Effectiveness of Different Approaches to Disseminating Traveler Information on Travel Time Reliability (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.

71 The objective of the open-ended survey was to determine whether drivers could correctly interpret both an indicator of normal trip time and an indicator of reliability (buffer time or total trip time estimate). Conceptually, a driver’s average or normal travel (trip) time plus the buffer time the driver adds for contingencies (uncertainty of conditions) equals the driver’s estimated total trip time for planning purposes: normal trip time buffer time total trip time estimate+ = TTI researchers developed the open-ended, paper-based survey to determine which combinations of terms represent- ing a normal trip time, a buffer time, and the ultimate trip time to plan for would be best for conveying trip time reli- ability information. research Method Survey Design The following potential terms were tested for comprehension and preference in conveying the normal travel time at a par- ticular time of day: • Average travel time; • Estimated travel time; • Expected travel time; and • Typical travel time. Terms tested to represent the buffer time for a given trip (as in the preceding equation) included the following: • Added time; • Cushion time; • Extra time; and • Recommended cushion time. Finally, the terms tested to represent the total trip time that drivers should plan for to protect themselves against being late (i.e., “total trip time estimate” in the preceding equation) were the following: • 95th percentile travel time; • The majority of the time the travel time is xx minutes or less; • Most of the time the travel time is xx minutes or less; and • Travel time for planning. These terms were based on focus group results and the expertise and experience of the TTI research team, or they were being evaluated in the usability studies. Some terms initially considered in the focus groups and usability surveys were not evaluated here to keep this part of the study short enough to mesh with the current surveys and experiments already in progress at the time. The terms worst-case and maximum were not tested here, as agency concerns regarding liability and cred- ibility would likely preclude them from using such absolute terms. Similarly, the term most common could be perceived by motorists in a way that would reduce agency credibility and so was not evaluated in this portion of the surveys. Negative feed- back from focus groups led researchers to remove the terms buffer and leeway from analysis. The term departure window was considered to be adequately examined in the existing usability study. The term added travel time was not included in the initial survey and was not offered by focus group partici- pants; however, researchers believed that participants might not have suggested it because a version of the term was used by facilitators to explain the concept. Thus, added travel time was evaluated here. Researchers presented terms to participants in the context of two scenario-based, multipart questions. Scenario 1 pre- sented participants with a term for the normal trip time and a term for the buffer time (as defined in the above equation). Scenario 2 presented participants with a term for the normal trip time and a term for the total trip time estimate. Because of the large number of terms to be tested in these scenarios, C H a P T e r 8 Open-Ended Survey

72 not every term could be tested in every combination with every participant. A total of 16 versions of the survey were created, as shown in Table 8.1. Participants were divided across the 16 survey versions, with 11 or 12 participants taking each version. Participants The open-ended survey was administered to participants who had completed either the computer survey or travel behavior laboratory experiment. Thus, a total of 189 participants com- pleted open-ended surveys in five cities. Table 8.2 lists the num- ber of open-ended survey participants in each city. Detailed demographic information on the participants was not recorded for this survey but was believed to be representative of the driving populations in each location. Procedure A researcher administered the open-ended survey to partici- pants who had completed either the computer survey or the travel behavior experiment. The researcher read the ques- tions to the participant and recorded the participant’s answers. The individual interviews took place in a separate room, away from the computer testing, and took 15 min to 20 min to complete. For some of the questions, cards describing the travel sce- nario were placed in front of the participants to lessen their working memory load. The interviewer read aloud the text on the card, and the participant was allowed to re-read the text if desired and examine the questions and answer alterna- tives while discussing the item. The survey materials from Version A are presented in Appendix E. Survey participants were presented with two hypotheti- cal scenarios. Each scenario placed participants in a hotel in the morning, and their destination was a meeting in a downtown office building. They were able to get travel time Table 8.1. Experimental Design for Open-Ended Survey Survey Version Scenario 1 Terms Scenario 2 Terms Number of ParticipantsNormal Trip Time Buffer Time Normal Trip Time Total Trip Time Estimate A Average Cushion Expected 95th percentile 12 B Average Extra Typical Trip time for planning 11 C Average Added Estimated Most of the time 12 D Expected Recommended cushion Average Majority of the time 12 E Expected Cushion Average 95th percentile 12 F Expected Extra Typical Most of the time 12 G Expected Added Estimated Trip time for planning 12 H Typical Recommended cushion Expected Majority of the time 12 I Typical Cushion Average Trip time for planning 12 J Typical Extra Expected 95th percentile 12 K Typical Added Estimated Most of the time 12 L Estimated Recommended cushion Typical Majority of the time 12 M Estimated Cushion Average Most of the time 12 N Estimated Extra Expected Trip time for planning 12 O Estimated Added Typical 95th percentile 11 P Average Recommended cushion Estimated Majority of the time 11 Table 8.2. Number of Open-Ended Survey Participants per City City Participated in Computer-Based Survey Participated in Travel Behavior Experiment Total Participants per City Dallas 31 14 45 Denver 0 0 0 Miami 0 0 0 San Jose 54 20 74 Hartford 53 17 70 Total 138 51 189

73 information from a hotel website before they left the hotel for the meeting. Scenario 1 read as follows. Blank spaces were replaced with one of the four terms being evaluated to convey the normal trip time and with one of the four terms representing the buffer time. Imagine that you are in a hotel in an unfamiliar city and have an important business meeting first thing this morn- ing at a downtown office building. You will need to leave the hotel during the peak period in order to arrive at the business meeting. You are trying to decide how much travel time to allow for your drive to downtown. You cannot be late to the meeting, but you do not want to arrive too early. You know that you can park right next to the building and do not need to allow additional time to search for parking. You also know that you can get current travel time infor- mation from the hotel website. Before you drive to your downtown meeting, you check the website and obtain the following information . . . The _______ travel time to downtown is 40 minutes and the _______ time to downtown is 30 minutes. As shown in the following text, scenario 2 was presented in an identical format but with the normal travel time to down- town set at 30 min instead of 40 min and a total trip time estimate (representing the sum of the normal time and a buffer time) set at 55 min. Again, the blanks were replaced with one of the four terms being evaluated to convey the nor- mal trip time and one of the four terms representing the total trip time estimate. As noted in Table 8.1, for each survey ver- sion, different terms were used to represent the normal trip time in scenario 1 and scenario 2. Imagine that you are again in a hotel in another unfamiliar city and have an important business meeting this morning at a downtown office building. As before, you will need to leave the hotel during the peak period in order to arrive at the business meeting. You are trying to decide how much travel time to allow for your drive to downtown. You can- not be late to the meeting, but you do not want to arrive too early. You know that you can park right next to the build- ing and do not need to allow additional time to search for parking. You also know that you can get current travel time information from the hotel website. Before you drive to your downtown meeting, you check the website and obtain the following information . . . The _______ travel time to downtown is 30 minutes and ___________ travel time to downtown is 55 minutes [or less]. After each scenario was provided to a participant (with the blanks replaced with the appropriate terms), the participant was asked the following questions: • How much time would you allow for your drive to downtown? • How did you decide on that time? • What does the (average, estimated, expected, or typical) travel time mean to you about how long your drive will take? • For scenario 1: What does the (added, cushion, extra, or recommended cushion) time mean to you about how long your drive will take? • For scenario 2: What does the (95th percentile time, major- ity of the time, most of the time, or travel time for planning time) mean to you about how long your drive will take? • How likely are you to reach downtown in (70 for scenario 1, 55 for scenario 2) minutes or less? • For scenario 1: If the (added, cushion, extra, or recommended cushion) time had been 10 minutes, how much time would you have allowed for your drive? Why? (Note: The response implying the desired use of that term would be 50 min.) • For scenario 2: If the (95th percentile time, majority of the time, most of the time, or travel time for planning time) had been 45 minutes, how much time would you have allowed for your drive? Why? (Note: The response implying the desired use of that term would be 45 min.) At the end of each scenario, the participants were shown the remaining possible terms being tested to convey uncertainty and asked to indicate whether each one meant the same thing as the term they had originally seen in the website message. If not, participants were asked to explain why the term was differ- ent. Finally, participants were asked which term they preferred to convey travel time uncertainty information. After participants answered questions concerning both sce- narios, they were asked whether they preferred the travel time uncertainty information in the first scenario (buffer time) or the second scenario (total trip time estimate). results The following sections discuss the use of the buffer time terms and total trip time estimate terms by participants when determining the total trip time they would allow for each scenario provided. To evaluate effectiveness, researchers examined the distribution of participant-selected trip times relative to the total trip time intended by the message. For this evaluation, the greater the percentage of participants select- ing travel times within the 10-min to 15-min range around that total trip time value, the more effective the message was deemed. Trip times much below the intended total trip time would result in motorists being late more frequently than

74 • Participants could add some time other than the buffer time value to the normal trip time to come up with the trip time they would allow for (the amount of time added may or may not be influenced by the buffer term). The percentage of participants who selected various trip times in each of those categories is presented in Table 8.3. Overall, few participants indicated they would allow for only the normal trip time. Researchers could not determine whether that resulted from being provided a buffer time or that was simply the participants’ normal trip planning behavior. In addition, very few participants stated they would only allow for the buffer time provided in the website message, indicating that only a small percentage of participants were confused by the intent of the buffer term. Among the four buffer terms, recommended cushion time resulted in the greatest percentage of participants adding the buffer time to the normal trip time (28%), followed closely by added time (23%). When the term cushion time was used, only 8% of participants added the nor- mal trip time and buffer time to determine the amount of time they should allow for the trip. Still, the majority of participants added some other value to the normal trip time when choosing the trip time they would allow for in scenario 1. Additional insights into the trip times that participants indicated they would allow for (as a function of the trip time uncertainty term) are presented in Table 8.4. As expected, the percentages for the “35–44 min” and “25–34 min” groups are similar to those of the “used normal trip time only” and “used buffer term only” rows in Table 8.3, respectively. In addition, those in the “65–74 min” column correlate with the “added normal trip time and buffer time values together” column in Table 8.3. The distribution of trip times that would be allowed desired; and trip times much higher would result in exces- sively early arrival times for the majority of trips made. Use of Buffer Time in Total Trip Time Estimation Researchers analyzed the responses to the questions regard- ing how much time the participants would allow for their trip. They initially looked at each pair of normal trip time and buffer terms to determine whether the normal trip time term affected the use of the buffer term. Researchers did not iden- tify any clear trends that indicated the normal trip time term affected the use of the buffer term, so the data for each buffer term were aggregated across the normal trip time terms for further analysis. Researchers explored how the type of buffer term used may influence the extent to which motorists do or do not incorporate the amount of buffer time presented in a mes- sage into their travel plans. At the simplest level of analysis, researchers envisioned four possible ways in which motor- ists would use the times presented in scenario 1 (the normal trip time and the buffer value) to estimate the time they would allow for their trip: • Participants could ignore the buffer value and base their trip time strictly on the normal trip time value provided. • Participants could ignore the normal trip time value and base their trip time strictly on the buffer time value provided (implying confusion as to the intent of the buffer term). • Participants could combine the normal trip time and the buffer time values to come up with the trip time they would allow for (the intended use of the buffer time value). Table 8.3. Participant Use of Buffer Term When Planning a Trip Participant Use of Buffer Term Percentage of Participants Buffer Term Viewed Average Added (n 5 47) Cushion (n 5 48) Extra (n 5 47) Recommended Cushion (n 5 47) Used normal trip time only 4 17 4 6 8 Used buffer term only 2 4 2 6 4 Added normal trip time and buffer time values together 23 8 15 28 19 Added normal trip time and some other value together 66 67 66 55 63 Total 95a 96b 87c 95d 94e a Does not add up to 100% because researchers categorized 5% of responses as “other.” b Does not add up to 100% because researchers categorized 4% of responses as “other.” c Does not add up to 100% because researchers categorized 13% of responses as “other.” d Does not add up to 100% because researchers categorized 5% of responses as “other.” e Does not add up to 100% because researchers categorized 6% of responses as “other.”

75 the term best able to encourage selection of the trip time which most closely aligns with the amount of trip time uncer- tainty that the operating agency would be attempting to con- vey. When the terms added time and extra time were shown, a greater percentage of participants chose travel times that exceeded the 70-min intended value. That result suggests a potentially lower level of confidence implied by those two terms and led participant to allow more buffer time than was shows some small differences by type of trip time uncertainty term used. Stated another way, Figure 8.1 shows the cumulative distri- bution of trip times that participants would allow—as a func- tion of the buffer time used. Among the participants who saw the term recommended cushion, 70% chose a trip time that was within 15 min of the total trip time provided in the web- site message. Thus, recommended cushion time is apparently Table 8.4. Distribution of Trip Times Allowed for 70-Minute Trip Trip Times Allowed for 70-min Trip Percentage of Participants Buffer Term Viewed Average Added (n 5 47) Cushion (n 5 48) Extra (n 5 47) Recommended Cushion (n 5 47) <25 min 2 4 2 25–34 min 4 6 6 8 6 35–44 min 6 17 6 8 9 45–54 min 6 32 11 11 15 55–64 min 28 27 24 30 27 65–74 min 24 8 19 28 20 75–84 min 11 8 15 13 12 ≥85 min 21 15 2 9 Total 100 100 100 100 100 Note: Normal trip time = 40 min; buffer time provided = 30 min. Figure 8.1. Cumulative distribution of trip times allowed for 70-minute trip.

76 buffer terms were added time (74%), extra time (70%), and cushion time (69%). Next, participants were asked how much time they would have allowed if the normal trip time remained at 40 min, but the trip time uncertainty value was only 10 min. The responses to that question are summarized in Table 8.6 and shown in Figure 8.2. The responses made clear that drivers are not likely to have a high degree of confidence in small buffer values, regardless of the term used. Although recommended cushion time again led the greatest percentage of participants to select the sum of the normal and uncertainty trip time values (i.e., 50 minutes), a large portion of participants chose trip times longer than that amount. Furthermore, responses did not vary significantly as a function of trip time uncertainty term used. presented in the message. Participants also had less confi- dence in the term cushion time, but instead of allowing more time than presented, they chose a value less than the sum of the normal trip time and buffer time. That result suggests the term cushion time may convey a sense of excessive uncertainty in the trip time, greater than desired. The same trends regarding the confidence participants expressed in use of the trip time uncertainty terms are evi- dent in Table 8.5. It shows participant responses to the ques- tion “How likely are you to arrive at your destination in 70 minutes or less?” Overall, 85% of participants viewing recommended cushion time responded “absolutely certain (100% chance)” or “very likely (95% chance).” The percent- age of participants selecting those two categories for the other Table 8.5. Participant Confidence in Arriving at Destination in 70 Minutes or Less Participant Confidence Percentage of Participants Buffer Term Viewed Average Added (n 5 47) Cushion (n 5 48) Extra (n 5 47) Recommended Cushion (n 5 47) Absolutely certain (100% chance) 21 17 15 23 19 Very likely (95% chance) 53 52 55 62 56 Probably (75% chance) 19 21 26 11 19 Unsure (50/50 chance) 7 8 4 2 5 Doubtful (less than 50/50 chance) 0 2 0 2 1 Total 100 100 100 100 100 Note: Normal trip time = 40 min; buffer time provided = 30 min. Table 8.6. Distribution of Trip Times Allowed for 50-Minute Trip Trip Times Allowed for 50-Min Trip Percentage of Participants Buffer Term Viewed Average Added (n 5 47) Cushion (n 5 48) Extra (n 5 47) Recommended Cushion (n 5 47) <25 min 2 6 9 9 6 25–34 min 2 4 0 2 2 35–44 min 0 10 2 6 5 45–54 min 38 35 38 45 39 55–64 min 45 33 43 30 38 65–74 min 9 6 4 4 6 75–84 min 2 2 2 4 3 Total 98a 96b 98c 100 99d Note: Normal trip time = 40 min; buffer time provided = 10 min. a Does not add up to 100% because 2% of participants did not provide an answer. b Does not add up to 100% because 4% of participants did not provide an answer. c Does not add up to 100% because 2% of participants did not provide an answer. d Does not add up to 100% because 1% of participants did not provide an answer.

77 message meant the same thing as the remaining buffer terms (those not originally viewed by the participant). The percent- age of participants who indicated that various terms implied the same meaning is provided in Table 8.7. Overall, two clear trends emerged. Participants who saw either added time or extra time rated the other term as having a similar meaning. Likewise, those who saw either cushion time or recommended cushion time viewed the other term as having the same mean- ing. As shown in Figure 8.2, added time and extra time resulted in similar trends, so it is not surprising that these two terms These results imply that drivers may have difficulty accepting and trusting a route-specific buffer value that is smaller than the generic buffer they have developed through years of driving experiences and other influences. Similar Meanings and Preference of Buffer Terms Another portion of the survey addressed whether participants thought the buffer term provided in the scenario 1 website Figure 8.2. Cumulative distribution of trip times allowed for 50-minute trip (normal trip time 5 40 min, buffer 5 10 min). Table 8.7. Assessment of Similar Meanings of Buffer Terms Terms Compared Percentage of Participants Buffer Term Viewed Average Added (n 5 47) Cushion (n 5 48) Extra (n 5 47) Recommended Cushion (n 5 47) Added means the same as Cushion 43 50 46 Added means the same as Extra 81 66 73 Added means the same as Recommended Cushion 43 43 43 Cushion means the same as Extra 50 40 45 Cushion means the same as Recommended Cushion 73 72 73 Extra means the same as Recommended Cushion 36 49 43 Note: Shaded cells indicate combinations that were not evaluated.

78 • Participants could ignore the normal trip time value pro- vided and base their trip time strictly on the total trip time estimate provided (the intended use). • Participants could combine the normal trip time and the total trip time estimate to come up with the trip time they would allow for (the intended use of the buffer time term). • Participants could add some other time (their own buffer) to the normal trip time or the total trip time estimate to come up with the trip time they would allow for (the amount of other time added may or may not be influenced by the total trip time estimate). The percentage of participants who selected trip times in each of those categories is presented in Table 8.9. Few partici- pants indicated they would allow for only the normal trip time provided in the website message. In addition, few par- ticipants combined the values for the two terms provided. Among the four total trip time estimate terms, majority of the time resulted in the greatest percentage of participants basing their trip time strictly on the total trip time estimate pro- vided. Still, on average, 37% of participants added some other value to the total trip time estimate to come up with the trip time they would allow for in scenario 2. Additional insights into the trip times that participants indicated they would allow for (as a function of the trip time uncertainty term) are presented in Table 8.10. The majority of the trip time values were in the “45–54 min” and “55–64 min” groups, which were within 10 min of the total trip time estimate value (55 min). Those times reflect the responses of participants who used the total trip time estimate, the total trip time estimate plus some other small time, or the normal trip time plus some other larger time. The distribution of trip times that participants would allow for shows some small differences by type of total trip time estimate term used. The distribution is shown in Figure 8.3. were considered to have similar meanings. However, the trip times allowed by participants for cushion time and recommended cushion time (shown in Table 8.7) were not similar. The percentage of participants who preferred each of the terms tested is shown in Table 8.8. Overall, the results did not indicate a clear preference for any of the terms: no single term was preferred by more than one-third of the participants, on average. However, the preference data did reflect that pairs of terms were viewed as having similar meanings: cushion time and recommended cushion time were preferred over added time or extra time. Use of Total Trip Time Estimate Again, researchers analyzed the responses to the questions regarding how much time participants would allow for their trip. They initially looked at each normal trip time or total trip time estimate pair to determine whether the normal trip time terms affected the use of the total trip time estimate term. Researchers did not identify any clear trends that indicated the normal trip time term affected the use of the total trip time estimate term, so the data for each total trip time estimate term were aggregated across the normal trip time terms for further analysis. The next analysis explored how the total trip time estimate terms influence the extent to which motorists may or may not use the total trip time provided to plan their travel. At the sim- plest level of analysis, researchers envisioned four possible ways in which drivers would use the times presented in scenario 2 (the normal trip time and the total trip time values) to estimate how much time they would allow for their trip: • Participants could ignore the total trip time estimate and base their trip time strictly on the normal trip time value provided. Table 8.8. Preferences Among Buffer Terms Preferred Term Percentage of Participants Buffer Term Viewed Average Added (n 5 47) Cushion (n 5 48) Extra (n 5 47) Recommended Cushion (n 5 47) Added 17 17 15 15 16 Extra 15 10 13 32 17 Cushion 34 25 45 30 33 Recommended cushion 32 48 26 23 32 Added or extra 2 0 0 0 1 None 0 0 3 0 1 Total 100 100 100 100 100

79 many participants who viewed the term travel time for plan- ning chose to use a value less than the total trip time estimate in deciding how much time to allow. This result suggests that the term travel time for planning may convey more of a sense of a maximum trip time. Table 8.11 shows participants’ confidence in how likely they would be to arrive at the destination in 55 min or less for the trip time uncertainty terms used in scenario 2. Strangely, the term most of the time resulted in the largest percentage of participants (75%) selecting “absolutely certain (100% chance)” or “very According to these data, the term majority of the time resulted in the largest percentage of participants (79%) choosing times within 10 min of the total trip time estimate (55 min) conveyed in the website message for scenario 2. When the terms most of the time and 95th percentile were shown, a greater percentage of participants chose travel times that exceeded the total trip time estimate presented in the message (23% and 35%, respec- tively). This results suggests participants may have less confi- dence in these two terms and that led to decisions to allow more time than was presented in the message. Conversely, Table 8.9. Use of Total Trip Time Estimate for Trip Planning Participant Use of Total Trip Time Estimate Percentage of Participants Total Trip Time Estimate Term Viewed Average 95th Percentile (n 5 46) Majority of the Time (n 5 48) Most of the Time (n 5 48) Travel Time for Planning (n 5 47) Used normal trip time only 4 4 0 4 3 Used total trip time estimate only 24 37 25 31 29 Added normal trip time and total trip time estimate values together 7 0 0 2 2 Added normal trip time and some other value together 13 15 29 40 24 Added total trip time estimate and some other value together 50 40 38 19 37 Total 98a 96b 92c 96d 95e a Does not add up to 100% because researchers categorized 2% of the responses as “other.” b Does not add up to 100% because researchers categorized 4% of the responses as “other.” c Does not add up to 100% because researchers categorized 8% of the responses as “other.” d Does not add up to 100% because researchers categorized 4% of the responses as “other.” e Does not add up to 100% because researchers categorized 5% of the responses as “other.” Table 8.10. Distribution of Trip Times Allowed for 55-Minute Trip Trip Times Allowed For Percentage of Participants Total Trip Time Estimate Term Viewed Average 95th Percentile (n 5 46) Majority of the Time (n 5 48) Most of the Time (n 5 48) Travel Time for Planning (n 5 47) <25 min 2 0 0 0 1 25–34 min 4 4 4 8 5 35–44 min 7 2 4 11 6 45–54 min 7 11 25 23 16 55–64 min 45 68 44 50 52 65–74 min 11 2 7 4 6 75–84 min 11 9 8 2 7 ≥85 min 13 4 8 2 7 Total 100 100 100 100 100 Note: Normal trip time = 30 min; total trip time estimate provided = 55 min.

80 remained at 30 min but the total trip time estimate was 45 min (10 min less than in the original scenario 2). The responses to that question are summarized in Table 8.12 and shown in Figure 8.4. The majority of the trip time values were in the “45–54 min” and “35–44 min” groups, which was again within 10 min of the total trip time estimate value (45 min). This result shows that most of the participants did shift the trip time they would allow for to match the total trip time estimate presented. However, for all of the trip time uncertainty terms used, a larger percentage of participants chose a trip time longer than the total trip time estimate, showing less confidence in the smaller total likely (95% chance).” This result was unexpected because 23% of participants chose a travel time that exceeded the total trip time. For travel time for planning, the percentage of par- ticipants selecting those two categories was 73%. That result was expected because 92% of participants chose a travel time that was less than the total trip time estimate. For both major- ity of the time and 95th percentile, the percentage of partici- pants selecting the same two categories corresponded to only 62% and 54%, respectively. The final question for this scenario asked participants how much trip time they would allow if the normal trip time Figure 8.3. Cumulative distribution of trip times allowed for 55-minute trip (normal trip time 5 30 min; total trip time estimate provided 5 55 min). Table 8.11. Participant Confidence in Arriving at Destination in 55 Minutes or Less Participant Confidence Percentage of Participants Total Trip Time Estimate Term Viewed Average 95th Percentile (n 5 46) Majority of the Time (n 5 48) Most of the Time (n 5 48) Travel Time for Planning (n 5 47) Absolutely certain (100% chance) 2 9 21 13 11 Very likely (95% chance) 54 53 54 60 56 Probably (75% chance) 20 27 17 15 20 Unsure (50/50 chance) 17 9 8 10 11 Doubtful (less than 50/50 chance) 7 2 0 2 2 Total 100 100 100 100 100 Note: Normal trip time = 30 min; total trip time estimate provided = 55 min.

81 Table 8.12. Distribution of Trip Times Allowed for 45-Minute Trip Trip Times Allowed For Percentage of Participants Total Trip Time Estimate Term Viewed Average 95th Percentile (n 5 46) Majority of the Time (n 5 48) Most of the Time (n 5 48) Travel Time for Planning (n 5 47) <25 min 4 0 2 2 2 25–34 min 4 4 2 4 4 35–44 min 0 11 15 17 11 45–54 min 48 49 44 50 48 55–64 min 26 28 27 25 26 65–74 min 3 2 2 0 1 75–84 min 13 4 4 2 6 ≥85 min 2 2 4 0 2 Total 100 100 100 100 100 Note: Normal trip time = 30 min; total trip time estimate provided = 45 min. 0% 25% 50% 75% 100% < 25 25 -3 4 35 -4 4 45 -5 4 55 -6 4 65 -7 4 75 -8 4 ≥8 5 Trip Time Allowed for (minutes) 95th Percentile Majority of the Time Most of the Time Travel Time for Planning Total Figure 8.4. Cumulative distribution of trip times allowed for 45-minute trip (normal trip time 5 30 min; total trip time estimate provided 5 45 min).

82 The percentage of participants who preferred each of the terms tested is shown in Table 8.14. These findings revealed no clear consensus as to the most preferred total trip time estimate term. However, the preference data did reflect that majority of the time and most of the time were similarly preferred. Preference for Travel Time Uncertainty Terms Participants were also asked whether they preferred the travel time uncertainty information in the first scenario (buffer time) or the second scenario (total trip time estimate). Overall, 60% of participants preferred to be provided the normal travel time and the total trip time estimate (scenario 2). Of these participants, about half preferred the total trip estimate term because the total number was provided (i.e., they did not have to add multiple values). Another 20% thought the total value would be more accurate or exact than the two pieces of information that had to be added together. trip time estimate value. Here again, researchers inferred that drivers may have difficulty accepting and trusting a total trip time estimate that is closer to the normal trip time. Meanings of and Preference for Total Trip Time Estimate Terms Next, researchers asked whether participants thought the total trip time estimate term provided in the scenario 2 web- site message meant the same thing as the remaining total trip time estimate terms. The percentage of participants who indicated that various terms implied the same meaning is pro- vided in Table 8.13. Overall, one clear trend emerged: partici- pants who saw either majority of the time or most of the time rated the other term as having a similar meaning. This result was not surprising considering the similar trends for these two terms shown in Figure 8.1. No other term combinations were viewed as being similar by most participants. Table 8.13. Assessment of Similar Meanings of Total Trip Time Estimate Terms Terms Compared Percentage of Participants Total Trip Time Estimate Term Viewed Average 95th Percentile (n 5 46) Majority of the Time (n 5 48) Most of the Time (n 5 48) Travel Time for Planning (n 5 47) 95th percentile means the same as majority of the time 68 45 56 95th percentile means the same as most of the time 49 54 52 95th percentile means the same as travel time for planning 13 32 22 Majority of the time means the same as most of the time 91 79 85 Majority of the time means the same as travel time for planning 26 47 36 Most of the time means the same as travel time for planning 29 64 46 Note: Shaded cells indicate combinations that were not evaluated. Table 8.14. Preference for Total Trip Time Estimate Terms Preferred Term Percentage of Participants Total Trip Time Estimate Term Viewed Average 95th Percentile (n 5 46) Majority of the Time (n 5 48) Most of the Time (n 5 48) Travel Time for Planning (n 5 47) 95th percentile 21 15 10 13 15 Majority of the time 34 30 63 30 39 Most of the time 26 38 17 47 32 Travel time for planning 15 17 10 8 12 95th percentile or majority of the time 2 0 0 0 1 None 2 0 0 2 1 Total 100 100 100 100 100

83 of participants adding the buffer time to the normal trip time, while the term cushion time resulted in the least. • Participants who saw recommended cushion time were more likely to select a total trip time that was clustered closely around the 70-min total trip time implied by the website message. This term also produced the highest confidence among participants that they would arrive at their destina- tion in 70 min or less. Even though recommended cushion time and cushion time were viewed as having a similar mean- ing, participants who saw cushion time were more likely to choose a trip time less than 70 min, implying that partici- pants had less trust in this term. • Participants viewed the terms added time and extra time as having a similar meaning. In addition, those who saw added time or extra time were more likely to choose a trip time that exceeded 70 min. This result suggests that they had some- what less confidence in these terms and so added even more time to ensure on-time arrival; or it may suggest that they did not fully understand the intent of these time values. • Results did not indicate a clear preference for any of the buff er terms. The terms recommended cushion time and cushion time were selected most frequently, but only by about one-third of the participants for each. Similar Meanings of and Preference for Normal Trip Time Terms Researchers also asked whether participants thought the nor- mal trip time term provided in the scenario 2 website mes- sage meant the same thing as the remaining normal trip time terms. As shown in Table 8.15, on average, 76% of partici- pants viewed the terms estimated and expected as having a similar meaning. Likewise, 70% of participants viewed aver- age as similar to typical. No other clear trends emerged. The results of the participants’ preferences for the normal travel time terms are summarized in Table 8.16. These find- ings revealed no clear consensus as to the most preferred nor- mal travel time term. Conclusions Buffer Time Terms • Approximately 20% of participants added the normal travel time and buffer time to determine the amount of time they should allow for the trip. The particular buffer time term used may have had a small influence on this value. The term recommended cushion time resulted in the greatest percentage Table 8.15. Assessment of Similar Meanings of Normal Travel Time Terms Terms Compared Percentage of Participants Normal Travel Time Term Viewed Average Average (n 5 48) Estimated (n 5 47) Expected (n 5 48) Typical (n 5 46) Average means the same as estimated 63 66 64 Average means the same as expected 42 48 45 Average means the same as typical 71 70 70 Estimated means the same as expected 68 83 76 Estimated means the same as typical 40 37 39 Expected means the same as typical 52 43 49 Note: Shaded cells indicate combinations that were not evaluated. Table 8.16. Preference for Normal Travel Time Terms Preferred Term Percentage of Participants Normal Travel Time Term Viewed Average Average (n 5 48) Estimated (n 5 47) Expected (n 5 48) Typical (n 5 46) Average 13 30 13 24 20 Estimated 58 21 52 33 41 Expected 16 40 25 39 30 Typical 13 9 10 4 9 Total 100 100 100 100 100

84 destination. Overall, 75% and 67% of participants, respec- tively, indicated that their trip would certainly or very likely take less than the value shown, further indicating that they recognized that value as an upper limit to how long a trip might take. • A slight preference was detected among participants for the use of total trip time terms over the use of buffer terms. This result most likely occurred because many drivers would prefer not to have to do the math (adding a normal trip time and a buffer time) to estimate the time they should allow for a trip. Comprehension of Terms for Normal Travel Time • None of the four terms evaluated to convey normal travel time appeared to affect participants’ selection and use of the buffer time or total trip time estimates. • Participants viewed the terms estimated and expected as having similar meanings. In addition, participants judged the terms average and typical to have similar meanings. No other clear trends emerged as to whether participants believed that the various normal travel time terms had similar meanings. • No clear trends emerged with regard to participant prefer- ence for the normal travel time terms. The terms estimated and expected garnered the most support but only by about one-third or so of the participants for each. recommendations • Because the four normal travel time terms evaluated per- formed similarly and no clear preference was identified, researchers believe that any of the four terms may be used to convey the normal travel time. • Both types of travel time uncertainty terms (buffer time and total trip time estimate) similarly influenced the extent to which motorists might or might not use the travel time uncertainty in their travel time decisions. However, partici- pants did prefer to be provided with the total trip time esti- mate. Researchers think that providing the total trip time estimate decreases driver workload and the potential for math errors, which could otherwise lead users to distrust the information provided by an agency. Thus, a total trip time estimate term appears to be preferable to a buffer term when trying to convey trip time uncertainty information. Further- more, researchers recommend the use of the term majority of the time to describe the total trip time estimate. • The above notwithstanding, if an agency still desires to use a buffer time to convey travel time uncertainty, researchers recommend the use of the term recommended cushion time. • When the buffer time was reduced from 30 min to 10 min, participants did not reduce their total trip time estimate by that same amount. This result suggests that participants did not totally trust the buffer time values, regardless of the term used. Trip time providers may have difficulty getting drivers to accept a trip uncertainty value that is less than the general uncertainty values they have learned through their own experiences when making trip time decisions. Total Trip Time Estimate Terms • Results of the survey indicated that approximately 30% of participants used a total trip time measure for determining the amount of time they should allow for a particular trip. The particular total trip time term used may have had a small influence on this value. The term majority of the time resulted in the greatest percentage of participants who strictly used the total trip time measure, while the term 95th percentile resulted in the least. • The term majority of the time resulted in a greater percentage of participants who would select the time provided, or a value near it, when deciding how much time to allow for a particu- lar trip. Even though participants judged the terms majority of the time and most of the time to have similar meanings, par- ticipants who saw most of the time were more likely to choose a trip time that exceeded 55 min, implying a lower level of confidence for this term. Participants who saw the term 95th percentile were also more likely to choose a travel time that exceeded 55 min. Conversely, participants who viewed the term travel time for planning were more likely to choose a value less than 55 min, suggesting that this term may convey more of a sense of a maximum trip time; if drivers relied on this term, they might be late more often than they had hoped. • Participants showed no clear preference among the total trip time terms. The terms majority of the time and most of the time were selected most frequently but only by about one-third of the participants for each. • Changing the value of the total trip time estimate (relative to the normal trip time estimate) did affect the time par- ticipants chose to allow for a trip. However, as with the buffer time terms, participants’ selection of a time to allow for a trip appeared to depend on how the value of the total trip time estimate compared with their own personal buffer time that had developed over time on the basis of their own experiences. Terms that implied a buffer of less than 20 min appeared to be less trusted by drivers. Comparison of Travel Time Uncertainty Information • Use of a buffer time or a total trip time estimate resulted in similar levels of confidence in arriving on time at a