Click for next page ( 45


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 44
44 CHAPTER SIX RESPONSE RATES Response rates are critical to both survey quality and survey Some agencies use an estimate of the number of unique costs. High response rates minimize nonresponse error and riders as the base figure. This is appropriate in customer thus reduce the impact of nonrespondents being different satisfaction surveys that do not expect riders to complete from those responding to the survey. High response rates multiple surveys. Thus, LYNX and Pace Suburban Bus used also reduce the costs of carrying out the survey by reducing an estimate of unique riders based on boardings, transfer the number of survey worker hours needed to obtain the tar- rates, and trips per day per customer. Both were one-day geted number of completed surveys and, in the case of self- surveys that attempted to include all riders using the bus on administered surveys, by reducing the number of question- the survey day(s). naires that need to be printed. The calculated response rate is also affected by the choice of numerator. Agencies generally reported the MEASURING RESPONSE RATES number of surveys "completed and returned." As discussed here, some agencies require that every question be Careful thought needs to be given to calculating response answered for a survey to be considered complete, whereas rates and comparing response rates between surveys. Ideally, others set a lesser standard; a choice that affects the response rates are computed as the number of surveys measured response rate. returned and usable (the numerator) as a percentage of the number of riders asked to participate in the survey (the denominator, or base). For example, if 1,000 riders are RESPONSE RATES REPORTED offered questionnaires as they board a sample of buses and BY TRANSIT AGENCIES 400 accept and return their questionnaires, the response rate is 40%. Similarly, if 1,000 riders are approached in a transit Response rates reported by transit agencies vary widely, center for an interview and 600 agree to be interviewed and from a low of 13%, for a survey distributed on-board buses complete the interview, the response rate is 60%. by bus operators in Lodi to 90%, for an on-board bus survey distributed by university students in Ann Arbor. In practice, agencies reported response rates using a variety Within this very broad range, response rates for the major- of different numerators and denominators, as summarized in ity of on-board and intercept surveys ranged from 33% to Table 19. Depending on which counting method is used, and 67%, with one-half of agencies reporting response rates in whether the study population is people or trips, response rates this range. can be difficult to compare on an apples-to-apples basis. Response rates vary not only between agencies and sur- One "base" (denominator) often used is the number of veys, but also between routes and modes for a given transit customers asked to participate in the survey. This base is agency survey. For example, a TARC survey experienced appropriate when the study population is trips (rather than response rates ranging from 23% to 53% among lines with at people) and customers are asked to complete a second or least 400 surveys distributed. third survey if they are encountered by a survey worker a sec- ond or third time. In practice, this base is most workable in The following detailed profiles and Table 20, a summary large systems because customers are rarely encountered of response rates, provide overall response rates and key fea- more than once by a survey worker. Thus, the problem of tures of 29 surveys for which transit agencies reported suffi- riders being disinclined to complete a survey more than once cient information to compute a response rate. Surveys are does not arise. An example of this situation is the SANDAG presented in two groups based on the completeness of the O&D survey in San Diego. count of riders being asked to participate in the survey. The first group is comprised of agencies that reported the total Another often-used base is the number of surveys distrib- number of riders who were offered a questionnaire or were uted. This number is almost always less than the number of asked to be interviewed. The response rates in these cases customers asked to participate in the survey, and therefore therefore reflect both refusals and unreturned surveys. (In a overstates the response rate. few cases, the number of people offered surveys is calculated

OCR for page 44
45 TABLE 19 BASES AND NUMERATORS USED IN RESPONSE RATE CALCULATION Measure Comments Base Fares paid or boardings, If data are collected for period a Riders excluding transfers and other than survey days, this adjusted for number of riders may differ from number of making multiple trips riders on survey days Requires calculation of transfers (when boarding data are used) and number of riders making multiple trips per day b Trips Passenger boardings, Need to weight surveys to measured through farebox or account for transfers automatic passenger counters, adjusted for transfers Asked to Number of customers Must track people refusing participate approached for interview or offered questionnaire Surveys Number of questionnaires Must track number actually distributed distributed to customers taken by customers; this can be difficult to discern if surveys are left in envelops or seats for customers to pick up Respondents Surveys Count of returned surveys May include unusable surveys returned that are not blank with few answers marked Complete Count of returned surveys Strictness of rule for counting surveys that are fully or partially as "complete" will affect completed response rate and data quality a Number of people using transit--each rider asked to complete the survey once. b Passengers asked to complete a survey for each trip. as total daily ridership, because an attempt was made to offer 90% response rate. Refusals included in base number. a survey to every rider.) High response rate attributed to using University of Michigan students as surveyors; students are personable Surveys for which transit agencies did not track the num- and enthusiastic survey workers; incentive; conducting ber of refusals are grouped separately. The true response rate the survey regularly (every 2 years); and the university setting. for these surveys is somewhat lower (probably by 5 to 15 per- centage points) as a result of not counting riders who refused to take a survey. King County Metro Transit Division (Seattle, WA) Surveys in which refusals are included in calculating response rates Ride Free Area (RFA) Survey Short interviews conducted on board buses in downtown AATA (Ann Arbor, MI) Ride Free Zone. Survey workers selected every 3rd or 5th person boarding for very short interview. Rider survey Incentives: None. Self-administered survey on board buses. Survey staff dis- tributed surveys to passengers as they boarded. Surveys 1,899 riders were approached for interviews. 1,663 surveys were returned to surveyors. were completed. Incentives: Pens with agency name, website, and phone 88% response rate. Refusals included in base number. number. High response rate attributed to short, personal inter- 2,700 riders were offered surveys. 2,433 surveys were views and general willingness of riders to participate in completed and returned. surveys.

OCR for page 44
46 TABLE 20 SUMMARY OF SURVEY RESPONSE RATES Survey Percentage completed and distributed Modes Method. by Self-administered Commuter rail Bus operators Survey staff Incentives Interview Light rail returned Subway Bus Agency Project Base Refusals included in calculating response rate AATA Rider survey 2,700 offered 90 Pens King Co. Metro (Seattle) Ride Free Area (RFA) Survey 1,899 approached 88 None R-7 Origin, Destination and Trip SEPTA (Philadelphia) 535 offered 86 None Purpose Study MARTA (Atlanta) Systemwide Survey 5,000 approached 80 None Gulf Regional Planning Customer Appreciation Day 110 approached 73 Food Commission Survey Douglas Branch Blue Line 2,478 distributed + est. CTA 71 None Passenger Survey refusals Intercity Transit (Olympia, Customer Satisfaction 2,672 offered 70 None WS) Lane Transit District 2004 Origin/Destination Study 8,338 offered 63 None (Eugene, OR) Onboard Transit Passenger SANDAG (San Diego) 79,220 offered 54 None Survey Metra (Chicago area) On-Board Survey 50,000 offered 50 None 2001 LYNX Market Research 33,470 est. daily LYNX (Orlando) 45 None Study ridership Santa Clara VTA 2000 On-Board Survey 44,633 offered 41 None

OCR for page 44
TABLE 20 (Continued) Survey Percentage completed and distributed Modes Method. by Self-administered Commuter rail Bus operators Survey staff Incentives Interview Light rail returned Subway Bus Agency Project Base Metrolink (L.A. area) 2004 Onboard Survey 32,960 boardings 41 Drawing TriMet O&D Systemwide Survey TriMet (Portland, OR) 205,000 offered 40 None 2000 Santa Monica Big Blue Bus Line-by-Line Analysis 13,000 offered 36 None Free TARC (Louisville, KY) Project Gobility 12,906 offered 33 ride(s) Pace Suburban Bus (Chicago 58,000 est. daily CSI/User Study 14 Drawing area) ridership Refusals not included in calculating response rate Fort Worth Transportation Customer Satisfaction Survey 500 distributed 80 None Authority Potomac & Rappahannock Full on-board surveys of local Transportation Commission 3,647 distributed 70 None and commuter bus riders (Woodbridge, VA) West Side Customer Travel CTA 8,230 distributed 67 None Survey Spring 2004 On-Board Customer Metro (Los Angeles) 27,280 distributed 52 Drawing Satisfaction Survey Free RTD (Denver) Customer Satisfaction Survey 9,000 distributed 41 ride(s) Orange County (CA) Free 2001 On-Board Survey 25,000 distributed 38 Transportation Authority ride(s) Metro (St. Louis) Metro On-Board Survey 10,000 distributed 35 Drawing 47

OCR for page 44
48 TABLE 20 (Continued) Survey Percentage completed and distributed Modes Method. by Self-administered Commuter rail Bus operators Survey staff Incentives Interview Light rail returned Subway Bus Agency Project Base WMATA (Washington, DC) Metrorail Passenger Survey 207,788 distributed 28 None GCRTA (Cleveland) Annual Onboard Survey 4,000 distributed 23 None CTTransit (Hartford, CT) Bi-annual passenger survey 22,000 distributed 20 None DART (Dallas) Customer Satisfaction Survey 40,000 distributed 14 Drawing City of Lodi Customer Service 400 distributed 13 None Notes: Offered = number of passengers offered a questionnaire. Includes refusals. Approached = number of passengers approached for personal interview. Includes refusals. Distributed = number of surveys distributed. Does not include refusals unless otherwise noted. Boardings = number of riders boarding bus/train. Includes customers who had completed survey previously. Estimated daily ridership = Number of unique customers. Persons transferring and persons taking 2+ trips per day are counted once. See chapter six for detailed information.

OCR for page 44
49 SEPTA (Philadelphia, PA) Intercity Transit (Olympia, WA) R-7 Origin, Destination, and Trip Purpose Study Customer Satisfaction Survey Self-administered survey on board commuter rail trains. Self-administered survey on board buses. Survey workers Survey workers distributed short one-page surveys to distributed two-page (front and back) surveys to passen- passengers as they boarded. Surveys were returned to gers as they boarded. Surveys were returned to surveyors surveyors. and bus operators. Incentives: None. Incentives: None. 535 riders were offered surveys. 460 surveys were com- 2,672 riders were offered surveys. 1,885 surveys were pleted and returned. completed and returned. 86% response rate. Refusals included in base number. 70% response rate. Refusals included in base number (18% of riders boarding refused to take a survey). MARTA (Atlanta, GA) Suburban/rural area, many riders users for many years; riders appreciated being asked their opinion. Systemwide Survey Survey workers interviewed passengers on board buses and subway cars. Every fifth rider selected for a short Lane Transit District (Eugene, OR) interview. 2004 Origin/Destination Study Incentives: None. Self-administered survey on-board buses. Survey workers 5,000 riders were approached for interviews. 4,000 surveys distributed two-page (front and back) surveys to passen- were completed. gers as they boarded. Surveys were returned to surveyors or bus drivers. 80% response rate. Refusals included in base number. Incentives: None. High response rate the result of short, personal interviews. 8,338 riders were offered surveys. 5,528 surveys com- pleted and returned. Gulf Regional Planning Commission (Gulfport, MS) 63% response rate. Refusals included in base number Customer Appreciation Day Survey (10% refused). (Note that 73% of passengers returned survey; response rate is reduced by 10% as a result of Survey workers approached riders as they alighted buses some riders filling out multiple surveys; these were set at transfer station and train, beach, and mall hubs. Sur- aside.) veyors offered riders the option of being asked questions or self-administering the survey; most riders chose to be High response rate attributed to ridership being skewed to interviewed. Self-administered surveys were returned to younger riders, many of whom are university students and survey workers. are likely to complete the survey. Incentives: Soft drinks and cookies offered. 110 riders were approached for interviews. 80 surveys SANDAG (San Diego, CA) were completed. Onboard Transit Passenger Survey 73% response rate. Refusals included in base number. Self-administered survey on board buses, light rail, and Interviewing increased response rates, particularly given commuter rail. Survey workers distributed one-page low literacy rate and demographic groups involved. surveys to passengers as they boarded. Surveys were returned to surveyors. CTA (Chicago, IL) Incentives: None. Douglas Branch Blue Line Passenger Survey 79,220 riders offered surveys. 42,740 surveys were completed and returned. Self-administered survey on board elevated train. Survey workers distributed two-page (front and back) surveys to 54% response rate. Refusals included in base number. passengers as they boarded. Surveys were returned to (Note that 65% of passengers returned survey; response surveyors. rate is reduced by 11% from setting aside incomplete surveys.) Incentives: None. High response rate attributed to use of short, simple ques- 2,230 surveys distributed. About 90% of riders entering the tionnaire and effective survey staff. train accepted a survey. 1,756 surveys were completed and returned. Metra (Chicago, IL) 71% response rate, based on refusal rate of 10%. On-Board Survey Very skillful and experienced survey worker generates high response rates. Focused nature of survey area Self-administered survey on board commuter rail trains. (branch of the Blue Line) may also have encouraged par- Survey workers distributed five-page survey to passen- ticipation. gers after they were seated. Large majority of surveys

OCR for page 44
50 were returned to survey workers or deposited in boxes at TriMet (Portland, OR) downtown terminals; surveys also returned by mail. TriMet Origin Destination Survey--Systemwide 2000 Incentives: None. Self-administered survey on board buses and light rail. 50,000 riders were offered surveys. 25,000 surveys were Survey workers distributed one-page surveys to passen- completed and returned. gers as they boarded buses. On light rail, two survey workers approached riders after they boarded. Surveys 50% response rate. Refusals included in base number. returned to envelopes posted by each exit and by mail. Good response rate attributed to length of most commuter Incentives: None. rail trips and by convincing passengers that results would be used to improve service. 205,000 riders were offered surveys. 81,100 surveys were completed and returned. LYNX (Orlando, FL) 40% response rate. Refusals included in base number. 2001 LYNX Market Research Study Self-administered survey on board buses. One-page Santa Monica Big Blue Bus (Santa Monica, CA) surveys placed in boxes on board buses. Signs and bus Line-by-Line Analysis operators encouraged riders to complete a survey. Self-administered survey on board buses. Survey workers Incentives: None. distributed one-page surveys to passengers as they 33,470 estimated number of daily riders, all of whom were boarded. Surveys were returned to surveyors or bus at least theoretically offered a survey. 15,000 surveys drivers. were completed and returned. Incentives: None. 45% response rate. Total ridership included in base number. 13,000 riders were offered surveys. 4,709 surveys were Bus operators were happy with their contract at that time completed and returned. and many actively encouraged riders to complete the 36% response rate. Refusals included in base number. survey. Note that some riders may have completed multiple surveys, in which case true response rate would be lower. TARC (Louisville, KY) Project Gobility Santa Clara VTA (San Jose, CA) Self-administered survey on board buses. Survey work- 2000 On-Board Survey ers distributed two-page surveys to passengers as they boarded. Surveys were returned to surveyors and by mail. Self-administered survey on board buses and light rail. Survey workers distributed two-page (front and back) sur- Incentives: Free ride ticket with the completion of the veys to passengers as they boarded. Surveys were survey. returned to surveyors or return box at rear exit. 12,906 riders were offered surveys. 4,211 surveys were Incentives: None. completed and returned. 44,633 passengers who had not previously been asked to 33% response rate. Refusals included in base number. complete a questionnaire on the line being surveyed were offered surveys. 18,351 surveys were completed and returned. Pace Suburban Bus (Arlington Heights, IL) 41% response rate. Refusals included in base number. CSI/User Study Self-administered survey on board buses. Bus operators Metrolink (Los Angeles, CA) distributed surveys or made surveys available on bus dashboard. Bus operators chose method to distribute that 2004 Onboard Survey they were most comfortable with. Bus operators were also asked to make announcements and car cards were posted Self-administered survey on board commuter rail. Survey in each bus to announce the survey. Surveys returned to workers distributed four-page surveys to passengers as on-board folder and by mail. they boarded. Surveys were returned to surveyors or by business reply mail. Incentives: Raffle of three 1st prizes--$500 U.S. Savings Bonds and five 2nd prizes--$100 U.S. Savings Bonds. Incentives: Drawing for free monthly passes. 58,000 estimated number of daily riders, all of whom were 32,960 passengers boarded during the surveying, includ- at least theoretically offered a survey. 7,937 surveys were ing many who had previously completed the survey; completed and returned. 14,834 surveys were actually distributed, with 13,470 com- pleted and returned. 14% response rate. Total ridership included in base number. 41% response rate based on total passengers, including those who refused the survey because they had com- Surveys were distributed to bus operators through pleted a survey earlier. Response rate for surveys dis- dispatchers. Not known how many surveys were actually tributed was 91%. distributed to passengers.

OCR for page 44
51 Surveys in which refusals are not included in calculat- RTD, Denver (Denver, CO) ing response rate Customer Satisfaction Survey Fort Worth Transportation Authority (Fort Worth, TX) Self-administered survey on board buses and light rail. On buses, bus operators distributed seven-page surveys (in Customer Satisfaction Survey envelopes with pencils and incentives) to randomly Self-administered survey at bus terminal and four transfer chosen passengers. Surveys were returned by mail facilities. Survey workers intercepted riders and asked (primarily) and to bus operators. On light rail, survey work- them to complete survey. Surveys returned to survey ers distributed 11-page survey to passengers on plat- workers or by mail. Also distributed some surveys on forms. Surveys were returned by mail. board; bilingual survey worker conducted some interviews Incentives: Two free ride coupons included in each survey in Spanish. packet; drawing for grocery gift certificates. Incentives: None. 9,000 surveys distributed. 3,654 surveys were completed 500 surveys distributed. 403 surveys were completed and and returned. returned. 41% response rate. Refusals not included in base number. 80% response rate. Refusals not included in base number. Anecdotally, however, only a small number of passengers refused to take a survey. High response rate for riders who agreed to participate, but difficult to obtain participation in this setting owing to Good response rate attributed to strong incentives, survey lack of time. being conducted periodically, and passengers wanting to provide feedback to the agency. Potomac & Rappahannock Transportation Commission Orange County Transportation Authority (Orange, CA) (Woodbridge, VA) 2001 On-Board Survey Full on-board surveys of local and commuter bus riders Self-administered survey on board buses and at transit Self-administered survey on board buses. Survey work- centers. Survey workers distributed surveys to passen- ers distributed one or two-page surveys to passengers as gers. Surveys were returned to surveyors and by mail. they boarded. (Survey length varied by type of service.) Surveys were returned to surveyors. Incentives: Free ride coupon on survey return. Incentives: None. 25,000 surveys distributed. 9,500 surveys were completed and returned. 3,647 surveys distributed. 2,544 surveys were completed and returned. 38% response rate. Refusals not included in base number. (Note that 56% of passengers returned survey; response 70% response rate. rate is reduced by 18% from setting aside of incomplete surveys.) CTA (Chicago, IL) MetroSt. Louis (St. Louis, MO) West Side Customer Travel Survey Metro On-Board Survey Self-administered survey on board buses. Survey workers distributed one-page surveys to passengers as they Self-administered survey on board buses and at light rail sta- boarded. Surveys were returned to surveyors or by mail. tions. Survey workers distributed two-page (front and back) surveys to passengers as they boarded buses and at light Incentives: None. rail stations. Surveys were returned to surveyors and by mail. 8,230 surveys distributed. Does not include riders refus- Incentives: Respondents eligible for contest drawing. ing to take a survey. 5,200 surveys completed and returned. 10,000 surveys distributed. 3,500 surveys were completed and returned. 67% response rate. Refusals not included in base number. 35% response rate. Refusals not included in base number. Metro (Los Angeles, CA) WMATA (Washington, DC) Spring 2004 On-Board Customer Satisfaction Survey Metrorail Passenger Survey Self-administered survey on board buses, subway, and Self-administered survey in subway stations. Survey light rail. Survey workers distributed surveys to passen- workers distributed surveys on platforms. Surveys were gers. Surveys were returned to surveyors. returned to surveyors and by mail. Incentives: Drawing for 10 free monthly passes. Incentives: None. 27,280 surveys distributed. 14,265 surveys were com- 207,788 surveys distributed. 57,700 surveys were com- pleted and returned. pleted and returned. 52% response rate. Refusals not included in base number. 28% response rate. Refusals not included in base number.