origin–Destination Survey Methods and Technologies. Washington, DC: The National Academies Press. doi: 10.17226/25428.
×
Page 44
Page 45
Suggested Citation:"Chapter 4 - Case Examples." National Academies of Sciences, Engineering, and Medicine. 2019. Public Transit Rider Origin–Destination Survey Methods and Technologies. Washington, DC: The National Academies Press. doi: 10.17226/25428.
×
Page 45
Page 46
Suggested Citation:"Chapter 4 - Case Examples." National Academies of Sciences, Engineering, and Medicine. 2019. Public Transit Rider Origin–Destination Survey Methods and Technologies. Washington, DC: The National Academies Press. doi: 10.17226/25428.
×
Page 46
Page 47
Suggested Citation:"Chapter 4 - Case Examples." National Academies of Sciences, Engineering, and Medicine. 2019. Public Transit Rider Origin–Destination Survey Methods and Technologies. Washington, DC: The National Academies Press. doi: 10.17226/25428.
×
Page 47
Page 48
Suggested Citation:"Chapter 4 - Case Examples." National Academies of Sciences, Engineering, and Medicine. 2019. Public Transit Rider Origin–Destination Survey Methods and Technologies. Washington, DC: The National Academies Press. doi: 10.17226/25428.
×
Page 48
Page 49
Suggested Citation:"Chapter 4 - Case Examples." National Academies of Sciences, Engineering, and Medicine. 2019. Public Transit Rider Origin–Destination Survey Methods and Technologies. Washington, DC: The National Academies Press. doi: 10.17226/25428.
×
Page 49
Page 50
Suggested Citation:"Chapter 4 - Case Examples." National Academies of Sciences, Engineering, and Medicine. 2019. Public Transit Rider Origin–Destination Survey Methods and Technologies. Washington, DC: The National Academies Press. doi: 10.17226/25428.
×
Page 50
Page 51
Suggested Citation:"Chapter 4 - Case Examples." National Academies of Sciences, Engineering, and Medicine. 2019. Public Transit Rider Origin–Destination Survey Methods and Technologies. Washington, DC: The National Academies Press. doi: 10.17226/25428.
×
Page 51
Page 52
Suggested Citation:"Chapter 4 - Case Examples." National Academies of Sciences, Engineering, and Medicine. 2019. Public Transit Rider Origin–Destination Survey Methods and Technologies. Washington, DC: The National Academies Press. doi: 10.17226/25428.
×
Page 52
Page 53
Suggested Citation:"Chapter 4 - Case Examples." National Academies of Sciences, Engineering, and Medicine. 2019. Public Transit Rider Origin–Destination Survey Methods and Technologies. Washington, DC: The National Academies Press. doi: 10.17226/25428.
×
Page 53
Page 54
Suggested Citation:"Chapter 4 - Case Examples." National Academies of Sciences, Engineering, and Medicine. 2019. Public Transit Rider Origin–Destination Survey Methods and Technologies. Washington, DC: The National Academies Press. doi: 10.17226/25428.
×
Page 54

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.

41 Five case examples were chosen to highlight the current state of practice of origin–destination (OD) transit surveys. Interviews were conducted in March and April 2018 with representatives from the following organizations: • Metropolitan Transportation Commission (San Francisco Bay Area) • Los Angeles County Metropolitan Transportation Authority • Chatham Area Transit (Savannah, GA) • Massachusetts Bay Transportation Authority (Boston) • TriMet (Portland, OR) These organizations provided a range of insights, from effective techniques in conducting rider surveys to the use of passive data to supplant the survey process. Metropolitan Transportation Commission Background The Metropolitan Transportation Commission (MTC) is the metropolitan planning orga- nization (MPO) for the nine counties (Alameda, Contra Costa, Marin, Napa, San Francisco, San Mateo, Santa Clara, Solano, and Sonoma) in the San Francisco Bay area (see Figure 11). MTC is responsible for planning, financing, and coordinating OD survey efforts for the 23 tran- sit agencies within the region. These 23 transit providers—AC Transit/Dumbarton Express, Altamont Commuter Express (ACE), BART, Caltrain, County Connection, Fairfield Suisun Transit (FAST), Golden Gate Transit & Ferry/Marin Transit, Livermore Amador Valley Transit Authority (WHEELS), Muni (SFMTA), Napa VINE, Petaluma Transit, Rio Vista Delta Breeze, SamTrans, San Francisco Bay Ferry, Santa Rosa CityBus, SMART, Soltrans, Sonoma County Transit, TriDelta Transit, Union City Transit, Vacaville City Coach, VTA, and WestCAT— operate express bus, ferry service, light rail, heavy rail, and commuter rail, together serving more than 1.7 million boardings per day. MTC was selected as a case example because of the scale of its survey efforts. MTC is nearly continuously conducting surveys for one of its 23 transit providers, affording it the opportu- nity to gain experience across modes and experiment with different survey approaches. Infor- mation in this section draws from interviews conducted with Shimon Israel, Senior Planner at MTC (March 30 and June 15, 2018). MTC funds and manages OD survey efforts, and hires contractors to staff and conduct the actual survey process. MTC works directly with the contractor to manage all contracting, invoicing, and scheduling. For the seven largest transit operators in the region, MTC funds 80 percent of the survey costs, with the operator responsible for the remaining 20 percent. C H A P T E R 4 Case Examples

42 Public Transit Rider Origin–Destination Survey Methods and Technologies For the smaller agencies, MTC funds the entire cost of the study. Because of the importance of acquiring accurate transit data, a critical element of maintaining the regional travel model, MTC has not had problems justifying survey expenses internally or to the public. Evolution of Survey Practice MTC began leading the OD surveys for all transit providers in the Bay Area in 2012 to improve data quality and standardization. Inconsistency in survey quality and frequency across opera- tors made comparing data between the systems challenging. As a regional organization, which also manages regional household travel surveys and a travel model, MTC was a natural home for such an effort. Based on the positive experience of Los Angeles (see LA Metro case example), MTC initially utilized a two-step computer-aided telephone interview (CATI) approach for its surveys. The two-step approach involved handing out a brief paper rider survey followed shortly by a more detailed telephone interview. MTC found that CATI surveys resulted in fewer data errors than paper surveys but yielded a low response rate. MTC decided to transition to a tablet-based interview approach for its next round of surveys, based on FTA guidance and the positive experience with tablets by some of MTC’s survey con- tractors. Tablets have yielded high response rates and high-quality data (MTC reports an aver- age response rate of more than 75 percent, although with the caveat that this figure comes from its contractors and has not been independently verified). In addition, unlike traditional paper surveys, trained interviewers can clarify questions for respondents taking a tablet-based survey. MTC’s contractors can track responses in real time and more quickly address quality issues in the field. A CATI option is still offered as an alternative mode in cases where a respondent is Source: Courtesy of the Metropolitan Transportation Commission. Figure 11. MTC planning area.

Case Examples 43 unable to complete a survey in person. MTC reports that CATI surveys represent a small share of total responses. Survey Methodology MTC follows most of the best practices in survey design and implementation advocated by FTA, such as the use of pre-testing, auto-validation of survey responses, real-time monitor- ing of survey responses and data quality, accommodations for surveying short trips, and use of on/off surveys for sample plan development and result expansion (K. Cervenka, telephone interview, June 13, 2018). The following summarizes MTC’s general approach for surveys; however, there is some variety in survey methodology based on the system and survey vendor conducting the work. • Frequency and Timing of Surveys: Surveys are conducted every fall and spring, resulting in each of the 23 agencies being surveyed on a 5- to 7-year cycle. MTC avoids surveying during the summer and winter to avoid seasonal-specific changes to transportation demand such as Christmas and summer break. • Survey Instrument Design: MTC utilizes a standard survey, with the option for agencies to add five or six custom questions. Surveys cover origin and destination trip data and demo- graphic information for Title VI federal requirements. During the survey development, MTC consulted with FTA as well as with some of its peers on survey design. • On/Off Surveys: Before conducting a survey, MTC will sometimes conduct a study to track the boarding and alighting pairs of riders. These studies are typically conducted for higher ridership routes (>2,000 boardings per day) or routes that serve multiple different types of markets along its alignment. On BART, the region’s rapid transit system, boarding and alighting locations are automatically collected through the electronic fare collection system. For other transit providers, MTC will conduct an on/off survey by distributing cards to riders as they board and collecting the cards as they alight. The boarding and alighting data are used to develop the sampling plan and expansion factors. • Pre-Testing: MTC conducts pre-tests before most of its rider surveys. Typically, the pre- tests focus on deploying the survey in challenging environments likely to be encountered in the field. For example, pre-testing on MUNI focused on how to best intercept passengers on crowded buses with rear-door boarding. MTC also uses pre-testing to gauge the public’s receptiveness to new questions, especially if the question is of a sensitive nature. At the end of the pre-test phase, the survey vendor provides MTC with a report outlining challenges identified during the pre-test and recommendations for addressing them. • Survey Recruitment: As mentioned previously, MTC relies largely on an interview- administered tablet approach for surveys. Survey staff are placed aboard vehicles or at station platforms. Participants are selected randomly based on the vendor’s selection algorithm. The selection process may vary depending on conditions such as location and crowded ness. For example, on BART the surveyor may be instructed to approach someone at a certain location on the platform, whereas on a crowded MUNI bus, they instead may be instructed to approach a rider at a particular clock-face location relative to where the surveyor is standing (e.g., approach the person at your 9 o’clock). • Survey Monitoring: One of MTC’s vendors can track survey responses in real time. Moni- toring tools allow management to identify and address systematic issues quickly—for exam- ple, instances where a surveyor is regularly skipping certain questions or showing bias in the race, gender, or age of people being approached. • Accommodations for Short Trips: Short trips are notably challenging to collect data on during OD surveys. MTC’s vendors utilize a range of strategies to accommodate such riders. Surveyors often will get on or off a vehicle with the rider so that they can complete

44 Public Transit Rider Origin–Destination Survey Methods and Technologies the survey during the trip. In instances where that is not feasible, the rider will be asked for contact information and contacted within 24 hours to complete a CATI survey. In instances where surveyors are below quota on certain trip pairs, surveyors will begin recruiting riders on platforms or vehicles solely based on the length or location of their trip. • Accommodations for Low–English Proficient Riders: Survey staff are often bilingual and can assist riders in completing surveys in their native language. In instances where a language cannot be accommodated by the survey staff, the rider’s contact information is collected and a follow-up CATI survey is performed. • Recruitment of Minors: To avoid age bias in the sample, MTC has measures to recruit minors. Surveyors are instructed to openly approach riders who appear 16-years or older. In cases where the rider appears younger than that, the surveyor asks an accompanying adult to act as a proxy. MTC does not typically sample school-tripper routes that are disproportion- ately ridden by minors. • Handling of Sensitive Questions: Certain sensitive questions, such as the participant’s income or gender, may raise privacy and completion bias concerns. Such questions are left for the end of the survey. Surveyors can hand the tablet over to respondents to complete such questions on their own. • Survey Expansion: Survey results are expanded based not only on ridership but trip charac- teristics observed in the on/off study conducted before the survey itself. Trip pairs underrep- resented in the data are assigned higher weights. In cases where a trip is not reflected in survey results, a synthetic record is added that does not include any demographic characteristics. Results While MTC strongly supports and recommends the use of tablets to conduct surveys, tablets are not without their challenges. MTC expressed concern that there may be bias in an inter- viewer’s selection and respondent’s willingness to participate. For example, interviewers may skip a rider who appears homeless, while potential respondents may be fearful of engaging with tablets due to privacy concerns and technology barriers. MTC is currently conducting pilot studies to determine the extent to which using tablets may introduce bias against populations within different racial and ethnic and/or household income categories. Additionally, MTC stressed the importance of well-trained survey staff. Because of the breadth of the survey effort, MTC must rely on contracted temporary workers to administer tablet surveys. These temporary workers may not all have the same survey administration skill level. In addition, temporary employees are incentivized to complete a certain number of surveys per shift. Incentivizing the volume of surveys administered may compromise the quality of some of the data (i.e., interviewers may rush to finish one survey and not administer the entire survey to obtain an additional response). The real-time monitoring tools mentioned earlier are important for managing quality and identifying any employee issues early on. MTC believes that the increased cost of using tablets is justified by improved data. There have been minor concerns over lost or stolen tablets; contractors, however, are responsible for the security of the tablets. MTC staff did note that certain transit operators did not allow surveying with tablets on late-night trips because of crime concerns, but this did not have a large impact on the overall data quality. Lessons Learned Moving forward, MTC will continue to look at the impact that tablets may have on respon- dent selection bias related to Title VI populations. MTC is aware that there may be human error in the survey sampling, but believes tablet results are still far superior to other surveying

Case Examples 45 methods. Because survey data are used to calibrate the regional travel model, accurate and high- quality data are essential. In a discussion about survey administration lessons learned, MTC staff noted the following: 1. Tablets obtain a higher response rate than other methods. 2. The quality of data from tablet surveys is excellent, justifying the increased cost. 3. Training and monitoring interviewers is extremely important to the success of the tablet method. 4. CATI is an important option to have available to obtain information on short trips and on persons with limited English proficiency. 5. Pre-testing is an important step in the process and allows surveyors to identify and address challenges early on. 6. Even with tablets, urban bus remains the hardest mode to survey due to the number of short trips and diverse populations, as well as the usage of rear-door boarding. TriMet The Tri-County Metropolitan Transportation District of Oregon (TriMet) operates bus, light rail, commuter rail, and paratransit service across a three-county region centered on Portland, Oregon. The agency carries more than 300,000 riders on a typical weekday. TriMet was selected as a case example for this report because the agency is a strong proponent of the tablet-based survey approach, and unlike many other large transit providers, has chosen to develop in-house surveying capacity in lieu of relying on outside contractors. The agency regularly designs and manages a range of rider surveys such as OD studies, before-and-after studies, fare studies, satisfaction surveys, and APC validation. This case example is the result of correspondence with Baofeng Dong, Bibiana McHugh, and Virginia Shank at TriMet during April 2018, along with findings documented by McHugh et al. (2017) in “Conducting Onboard Transit Rider Surveys with Electronic Handheld Tablets: An Agencywide Consolidated Approach,” which was discussed as well in the literature review chapter. Background and Overview of Survey Approach TriMet’s Information Technology-GIS department is responsible for managing all survey efforts. The agency internally handles development of the sampling plan, data processing, and data analysis. All survey equipment, including the Android tablets used to conduct the surveys, are furnished by the agency. TriMet developed a custom survey software in-house that uses Open Data Kit, an open-source set of tools for mobile data collection and management. The agency chose to conduct its surveys in-house to retain knowledge internally, improve efficiency, and reduce costs. In the past, the agency has used vendors to provide survey staff but after switching from paper to tablet surveys in 2014, has relied exclusively on student interns recruited from local colleges and universities. These students are interviewed by TriMet and paid through a temp agency. The agency’s 2016 systemwide fare study, which collected 17,719 completed surveys, required 18 interns and 3 supervisors. TriMet implements several measures to manage data quality, beginning with the recruit- ment of its interns. The agency looks for individuals who reflect the racial, linguistic, and ethnic diversity of its ridership and demonstrate a familiarity with the system, technical know-how, customer service skills, a professional attitude, and good work history. Recruits go through a 3-day training program. In the field, staff work closely with experienced supervisors. TriMet has

46 Public Transit Rider Origin–Destination Survey Methods and Technologies also leveraged the real-time capabilities of tablets to monitor survey progress through an online dashboard (see Figure 12). As mentioned previously, the agency conducts a wide range of surveys to support functions such as before-and-after reporting, Title VI requirements, service planning, and fare policy. TriMet has not conducted a systemwide OD study since the early 2000s, but has completed corridor-specific OD surveys to assess transit system improvements such as the MAX Orange Line. To meet Title VI reporting requirements, TriMet incorporated demographic, income, and English proficiency questions into its 2016 Fare Survey. 2014–2015 Orange Line Before Study Origin–Destination Survey The MAX Orange Line light rail between Downtown Portland and Milwaukie, Oregon, started revenue service in Fall 2015. As part of its before-and-after reporting requirements, TriMet conducted an OD survey along the future light rail corridor before its opening. The study area included light rail and bus lines. The agency’s approach to tablet surveys is fairly similar to that of MTC: • TriMet conducts an on/off study to determine boarding and alighting pairs before deploy- ing its full survey; agency staff distributed a scannable card to passengers aboard buses upon entry, and then collected the card when passengers alighted to determine each passenger’s start and end stop. On MAX light rail, TriMet modified this approach because it was not practical for station survey staff at all doors along the train. Instead, staff went through the Source: Courtesy of TriMet. Figure 12. Survey dashboard visualization of origin data in real time.

Case Examples 47 car asking each passenger about their boarding and alighting station. Boarding and alighting data were used to develop the survey sampling plan and expansion factors. • The full survey is an interview-administered tablet survey conducted aboard the vehi- cle. To reduce selection bias, a random-number generator determines which riders were approached by survey staff. The survey program would validate responses in real time to reduce data entry error. TriMet offered the full survey in English and Spanish, and limited English proficiency questions in 11 additional languages. TriMet did not offer a financial incentive for users to take the survey. Agency staff found that riders were self-motivated to take the survey by a desire to provide feedback and help improve TriMet service. Finally, TriMet implemented strategies to address the undersampling of short trips. Sur- veyors were instructed to ask riders a subset of required questions if their trip was too short to complete the full survey. In the expansion phase, data from the on/off study were used to increase the weight of short trips that were undersampled in the survey responses. Results TriMet has demonstrated cost savings and efficiencies gained from its tablet survey approach by comparing its 2012 Fare Study, which was conducted using paper and pencil, to its 2016 Fare Study, which was conducted entirely by tablet (McHugh et al., 2017). Some of the key advantages for the tablet approach include: • A 32 percent reduction in survey costs while yielding 4 percent more complete surveys. Tablet surveys enabled functions such as skip logic which shortened the time per survey, allowing for shorter shifts and more completed surveys per sample hour. The availability of survey results in real time eliminated oversampling. Additional cost savings were achieved due to the switch from staffing the survey with contractors to student interns. • A 48 percent improvement in response rate. • Reduction in data cleaning and processing requirements. Results were available 1 month after the survey was fielded, compared to 6 months for the 2012 survey. • Ability to monitor data quality in real time and identify issues as they arose. Key Take-Aways This case example highlights several key lessons for transit providers planning their own survey efforts: • Conducting a survey in-house can result in cost savings, build internal capacity, and give agency staff greater control over data quality and sampling strategies. TriMet was able to rely extensively on student-interns to staff the surveys. TriMet staff recognized that not all transit providers have the dedicated staff, expertise, and data infrastructure necessary to field surveys in-house (McHugh et al., 2017). A sustained commitment from management allowed TriMet to build its internal survey capacity. As the agency is conducting a near-continuous range of surveys, it can amortize its investment in staff and equipment over time. • Instead of relying on costly systemwide OD surveys to collect Title VI information, TriMet has incorporated Title VI questions into other survey efforts such as its 2016 Fare Survey. • Tablet-based surveys improved data quality compared to paper-surveys. TriMet (like MTC) noted many benefits of using tablets over paper, including the ability to validate responses in the field, elimination of data entry/transcribing requirements, and the use of real-time moni- toring of survey responses to quickly identify systematic data collection issues. Tablet surveys also have a much lower rate of invalid surveys.

48 Public Transit Rider Origin–Destination Survey Methods and Technologies • TriMet found that tablet surveys were less costly to administer than paper-based surveys. Among transit providers surveyed in this report, only 19 percent of respondents saw savings due to tablets compared to 26 percent who reported a cost increase (the remainder saw no change or did not know). TriMet staff suggested several factors that led to their cost savings compared to other survey respondents: – TriMet utilized open-source software instead of paying a vendor for survey tools. – The agency did not provide a paper-based option in parallel to its tablet survey, which reduced printing and postage costs. – TriMet partnered with local colleges to staff its 2016 survey, which reduced the cost of labor compared to the 2012 survey, which was staffed by a survey contractor. – The questionnaire design, sampling plan, data cleaning, survey expansion, and survey analysis are all done in-house. – TriMet did not supplement on-board tablet surveys with follow-up phone calls. – Survey vendors are still in the process of adopting tablet surveys and may not be taking full advantage of the technology to improve cost-efficiencies: � Eliminates the need for oversampling, � Reduces data entry costs, and � Fewer responses are thrown out for invalid answers. Los Angeles Metro Background The Los Angeles County Metropolitan Transportation Authority (LACMTA or Metro) is the third-largest transit system by ridership in the nation, with over 1.3 million riders each day on bus, light rail, and metro. LACMTA’s OD survey efforts stand out for two reasons: their large- scale use of the two-step CATI method, and their high response rate among non-English speak- ers. On April 2, 2018, the TCRP study team conducted an interview with LA Metro Principal Transportation Planner John Stesney about the agency’s OD study practices. Metro completed its most recent OD survey in 2011. Because 20 percent of transit trips within Metro’s service area occur on other local providers, Metro conducted the survey for all transit providers in Los Angeles County. Metro manages its own regional travel demand model and it was important to have an accurate regionwide snapshot of transit travel patterns. Survey Methodology Prior to 2011, Metro used paper surveys for its OD surveys. The low data quality from paper surveys incentivized Metro to find a better method. As part of a pilot study with the FTA to advance the state of survey practice, Metro found that paper surveys resulted in a high degree of response error, because participants frequently misunderstood or incorrectly completed ques- tions. The pilot study determined that a two-step CATI method would result in improved data quality, while an interview method would result in high-quality data but would be unrealistic due to time constraints, especially for short trips. In 2011, LA Metro hired a contractor to conduct its first OD survey after the FTA pilot pro- gram. The contractor selected had CATI experience and had conducted such surveys for other agencies. LA Metro attributes the success of its CATI survey to the expertise of the contractor and does not feel that the organization has the resources to re-create a CATI survey internally. In the first step of the CATI survey process, contractor staff handed out cards to riders with three contact information questions (see Figure 13). The card was printed in English on one side and in Spanish on the other. Riders could return the completed cards to staff members. Riders

Case Examples 49 would receive a follow-up phone interview within a week of completing the three-question on-board survey. Phone interviewers received training on both the survey method and service area. While conducting the phone survey, the interviewers had access to survey software that allowed them to visualize (e.g., see streets, routes, and stops) and verify the reasonableness of responses in real time, ensuring collection of higher quality data. LA Metro felt that the use of a phone survey reduced interviewer bias. All respondents had an equal likelihood of being selected, but the results may not reflect a truly random sample because of any response bias occurring in the first step of the survey. Results The contractor was also responsible for cleaning and analyzing the data. The data qual- ity obtained from the CATI survey approach was a significant improvement over the paper surveys. The survey had a high response rate at 65 percent (defined as number of approached respondents who completed both steps of the process; response rates were reported by the vendor and not independently verified by Metro). The agency believes offering multiple $500 gift cards helped increase the response rate. The agency initially experienced internal pushback from offering such a large incentive; however, it successfully made the case by dem- onstrating the savings a $10,000 investment in incentives can have for a $1.5 million study. Source: Courtesy of the Los Angeles County Metropolitan Transportation Authority. Figure 13. On-board “first-step” survey.

50 Public Transit Rider Origin–Destination Survey Methods and Technologies More than one-third (35 percent) of the surveys were completed in a language other than English. Phone staff were bilingual in Spanish, and a language line was available for respondents seeking to complete the survey in a language other than English or Spanish. While Metro felt that the CATI approach improved response quality over prior paper surveys, the organization did experience some issues with its approach. The survey may have under sampled LEP populations—notably, those speaking a language other than Spanish. Addi- tionally, respondents were hesitant to provide income information. While incentives were received positively by the public, there were a few incidents of people falsely claiming to have won the prize. Lessons Learned In a discussion about lessons learned, Metro noted the following: 1. Contractor staff training and expertise are key to successful survey implementation. 2. Metro felt providing a high-value monetary incentive helped increase the response rate. In the future, the agency would like to increase the amount of incentives provided they represent only a small share of the cost of conducting a large-scale survey effort. 3. The CATI method obtained a higher response rate and more accurate data than paper surveys. 4. Paper surveys, without an interviewer, are hard for respondents to understand. Chatham Area Transit Rider surveys are not the only method to collect information on OD patterns for transit riders. Many agencies are using passively collected data—records that exist for another pur- pose, or that are collected without active means—to derive actionable planning information. This case example considers a small transit operator that built an OD planning tool with the help of triangulated cellular device location data. Chatham Area Transit (CAT) is the primary transit agency in Savannah, Georgia, serving riders in Chatham County (2015 population 286,956). It operates approximately 70 vehicles on 17 routes, and serves about 13,000 weekday riders (FTA, 2016a). Over the last few years CAT observed declining bus ridership on their system, about 1 per- cent per year between 2013 and 2016. In an effort to reverse this decline and better serve their community, CAT elected to initiate a major network redesign. To inform the redesign, CAT undertook an OD study. Information in this section is drawn from personal interview with Grant Sparks, Planning Manager at CAT (March 29, 2018). Although CAT conducted a rider survey in 2013, the data collected at that time were inad- equate for a variety of reasons. As the purpose of the system redesign was to better serve the traveling population, including those not already taking transit, CAT determined that updating the previous study with the same methodology, which would only focus on existing riders, was not sufficient; the goal of the system redesign was also to capture potential riders that are not currently served. Survey Methodology CAT elected to take a passive data approach to their system redesign OD study, with a focus on the entire traveling population. In this approach, CAT assembled study data from previously collected sources, with three major elements: • Mobile device data purchased from a commercial vendor, • Stop-level boarding data from a previous study, and • Transit propensity scores based on Census data.

Case Examples 51 Working with the regional MPO, CAT purchased AirSage (https://airsage.com) data for their study region. AirSage obtains mobile device location data from cellular networks and application services and processes the data into zone-based flows distinguished by time of day, imputed trip purpose (home to work, home to other, etc.), and population (resident, visitor, inbound commuter). CAT worked with their consultants to carefully specify the data pur- chase. They determined that using data from October 2015 and April 2016 would provide the clearest picture of regular operations, avoiding both peak visitor season as well as disruptions from hurricanes. Because cellular triangulation has some geographic margin of error, CAT and the MPO aggregated traffic analysis zones in the central city into larger districts suitable for AirSage to use. CAT is in the process of installing APC and AVL systems in its vehicles, but these systems were not available for the OD study. As such, CAT and its consultants used stop-level boarding and transfer data from the 2013 OD survey to build a picture of existing transit patterns. The final piece of data for the study is a geographic transit-use propensity score that the consul- tants developed from aggregate Census data, considering job and household density in addition to income level and other demographic characteristics. The score is based on information from the TCRP Transit Capacity and Quality of Service Manual (Kittelson and Associates, Inc., et al. 2013). Results By overlaying the cellular OD trip patterns for all trips, the stop-level ridership data, and the transit propensity score, CAT made several observations that could help it improve its service offerings: • The transit network was originally designed on the assumption that all transit riders would want to come downtown, but a shopping mall and hospital on the edge of the urban core appear to be relatively larger trip attractors for likely transit riders. • Up to 60 percent of the CAT service area is in neighborhoods with a low propensity to use transit. • There is a discrepancy between transit service and desired travel paths. CAT has contracted with consultants to redesign its transit services, and they anticipate that the data and findings from this study will be useful toward that effort. Next Steps In this most recent OD study, the consultants’ focus was not solely on providing informa- tion in a single effort, but on building a framework for regular and updated insight that CAT can use independently. For example, when its APC and AVL systems come online in the next couple of years, these data will supply the stop-level boarding data that in this study were based on data from the 2013 travel survey. CAT can also purchase the mobile device data at regular intervals, and again look for places and paths where transit service and overall trip patterns align or conflict. CAT has no plans to conduct a major transit rider OD survey in the future. CAT will conduct regular customer attitude surveys as a supplement to the data-driven process described in this case example. CAT intends to meet Title VI requirements in its redesign by analyzing the impacts of service changes, including to neighborhoods not well served by the existing system. CAT plans to use service area Census demographic data in its Title VI analysis and the report it submits to FTA; the Title VI report in the Remix transit planning software will provide a starting point for this analysis.

52 Public Transit Rider Origin–Destination Survey Methods and Technologies Massachusetts Bay Transportation Authority Another major source of passive data for transit planning is data that agencies themselves collect through various systems. Electronic fare collection (EFC) systems record when particu- lar fare cards are used to pay fares, and on which vehicle; AVL data provide information on where the vehicles are located when a fare card is used. Information on passenger loads can be obtained from transit agency APC systems. This case example focuses on a large transit agency that has built a powerful model based on these passive systems to study its passenger flows. The Massachusetts Bay Transportation Authority (MBTA) is the regional transit agency for the greater Boston area. The MBTA operates commuter rail, heavy rail, light rail, and bus services with approximately 1.3 million unlinked weekday trips. Information in this section comes from an interview with L. Paget-Seekins and A. Gartsman, planning staff at the MBTA, on April 4, 2018. In 2015–2017, the MBTA contracted with Central Transportation Planning Staff (CTPS, the regional MPO) to administer a transit rider survey. The survey was developed primarily to fulfill the MBTA’s Title VI obligation to “collect information on the race, color, national origin, Eng- lish proficiency, language spoken at home, household income and travel patterns of their riders using customer surveys” (FTA, 2012, Chap. IV-8). The data will also inform the regional travel demand model operated by CTPS. This survey is not, however, the only data that MBTA uses to study person movements through its system and plan for service changes. Through a partnership with researchers at the Massachusetts Institute of Technology (MIT), the MBTA has developed a passive and inferred data model called the Origin-Destination-Transfer (ODX) model. ODX Methodology The ODX model utilizes data from the MBTA’s EFC, AVL, and APC systems to produce an inferred matrix of transit trip boardings, alightings, and transfers. The MBTA’s stream of fare- card, AVL, and APC data is stored on servers available to the MBTA and the research team. The model is aggregated each quarter to inform the MBTA’s schedule changes. The MBTA only records fare-card data when a rider validates their card upon entering a vehicle or subway station. This means that there is no fare-card record of transfers within the paid-fare area of a subway station or alightings anywhere within the system. For this reason, the ODX data generation process infers the boarding, alighting, and transfer points of a per- son’s trip by comparing tap-on points throughout a person’s day and applying rules for when a change in service counts as a new trip, or when it counts as a transfer. Figure 14 illustrates part of the logic used by the ODX model. The details of these rules are available in work by Wang et al. (2011) and Gordon et al. (2013). ODX source data do not include passengers who pay cash fares, which represent 3 percent to 4 percent of the MBTA’s riders; the MBTA observes that these riders are not randomly distributed, but are concentrated along certain routes and neighborhoods. MIT researchers have access to the raw and processed data and use it in their own research projects subject to a nondisclosure agreement to protect sensitive passenger information. The MBTA validates the ODX data in two basic ways. First, the ODX generation software produces statistics on the percentage of inferred points that can be tracked and referenced over time; the initial MBTA’s first blog post (MBTA, 2016a) on ODX reported that 97 percent of origin points, 75 percent of destinations, and 92 percent of transfers were inferred directly from

Case Examples 53 the data (the balance are distributed according to the inferred patterns). Second, a group of MIT students volunteered their fare-card information along with precise daily activity patterns generated from the Moves mobile application (ProtoGeo Oy, 2015). With these validation methods, the MBTA feels satisfied in using the ODX data for service and fare planning. ODX and On-Board Surveys The MBTA feels that it cannot fulfill its Title VI obligations with ODX data because they do not observe or impute rider characteristics from the fare-card data. There has been a policy deci- sion to not collect additional demographic information upon fare-card purchase or activation, and they do not feel that joining aggregate census data from inferred origin points would be adequate for meeting FTA standards for Title VI analyses, because the distribution of minority riders may not match the distribution of the minority population. Additionally, a detailed OD survey is the only way that MBTA feels it can study the modes passengers use to access transit Source: Dumas, 2017. Figure 14. Heuristic model used to infer arrival time from fare-card data.

54 Public Transit Rider Origin–Destination Survey Methods and Technologies (walk, park-and-ride, etc.). ODX data do not provide true trip origins and destinations, merely the boarding and alighting locations. That said, the MBTA is considering using ODX data to meet its obligations for reporting unlinked passenger trips and passenger-miles traveled to the National Transit Database. They are working to validate the ODX data against their existing survey-based method and are hoping to receive FTA approval. Additionally, the MBTA has used ODX data to study bus crowding along multiple-route corridors (MBTA, 2016b) and is using it to develop a new service plan (MBTA, 2017) with detailed and updated data that would be unavailable with traditional means.

Next: Chapter 5 - Conclusion »
Public Transit Rider Origin–Destination Survey Methods and Technologies Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s Transit Cooperative Research Program (TCRP) Synthesis 138: Public Transit Rider Origin–Destination Survey Methods and Technologies captures the state of the practice among agencies of different sizes, geographic locations, and modes and evaluates the opportunities for and challenges of conducting surveys in an era of emerging technologies.

The report presents the reality and complexity of conducting origin–destination surveys and will allow agencies to compare what they are currently doing with what others are doing, get ideas about what other strategies are possible, and make better decisions about surveying in the future.

The report includes case examples of five transit systems that present an in-depth analysis of various survey strategies and include two agencies that have leveraged passive data to complement or eliminate origin–destination surveys.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!