Click for next page ( 13


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 12
12 Without replenishment of the sample to include new riders, matter, question design, and graphics." The experience of panel members over time would reflect only long-time users the synthesis team shows that web-based surveys can be of the system. Therefore, every time the study goes out to useful in different ways depending on the type of study they survey, a sample of new or relatively new riders should be will support: origindestination, customer satisfaction, obtained so that the longitudinal panel reflects the ridership mode choice, planning, or other. These various types of sur- tenure for the transit system. veys will be discussed in the following subsections and spe- cific examples will be cited from project experience to Attrition issues also need to be watched closely and underscore the ways in which using web-based technology addressed by longitudinal researchers, as respondents who can benefit survey design. drop out of a study may be different from those who remain in the study (a form of nonresponse bias). Therefore, it is important to ensure that any respondent attrition sample is OriginDestination Surveys replaced with others of similar characteristics (3). This addi- tional replacement sample for the panel is typically con- Origindestination surveys can be well served by web- ducted along with replenishment. based technology, because when respondents are asked to describe their locations they can be instantly geocoded Owing to the issues of replenishment and attrition, miss- online to a latitude and longitude, making for substantial ing values in the data set are common for longitudinal stud- cost savings compared with other survey methods (e.g., ies. Analyzing longitudinal data with missing values can be Resource Systems Group: New York MTA Bridges & Tun- statistically complex (3). However, if enough of a sample is nels OriginDestination Study 2004; NY State Thruway collected and an analysis of attrition does not show significant Authority Westchester, Rockland, & Orange County Travel bias in attrition (e.g., attrition is found to be mostly random Study 2003; and Florida's Turnpike OriginDestination and not the result of a systematic effect) then it is possible to Study 2003). Two-thirds of respondents to this synthesis analyze the data with those records that are complete (2). The survey who are currently using web-based technology men- additional effort of conducting longitudinal panels, although tioned that they have collected geographical data by means significant at times, allows the transit researcher to gain sig- of the web, coded by latitude and longitude, and the other nificant insight and robustness for their study in comparison one-third noted that they have collected data coded by zip with cross-sectional studies (3). Furthermore, over time these code. Online geocoding is a very difficult technical aspect studies may be less costly, because most of the sample work of web-based surveys and is discussed further in chapters has already occurred and the survey instruments and analysis four and five. Clean geocoded data can be used in geo- routines are already in place, providing researchers with the graphic information system software to analyze and present potential to have a more robust study with lower costs than if information that is often very important to transit research, they were to conduct the study using more typical repeated such as the commuter shed of a station or the number of ori- cross-sectional sampling techniques. gins on the system within each zone. Cross-sectional studies are defined as sampling a cross In recent years, geocoding survey and analysis tools section of the population at a given time. Often, repeated have been used successfully in several major transit mar- cross-sectional sampling of customers is undertaken, where kets on a variety of projects. An example is the Metropol- the same survey is used with a new cross-sectional sample itan Transportation AuthorityNew York City Transit's each time (2). This method is much more common than lon- (MTA NYC) JFK AirportLower Manhattan 2005 study in gitudinal panels, with 63% of synthesis respondents indicat- which survey respondents were asked to provide origin and ing that they conduct repeated cross-sectional studies. destination information using one of three search methods Although differences in satisfaction scores are detected using in the geographic information system component of the repeated cross-sectional studies, the measurement of the dif- survey: by selecting a location on a map, by entering a spe- ference may be confounded owing to differences within the cific address, or by entering a nearby intersection. By sample itself, because of demographic differences or some clicking on the mapping option, the respondent is shown a other nonquantifiable difference between individuals. Cross- map of the local area and simply clicks on the area of his sectional studies require a larger sample than longitudinal or her location to indicate where the trip began or ended studies to measure changes over time. (see Figure 6). The map zooms in one or two times, enabling the respondent to select an exact location that is instantly assigned a latitude and longitude in the project AREAS WHERE WEB-BASED SURVEY database. This option enables the respondent to indicate TECHNIQUES ARE MOST EFFECTIVE the location relatively easily and allows researchers to Web-Based Technology's Effect on Survey Design screen the response (i.e., the geocode must reside within the study area or the respondent will be screened out of the Reasons cited for using web-based surveys in the synthesis survey). The system automatically geocodes the location survey were "the ability to present complicated subject in real time, thereby avoiding the need to geocode later,

OCR for page 12
13 FIGURE 6 Screen shot of geocoding technique on the JFK AirportLower Manhattan 2005 study. which is frequently based on erroneous word descriptions Mode Choice Surveys of geographical data. A mode choice study can be difficult to do using paper-based Another way origindestination surveys can be enhanced survey methods, particularly for stated preference surveys. for data validation is by using web-based technology to show Mode choice can be evaluated much more efficiently using maps of transit systems and linking them directly with the computer-based technology because customized branching schedules of specific lines and stations. An example of this can obtain a clearer picture of each distinct respondent's can be seen in a series of screen shots captured from NJ choices based on his mode path; and then realistic alternative TRANSIT's 2003 Rail ePanel study. In this survey the scenarios can be constructed to understand the respondent's respondent was first asked which commuter rail line they used behaviors to variables such as time, cost, and comfort. As (see Figure 7). will be discussed in chapter four, offering the survey by means of the Internet can increase response rates over the Once a specific line was selected (color coded to match NJ survey offered only to those respondents who can be TRANSIT's schedules), the respondent was directed to a page recruited in person. One respondent stated that a web-based showing only the stations on that rail line (see Figure 8). Each survey can be "an easy tool for the end user and our staff to rail line's train schedule has been processed into a database gather data on work trips for employees at large employers with exact times and stations, for weekdays and weekends, for in the county." the entire system. When the respondent chose his particular station, the schedule data that was linked to the survey offered Planning and Other Surveys the respondent only actual train times and train numbers available (Figure 9). Offering correct available train times and Respondents indicated that for planning surveys, web- numbers is one example of how web-based technology can based surveys are beneficial "as a way to gather public help improve data quality and, in this case, decrease item input on our planning studies, in addition to holding pub- nonresponse in surveys. A discussion of item nonresponse lic meetings which are usually poorly attended." The abil- follows in chapter four: Item Nonresponse in Web-Based ity to quickly and easily reach out to the public, provided Surveys. Problems resulting from guessing and/or faulty agencies have a satisfactory list of e-mails and/or a well- memory on the part of the respondent are therefore mitigated, publicized website, is another benefit to using web-based resulting in clean data for the planners at NJ TRANSIT. surveys.

OCR for page 12
14 FIGURE 7 Screen shot requesting commuter rail line from NJ TRANSIT's Rail ePanel survey. FIGURE 8 Screen shot to select boarding station of chosen rail line from NJ TRANSIT's Rail ePanel survey. A variety of uses for "other" surveys were also noted in chapter six. Another "other" type of survey mentioned by a the synthesis survey. One agency researcher described an respondent was a household travel survey, and this type of interactive map study that had been conducted where they survey can benefit greatly by being conducted online. First, needed to "solicit customer feedback on their experiences respondents have a difficult time remembering all of their with the interactive map" on their trip planning section of daily trips for an assigned travel day. With web-based tech- their website. Using a web-based survey, they were able to nology, respondents can be prompted to include all trips by "determine if there are any fatal flaws that need immediate simplifying data entry. Depending on what the respondent attention." This particular study is detailed as a case study in describes for activities and/or purposes, they can be shown

OCR for page 12
15 FIGURE 9 Schedule page from NJ TRANSIT's Rail ePanel survey. customized screens and drop-down boxes on those screens. The respondent can be prompted to enter all trips for the For example, if a respondent starts out a trip from home to survey day, and can be shown various trip purposes for each work by walking, he or she can be shown a drop-down box trip in drop-down boxes. In the example shown here, the with a variety of choices for the second mode on their trip to respondent went to work at his construction site, then went out work (see Figure 10). for lunch with five friends. Each trip requires a start and an end FIGURE 10 MI Travel Counts (Michigan DOT) activity input page.