Deployment, Use, and Effect of Real-Time Traveler Information Systems (2013)

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.

15 C H A P T E R 2 Research Approach The research approach of this study used a variety of methods to explore outcomes from a traveler perspective including access to, perception of, and need for real-time information. It also used interviews and surveys to understand the dissemination practices of agencies, as well as performance measurement of TI systems. Described in more detail below, the current study used a variety of low cost methods including interviews, surveys, focus groups, and traveler logs in addition to analytical tasks. This approach allowed us to not only understand satisfaction and perception, but also behavioral outcome proxies (through self-report logs and recent timeframe surveys). This combination (i.e., “toolbox”) of methods creates a better, more thorough assessment of effectiveness of real-time TI systems and practices. It also provides a better pool of information from which to develop guidance for future approaches to real-time TI. This section highlights several methods that were used in the project: interviews, surveys, focus groups, and traveler logs, along with analytical research for development and synthesis of information. Questions and Variables of Interest Several questions of interest were identified for the current study to explore: • What access do travelers have to real-time TI and what is their awareness of that access? • What is the perception of the public of real-time TI? • What characteristics do travelers need in real-time TI? • What are the gaps between what agencies provide and what travelers feel is important and need? • How do agencies use real-time data in planning and operations activities and are these uses reaching the public with the intended effect? • What are the reasons travelers do not use available real-time TI? • Are there any self-reported behavioral changes due to real-time TI? If so, what are the most effective information type and medium to be used in changing traveler behavior? How do pre-trip and en-route patterns of using real-time TI differ? • How frequently do these changes occur? • In what ways do the changes occur (e.g., leave earlier, later, change route)? • If behavior does not change, then why not? Is it a system constraint (e.g., lack of alternative routes/transit options, unreliable information) or personal constraints (inability to adjust trip time due to appointments, etc.)?

16 In addition, several variables of interest with respect to traveler’s perception of, access to, and need for real-time TI were selected for the current study: • Traveler type (e.g., commuters, high and low TI users, different ages); • Information type (e.g., travel time, congestion levels, incident information, weather, road closures); • Media used (e.g., mobile device, computer, radio, TV); • Temporal information need and use (e.g., pre-trip, en-route); • Traveler perception of importance; • Effect on traveler behavior; • Likeability/satisfaction; A particular area of interest for transportation managers and engineers is changing traveler behavior based on information disseminated. Ideally, one would like to have extensive empirical data on the actual changes in traveler behavior as a function of various types and sources of TI to directly measure the effect (and relative effectiveness) of TI. This is in fact quite difficult to do, particularly with limited project resources. Among a variety of reasons, the difficulty is associated with: • Travelers may have available a broad range of pre-trip and en route information (websites, radio and television, mobile devices, on-board devices, roadway signs, commercial signing, direct observation of traffic, and, in the future, emerging new sources through programs such as Connected Vehicles). These various sources may be operated by different entities and may be continually evolving in an uncontrolled way. All may be contributing to behavior in some way, and this confound makes it very difficult to evaluate the behavioral effects of one particular source of information. • The effects of real-time TI are not expected to be uniform over time or between locations. Changes in driver behavior may be meaningful only under certain conditions (e.g., only when an alternative route or mode is available), which limits when and how data are collected and analyzed and makes it difficult to quantify “driver response.” • The effects of real-time TI are also not expected to be uniform between individuals, or even between different trips for the same individual. This may be affected by route familiarity, personal characteristics, trip motivation, etc. • The effects of real-time TI will be dependent on the characteristics of the local roadway network, as well as local options for other modes of travel. • The effects of real-time TI may be expressed in a variety of ways: pre-trip route planning, en route diversion, trip timing, trip chaining, mode choice, cancelling the trip, vehicle occupancy, etc. Any given measure may in itself not fully capture the effects of the information. The Westat-CATT team recognized these pragmatic concerns and our approach aimed at deriving maximum useful information and insights within the scope of the resources. We recognized that the objective of this project should not be a large-scale survey that devotes most of the project resources toward a narrow range of sites or practices. However nice a statistically precise large survey might be, it would overly limit the range of practices and conditions we could examine. Information at each site may come from traffic observation, small-scale surveys, commuter daily trip logs, driver focus groups, practitioner feedback, system performance statistics, system use, etc. Available information may vary from site to site. Our approach is to learn what we best can about multiple aspects of TI effects from a diverse set of sites. Multi-modal considerations: For some applications of real-time travel time provision (e.g., changeable message signs), information relating to other modes is largely irrelevant in the United States (although we have seen European examples). For other applications (e.g., certain websites), information on additional modes may be appropriate and useful. The project focus was on the provision of real-time TI to drivers. We are fully aware of multi-modal considerations and attended to them as feasible, but they were not the focus.

17 Target traveler: Many types of travelers may make use of real-time TI. In previous work with providers of TI, we have found that the needs of commuters are generally the operational agency’s primary concern, particularly at those times when congestion is likely to occur (morning and evening peaks). However, in some areas tourists are a predominant concern, and for special events the target users might be quite different (e.g., rock concert, golf tournament). These various target users differ substantially in the information they need, willingness to change routes or modes, flexibility in travel options, local knowledge of the roads and transportation system, and ability to comprehend/use various messages and data. Consequently, while our assessments took into account the effectiveness of various TI provision practices across the range of users, the greatest emphasis was placed on commuter needs and behavior. Materials and Methods Because there is not a single best technique for a project of this size, the Westat-CATT team believed the best approach to understanding travelers’ use, perception, and response to real-time TI was the combination of a variety of low-cost methods that have proven successful in the past. This “toolbox” consisted of several components which were inexpensively adapted to a particular site. Also, these components were tailored to the particular region/site and built on findings from agency interviews. This approach also has the advantage of not being cost prohibitive for agencies who would like to develop their own evaluation process and use some of the same tools. Several types of instruments and materials were used for data collection, with selected draft sample documents included in Appendices C-F: a) focus groups including a moderator’s guide, notes, and audio recording, b) traveler logs completed by participants, c) web-based surveys completed by participants through SurveyMonkey, d) telephone interviews of agency representatives. The following sections describe the basic approaches taken to ensure highest quality implementation. Agency Survey A draft web-based survey was disseminated to the Expert Panel members and comments were incorporated prior to its distribution. An e-mail with a link to the survey and a paragraph description of its importance was sent out to 50 contacts representing 40 state DOTs. See Appendix B for the survey questions and responses. On the same day, an e-mail was also sent out to the graduates of the Operations Academy Senior Management Program asking that they have someone in their office who works on their agency’s TI program assist by completing the survey. Similarly, an e-mail with the survey information was sent out to the members of the 511 Coalition LISTSERV. A total of 28 fully completed and 14 partially completed surveys were obtained. Subsequently, the survey was sent out again with a request to those who had not initially filled out the survey to please do so. An additional six fully completed surveys were obtained. Agency Interviews Survey results were compiled and, based on this compilation, five follow-up phone interviews were held to gather additional detailed information from individuals who had filled out the web-based survey. The selection of individuals for follow up phone interviews was largely based on those whose agencies indicated efforts to evaluate the TI program activities and/or had interesting responses and comments related to their programs.

18 Overall Traveler Assessment and Implementation Approach The detailed assessment focused on four sites listed in detail below. Each site contained the following data collection approaches and used in forming a comprehensive profile of traveler behavior and perceptions in the area (in this order): • Reference to agency interviews and surveys from earlier in the project, as well as use of Geographical Information Services to target optimal corridors that contained heavy traffic and potential for alternate routing. • Web-based surveys with participants primarily targeted through zip codes (using online databases and postal mailouts) in corridors of interest. • Two focus groups at each location: One group of real-time TI frequent or heavy users (~eight) and one group of real-time TI non-users or light users (~eight). • Traveler logs: Participants in each group were given a seven-day traveler log at the completion of each focus group, along with instructions for its completion. To supplement the more detailed assessment sites, there were two additional sites where the research team did not visit in person but collected data utilizing web-based surveys. Site Selection Site selection took into account a variety of factors, including roadway and regional characteristics, alternative routing opportunity (including public transportation), diversity of travelers, level of congestion, input from earlier agency data collection, panel member suggestions, etc. In order to optimize information gained from each location, we coordinated sites with those used in the agency dissemination survey and interviews, which will allowed us to develop a better picture of the RT-TIS from the users’ and agencies’ perspectives. The four detailed assessment (surveys, focus groups, and traveler logs) sites were: • Washington DC metro area (Rockville, Maryland)—This was the pilot site due to its geographical convenience, research team familiarity, and diversity of travelers/commuters/tourists • Orlando, Florida—Particularly attractive for potential tourist issues • San Francisco, California—Constrained commuting corridor options and highly developed TI services • Teaneack, New Jersey (New York city commuting area)—Strong public transportation options The two remotely assessed (survey only) areas were: • Detroit, Michigan • Salt Lake City, Utah

19 Figure 1. Site locations and data collection methods. These sites were selected due to the commuting characteristics, including a mix of daily commuters and tourists in some areas and the systems in place that are available for receiving TI. See Appendix A for a listing of TI sources and types by city. They also all had high levels of congestion, which was important because that could provide an impetus for trip changing behavior and also allow us to ask about TI use during non-peak times (which may mimic less congested and more rural areas). Taken together, they provide a variety of characteristics, regions, user types, traffic patterns, transit options, and real-time TI approaches. Sample Characteristics and Size The sample size and characteristics for each methodological approach follows: • Focus groups/traveler logs—there were 74 total travelers in the focus groups, and 61 individuals completed and retured traveler logs. • Participants were selected to balance gender and distribute ages across a range from 21-78 years of age. • Both drivers and non-drivers (e.g., transit commuters and carpool passengers) were included. • Smartphone users were oversampled and included in both groups for each city. • Both drivers and transit commuters were included in groups, with the NY metro area and San Francisco metro areas having more public transit users represented. • Traveler surveys—there were a total of 1021 travelers over the age of 21 who lived in the six sites who responded to the surveys. • There were 194 complete responses for the Salt Lake City, UT and Detroit, MI sites.

20 • There were a total of 317 complete responses for the survey targeting Orlando, FL, San Francisco, CA, Teaneck, NJ, and Rockville, MD. • A third survey focusing specifically on recent trip changing behavior was deployed in all six metro areas. There were a total of 522 respondent who completed this survey (with 72 qualifying for the trip changing questions of interest). • Agency surveys—Forty-three agencies responded to a web-based survey regarding TI practices. Twenty-eight surveys were fully completed and 14 were partially completed. Survey respondents were recruited from existing state DOT contacts, colleagues of graduates of the Operations Academy Senior Management Program, and members of the 511 coalition LISTSERV. • Agency interviews—five follow-up phone interviews were held to gather additional detailed information from individuals who had filled out the agency web-based survey. Focus Groups A total of eight focus groups, consisting of eight to ten travelers in each, were conducted in the four metro areas outlined below. Each focus group was facilitated by a trained moderator and lasted for approximately two hours. Focus groups were held during evening hours as commuters were the target population. Two groups were conducted in each city – one group of non- or light users of TI and one group of heavy users of TI. Also, smartphone users were oversampled and included in both groups for each city. Both drivers and transit commuters were included in groups, with the NY metro area and San Francisco metro areas having more public transit users represented. Public transit options were less utilized in the Orlando area. Focus groups were held at equipped facilities in the following cities: • Rockville, MD (Washington, DC metro area) • Teaneck, NJ (New York City metro area) • Orlando, FL • San Francisco, CA Facilities were selected based on recommendations from others who had used focus group facilities in the areas studied as well as based on location being in the metro area of interest. Facilities were also chosen that were equipped with recording capabilities as well as respondent databases to select the participants. Participant recruitment was completed through two channels depending on the focus group location. For all locations, the same telephone screener was administered to potential participants. For the focus group pilot location in the Washington, DC metro area, recruitment of participants took place through an ad placed within Westat. The ad excluded Westat employees but allowed family and friends of employees to participate. Participants who called in or emailed were administered a telephone screener to determine demographic criteria and commuting habits. Eligible participants were selected to include groups with a balanced gender and age distribution across the lifespan. In the New York City metro area, San Francisco and Orlando metro areas participants were recruited from participant databases through the facility being used to hold the group. Westat provided the telephone screener to the facility as well as detailed information as to demographic criteria and commuting habits criteria for recruitment. The facility administered the screener and eligible participants were selected to include groups with a balanced gender and age distribution across the lifespan. Once participants were identified and selected, Westat received a list and checked to ensure that the participants met the provided criteria. The recruitment screener administered included but was not limited to questions regarding daily commute characteristics including distance traveled, AM and PM rush hour commuting, smartphone use for TI and TI sources used prior to or during a trip (i.e., websites, freeway signs and news).

21 Participants signed informed consent forms upon arrival at each of the groups. The facilitator followed the moderator’s guide to direct the discussion (see Appendix C for the sample guide). In addition, the sessions were recorded for data analysis and reporting. Notes were cleaned and compiled following each focus group session (see Appendices G-K for note summaries). At the end of the focus group session, participants were asked to complete brief trip logs for seven days following the session (see section 2.2). Trip Logs Focus group participants were asked to complete trip logs for seven calendar days following their focus group session. Participants were asked to complete the logs for trips of at least five miles of travel and were asked to complete the logs immediately following each trip if possible. The trip logs collected basic information such as weather, trip start time, and end time. Trip logs also contained questions regarding TI sources used both pre-trip and en route (i.e., TV, radio, 511 call, smartphone application, navigation device, or electronic travel time sign) and questions regarding the behavioral effect of receiving that information (i.e., did they change their route, change trip time or cancel a trip stop or the trip altogether, change trip destination, use a different type of transportation or let someone know expected arrival time). A sample trip log is presented in Appendix D. At the end of the seven days participants mailed in completed trip logs to the researchers using the pre-addressed, postage paid return envelopes received upon exiting the focus group session. The return rate for completed trip logs was approximately 83%. Traveler Surveys A survey was developed through SurveyMonkey that targeted travelers in the Detroit, MI and Salt Lake City, UT metro areas. There were 194 complete responses for this survey. A second identical survey was administered to travelers in the focus group cities of Orlando, FL, San Francisco, CA, New York City metro area, and the Washington, DC metro area. There were a total of 317 complete responses for the second survey. A third survey focusing specifically on recent trip changing behavior was deployed in all six metro areas. There were a total of 510 respondents who completed this survey (with 71 qualifying for the trip changing questions of interest). Prior to launching the surveys they were piloted among staff for time and suggestions to increase clarity. The surveys took approximately 10 minutes to complete and paralleled many of the questions asked as part of the focus groups. Also, there were opportunities for open-ended responses for certain questions, and some of those responses are incorporated below. See Appendices E and F for the survey questions.