Click for next page ( 53


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 52
52 GCRTA (Cleveland, OH) 13% response rate. Refusals not included in base number. Annual Onboard Survey Relatively low response rate attributed to use of mail for return of survey and the survey format, which may have Self-administered survey on board bus and light rail. Sur- appeared similar to comment card. Agency expects higher vey workers distributed four-page surveys to passengers response rate with on-board return of surveys and revised as they boarded. Surveys returned to survey workers and formatting. Distribution by drivers is an asset; passengers by mail. know drivers by name. Incentives: None. 4,000 surveys distributed. 935 surveys were completed and returned. FACTORS AFFECTING RESPONSE RATES 23% response rate. Refusals not included in base number; Many factors affect the response rate achieved for on-board number refusing tends to be small. and intercept surveys. Primary factors based on a compari- Experience of agency staff is that response rate with City son of response rates across different surveys and interviews Year survey workers used in this project (as well as other with transit agency personnel can be divided between people temporary workers) tends to be about one-half the response rate when using agency employees. factors and methodological factors. "People factors" include the enthusiasm and diligence of survey workers, rider inter- est in responding to transit surveys, and rider demographics. CTTransit (Hartford, CT) Methodological factors include whether surveys are self- Bi-annual passenger survey administered or conducted as personal interviews, the venue (on-board or in-station), length and complexity of the ques- Self-administered survey on board buses and at transit centers. Surveys were distributed on board buses from tionnaire, use of incentives, and the regularity with which envelopes; bus operators encouraged riders to take and surveys are conducted. complete a survey. Agency staff distributed surveys to rid- ers at bus stops and transportation centers. Survey returned on board and by mail. One-day blitz in each divi- Enthusiasm and Diligence of Survey Workers sion of CTTransit. Incentives: None. The importance of the survey staff and diligence with which 22,000 surveys distributed. 4,500 surveys completed and they go about their jobs cannot be understated. Agency staff returned. repeatedly cited survey staff as a primary factor in explaining both unusually high and unusually low survey response rates. 20% response rate. Refusals not included in base number. It is more difficult than one might expect to generalize DART (Dallas, TX) about who makes for effective survey staff aside from proper Customer Satisfaction Survey training and supervision. The experiences with student work- ers, for example, is quite varied. AATA found that Univer- Survey workers distributed surveys to passengers at tran- sity of Michigan students were personable, enthusiastic, and sit centers. Also used seat drops on buses and rail at start of trips, and stocked timetable holders on board buses and effective as survey workers. AATA's bus survey achieved a light rail. Surveys returned by mail and to operators. Major- response rate of 90%, owing in part to the effectiveness of the ity of surveys were returned by mail; some also returned students as well as a small incentive (pens) and riders' will- to bus operators. ingness to participate. CTA has also had good experience Incentives: Drawing for a monthly pass. with recruiting students and other part-time workers to dis- tribute and collect surveys. Conversely, Metrolink moved 40,000 surveys distributed. 5,950 surveys completed and returned. away from using student survey workers owing to accuracy and reliability problems. 14% response rate. Refusals not included in base number. Response rate affected by mail return. Response rates are sometimes relatively low when bus operators distribute the questionnaires. Bus operators are pri- City of Lodi (Lodi, CA) marily focused on their regular duties and may give less attention to encouraging riders to complete a survey and may Customer Service Survey be less able to provide assistance to riders in completing the Self-administered survey on board buses. Distributed by survey. CTTransit, Pace Suburban Bus, and the city of Lodi drivers on fixed-route buses and returned by mail. Distrib- experienced response rates of 13% to 20% using bus opera- uted by drivers and by mail to dial-a-ride service riders and tors to distribute on-board surveys. returned by mail. Incentives: None. However, RTD and LYNX achieved response rates of 41% 400 surveys distributed. 50 surveys completed and to 45% using bus operators to distribute surveys--comparable returned. to many on-board surveys distributed by dedicated survey

OCR for page 52
53 workers. RTD's response rate is particularly remarkable given Self-Administered or Personal Interviews the length (seven pages) of the survey. RTD attributes the high response rate to strong incentives that included two free Not surprisingly, personal interviewing tends to generate bet- ride coupons in each survey packet and passengers' desire to ter response rates than self-administered surveys. Interview- provide feedback to the agency. In addition, RTD conducts ing provides more individual attention to respondents, less on-board surveys on a periodic basis, conditioning riders to respondent burden to complete the questionnaire, and assis- the survey process. LYNX attributes its high response rate to tance with understanding questions. Interviews also are more bus operators actively encouraging riders to take and complete likely to involve shorter questionnaires than self-administered a survey. surveys, which can also influence the response rate. Short, personal interviews conducted by King County Metro, Whether students, part-time or full-time survey workers, MARTA, and Gulf Regional Planning Commission achieved or bus operators handle the task, it appears that having an response rates of 80% or higher. individual actively distributing and collecting questionnaires is important. A DART survey was distributed, in part, by Interviewing is particularly effective when riders lack leaving questionnaires on seats and in timetable holders (as English proficiency. In the Gulf Regional Planning Com- well as some distribution by survey workers). The DART mission survey, surveyors offered respondents the option of survey response rate was 14%. being interviewed or completing the questionnaire them- selves. Most riders opted for the interview. Agency staff reported that offering the interviewing option was particu- Rider Interest in Responding to Transit Surveys larly important for riders with low literacy skills. As mentioned for AATA and RTD, rider interest can be just Venue (On-Board or In-Station) as important as survey worker enthusiasm. To some extent, rider interest is a function of their view of the agency and ser- The importance of venue is not that on-board or in-station vice and perhaps demographic characteristics. However, venues tend to produce higher response rates. Rather, venue there are steps that agencies can take to increase interest, affects the ease of approaching riders and the amount of time including advance notification of the survey and an explana- that riders have to complete the survey. In many cases, the tion of how the survey will benefit them through improved on-board environment offers the better venue for these very service. Publicizing survey results may also spur interest; reasons; riders can be easily approached as they board the AATA reports that its customers like to complete the survey bus or railcar. They usually have sufficient time during the and see the results. trip to complete a questionnaire. The on-board environment can pose problems, however. Rider Demographic Characteristics The primary problem is trip length--riders taking short trips may not have time to complete a questionnaire. On train cars Certain groups tend to be consistently more likely to respond with multiple doors, survey workers may be challenged to to surveys. Response rates tend to be higher among express keep track of which riders have entered at each station, for bus, light rail, and commuter rail riders than bus riders. For purposes of offering a survey. example, response rates were in the 40% to 50% range on local bus routes in Denver and Dallas, but more than 60% for In-station locales can offer distinct advantages. For exam- express and regional bus routes in Denver and on light rail ple, in the Gulf Regional Planning Commission survey lines in Dallas. Mode and express versus local service reflect agency staff offered food to respondents who completed respondent income and length of the ride, both of which are an interview or survey. Offering food is more practical in a positively correlated to response rates. station than on board a bus. Some agencies report that frequent riders, long-time rid- Length and Complexity of the Questionnaire ers, students, and tourists are inclined to have relatively high response rates. In Ann Arbor and Eugene, Oregon, In mail surveys, length tends to be inversely related to response large student populations contributed to relatively high rates (Dillman 2000). One would expect the same to be true for survey response rates. on-board and intercept transit surveys. Indeed, SANDAG and SEPTA survey response rates were high in part from the use of Short routes, which do not allow time for riders to com- a short questionnaire (see Appendix C). At the other end of the plete a survey, are more likely to experience lower response response rate spectrum, GCRTA attributed a 23% response rate rates. Riders who lack English proficiency also tend to be to the need for mail-back return of a four-page questionnaire. less likely to complete surveys. Some agencies substituted personal interviewing for self-administered surveys to Long questionnaires, however, do not necessarily result in address low response rates among these groups. low response rates. Excellent examples are the four- to

OCR for page 52
54 seven-page surveys for RTD, Metrolink, and Metra. This Incentives shows that response rates are a function of a combination of factors and not simply survey length. In several cases, incentives appear to have had a significant impact on obtaining good response rates. Transit staff in Ann The threshold for considering a survey complete and Arbor and Denver reported that the distribution of pens and usable also affects response rates. Some agencies considered free rides, respectively, stimulated response rates. On the a survey complete if three or more questions were answered, other hand, TARC and Orange County Transportation whereas other agencies used only surveys with complete Authority surveys in which free rides were offered do not O&D data. The criteria for "complete and usable" depend on show higher response rates than similar surveys in which no survey purposes. For surveys that include O&D questions, incentives were offered. the definition of complete frequently depends on whether the O&D information is the central survey purpose or one of sev- eral purposes, in which case surveys with incomplete or Regularity with Which Surveys Are Conducted unusable O&D data still have value for the analysis. Several agencies reported that conducting surveys on a regular Differences in the definition of complete can significantly basis generally improves response rates. Riders learn to expect affect the reported response rate. A VTA O&D survey used a survey every year or two. Riders also become educated about only those surveys in which all questions were answered; the the purpose of the survey. If riders see improvements to service response rate was 41%. LYNX reported a slightly higher that may be attributed to feedback through surveys as well as response rate of 45%. However, if the LYNX survey had other means, they are likely to feel empowered and thus more counted only those surveys with complete O&D information, likely to respond. the response rate would have been just under 30%. Similarly, the CTA West Side survey response rate would be 48% On the other hand, agencies mentioned that surveying instead of 67% if only surveys with complete O&D infor- riders too often induced respondent fatigue, with a negative mation were counted. impact on response rates.