Understanding How Individuals Make Travel and Location Decisions: Implications for Public Transportation (2008)

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.

123 Overview of the Chapter This chapter explores the components of TPB for what the model can tell about what affects individuals’ intentions to increase the use of environmentally friendly modes, such as walking and transit. The Internet panel survey in Phase 2 of the project included two sets of questions for the TPB, and these provide the data to allow examination of the relation- ship between intent and ATT, SN, and SCF about mode change. Further, the two sets of questions provide data to allow examination of beliefs that affect ATT, SN, and SCF. The two sets of questions are referred to as the initial TPB set and the final TPB set. In the initial set, respondents were asked to give their opinions about making more trips by walking and public transportation and reducing trips by pri- vate automobile. In the final set, they were asked to give opin- ions about how a series of transportation options might allow them to increase their use of alternatives to the private auto- mobile. The transportation options included good transit service to downtown, good transit service to the rest of the region, a shuttle bus to the local center, a shared ride service that is less expensive than a taxi, car sharing, a smart card that could pay for all services, and a smart phone that provided real-time information on schedules and a 911 emergency communication capability. In between these two sets of questions, the respondents were exposed to some messages that communicated the value of public transportation. Around one-third of the re- spondents received a message on saving money, one-third received a message about reducing pollution and improving public health, and one-third received no message. The ob- jective in this set of exercises was to test whether intent would change given the messages and service options, and also to see if variables associated with this change could be isolated. This chapter summarizes statistical analyses of the hypoth- esized relationships for the TPB. Regression analysis is used to examine the relationship between the intent and direct measures of ATT, SN, and SCF. It was examined whether re- spondents’ answers regarding their final intent are related to their initial intent or whether they are, as the TPB implies, re- lated to their final ATT, SN, and SCF. Following the analyses of intent, the relationship between behavioral beliefs and ATT was examined. Respondents pro- vided direct measures of the desirability or importance of each behavioral belief as variables called outcome evaluations. A second measure of the importance of the behavioral beliefs was obtained by regressing final behavioral beliefs on final at- titude. The coefficients from the regression are a statistically derived set of importance weights for the behavioral beliefs. Changes in attitude may be explained by changes in the be- havioral beliefs that are important contributors to attitude. Following the analysis for ATT, similar analyses are shown for the normative beliefs and SCF. The results of these analyses indicate the types of changes that may have the most poten- tial for improving the use of transit and walking. Can Respondents’ Ratings Be Trusted? One problem with the study design is that the respondents may have anticipated what the researchers would like to find and may have answered in a way to please the researchers. Be- cause respondents were asked to consider how they might use transit and walk more and drive less, they may well have in- dicated more interest in transit and walking than they actu- ally would have felt. There are several ways to try to overcome this problem. The first is to remind respondents of the disadvantages, as well as the advantages, of using transit and walking. As re- spondents consider the negative aspects of changing modes, they may be less inclined to exaggerate their attitude and in- tent to walk and use transit more. Another way to combat the tendency to exaggerate a pos- itive response is to examine the change when respondents C H A P T E R 1 2 Use of the TPB to Understand Mode Choice

were exposed to different messages and different alternatives. Since it is likely that a tendency to exaggerate will show up from the beginning, when we examine a change in intent or other variables, much of the effect of exaggeration should be eliminated, providing that it is similar in each case. Relationships of the Directly Measured TPB Variables Table 12-1 summarizes the change in ratings for the directly measured TPB variables with regard to increasing use of alternative modes and decreasing use of the private auto- mobile. Intent is measured by the respondents’ agreement on three statements regarding (a) their plans to walk and take public transportation more, (b) their intent to walk and take public transit more, and (c) whether they will make an effort to walk and use public transportation more. ATT was the degree to which the respondent thought that walking and using public transportation more would be (a) more pleas- ant, (b) more interesting, and (c) desirable. SN was what was “expected of me.” SCF was measured by (a) the respondents’ confidence in being able to walk and take public transporta- tion more, (b) their ease of doing so, and (c) the extent to which they thought it was possible. Intent increased significantly from the initial to the final set of TPB statements. The increase was on the order of 0.8 units on a rating scale between one and seven. The ATT did not in- crease significantly. The lack of change in attitude toward the behavior is one indication that the respondents were not simply answering all questions in a manner to please the researchers. Referring again to Table 12-1, there were significant changes in SN between the two TPB exercises. The rating of “it is expected of me” increased by nearly a point on the seven-point scale. There were also changes in the SCF of the respondents between the two TPB exercises. Respondents indicated significantly greater confidence that they could walk and take transit more with the alternatives than without them. So while there was not a significant change in the respon- dents’ own attitudes toward increasing their use of walking and taking transit, they did think that with the seven alterna- tive transportation options available, walking and taking transit more would be more acceptable to others. In other words, there would be a normative expectation that the respondent would walk and use transit more. The results also indicated respondents were more confident in being able to walk and take public transportation more, that it would be easier to do so, and that it was more likely possible. Regression was used to examine the relationship between the direct measure of intent (to walk and use public trans- portation more and reduce trips by automobile) and direct measures of ATT, SN, and SCF. Regressions were run sepa- rately for both of the sets of questions about intent—one set of questions before the messages and another set after the mes- sages and with respondents told to assume they had access to the seven alternative transportation options. Table 12-2 and Table 12-3 show these regressions. To be consistent between the initial set of TPB questions and the final set, one variable, “It is expected of me,” was used to measure SN. Table 12-2 shows the results for the initial set of questions. Table 12-3 shows the results for the final set. All of the inde- 124 TPB Measure Differences Intent 0.80* Attitude (average of three measures). 0.16 Subjective Norm (one measure: “it is expected of me”) 0.88* Substitute Measure of Subjective Norm (average of four normative beliefs) 1.67* Self-Confidence (average of three measures) 0.53* *Difference is significant at p < .05 Table 12-1. Differences in direct measures for the theory of planned behavior. Dependent Variable: Initial Intent Independent Variable Coefficient t-Statistic Probability Constant -0.37 -3.6 .0004 Initial Attitude 0.47* 16.3 .0001 Initial Subjective Norm 0.28* 11.0 .0001 Initial Self-Confidence 0.33* 12.1 .0001 * indicates a significant coefficient at p < .05 R2 = 80%, 501 observations. Table 12-2. Regression of initial intent to walk and take transit more.

125 pendent variables are highly significant. As can be seen in each of the tables, attitude has the largest coefficient. The coefficients for subjective norm and self-confidence are sim- ilar in magnitude to each other and less than that for attitude. One question regarding this set of equations is whether the respondents rated their intentions the same as they had previously, or whether they were consciously thinking about the use of the transportation alternatives. To look at this question, the final regression in Table 12-3 was redone to include the directly measured intent from the prior TPB exercise as another independent variable. If there had been no change in the way respondents answered the rating ques- tions, then it would be expected to see a larger and highly significant coefficient for the initial intent measure. Also, there is confidence that final intent is corresponding with the final ATT, SN, and SCF, following the theory of planned behavior. Table 12-4 shows the result of that regression. The initial intent measure is not significant at p < .05, and the coeffi- cients of the other variables remain significant and similar to the prior regression. Thus there is some confidence that the respondents are evaluating their intent to use walking and public transportation differently with and without the alter- native transportation services. Also, there is confidence that final intent is corresponding with the final ATT, SN, and SCF following the theory of planned behavior. Relationship Between Behavioral Beliefs and Attitude Since attitude is the most critical driver of intent, it is im- portant to understand what factors drive ATT. There also is in- terest in determining if those factors change when respondents are asked to consider the effect of additional services. Recall that the formal TPB model says that ATT is influ- enced by a linear combination of behavioral beliefs weighted by outcome evaluations. Table 12-5 shows the set of outcome evaluations that pair with the final behavioral beliefs. The outcome evaluations are shown in order of their average ratings for the sample of 501 respondents. According to Table 12-5, having reliable transportation is rated the highest of the outcome evaluations, and the variation among respondents is less for this than for all others. Saving money and improving health by walking more follow in order. The least desirable outcomes were to have fewer cars in the household and to be dependent on someone else for travel. In addition to the self-stated outcome evaluations, regres- sion can be used to measure how respondents weight their behavioral beliefs in forming their attitude for increasing their use of walking and public transportation. The regression coefficients provide a statistical measure of weights similar to the outcome evaluations. Table 12-6 shows a regression of the six final behavioral beliefs on final attitude (see Table 10-16. for a description Dependent Variable: Final Intent Independent Variable Coefficient t-Statistic Probability Constant 0.11 1.23 .2203 Final Attitude 0.55* 17.6 .0001 Final Subjective Norm 0.25* 11.3 .0001 Final Self-Confidence 0.23* 8.09 .0001 * indicates a significant coefficient at p < .05 R2 = 85%, 501 observations. Table 12-3. Regression for final intent to walk and take transit more. Dependent Variable: Final Intent Independent Variable Coefficient t-Statistic Probability Constant 0.09 1.04 0.2985 Final Attitude 0.54* 16.3 0.0001 Final Subjective Norm 0.25* 11.1 0.0001 Final Self-Confidence 0.21* 7.4 0.0001 Initial Intent (without services and message) 0.04 1.53 0.1256 * indicates a significant coefficient at p < .05 R2 = 85%, 501 observations. Table 12-4. Alternative regression for final intent to walk and take transit more.

126 Statement, Rated on a Seven-Point Scale Mean (SD) For me to have a reliable type of transportation to take to my destination would be: (extremely unimportant to extremely important) 6.5 (1.0) For me to reduce the cost of my daily transportation would be: (extremely undesirable to extremely desirable) 5.9 (1.4) For me to improve my health by walking more would be: (extremely unimportant to extremely important) 5.8 (1.3) For me to reduce pollution by using my car less would be: (extremely unimportant to extremely important) 5.3 (1.7) For me to reduce the time I spend driving would be: (extremely unimportant to extremely important) 5.3 (1.7) For my household to own fewer cars would be: (extremely undesirable to extremely desirable) 3.1 (1.9) For me to be dependent on someone else to get me to my destination on time would be: (extremely undesirable to extremely desirable) 2.8 (1.8) Dependent Variable: Final Attitude Independent Variable: Final Behavioral Beliefs Coefficient t-Statistic Probability Constant 1.02* 3.06 .0023 I would rely on alternative transportation and walking 0.29* 4.88 .0001 I would improve health and reduce pollution 0.27* 3.86 .0001 I’d save money 0.13* 2.68 .0077 I would reduce the amount of time I spend driving 0.07 1.39 .1642 My household could get by with fewer cars 0.05 1.51 .1323 I would be dependent on someone else -0.17* -3.58 .0004 *indicates a significant coefficient at p < .05 R2 = 32.2% 460 observations. Table 12-5. Outcome evaluations that pair with final behavioral beliefs. Table 12-6. Regression for final attitude with seven alternative services available. of the wording of the beliefs). Because the measures about health and reducing pollution were highly correlated, an average value was substituted for the three individual measures. Although not directly comparable with Table 12-5, the order of the coefficients is similar. The largest coefficient is for “I would rely on alternative transportation,” followed closely by the coefficient for improving health and reducing pollu- tion. “I would save money” is also significant and positive, although its magnitude is around half of the top two. The only other significant coefficient is for “I would be depend- ent,” which is negative. The results shown in Table 12-6 may indicate why respon- dents’ ratings for attitude did not change significantly from the initial to the final TPB exercise. Recall that respondents rated their ability to “rely on alternative transportation” higher in the final exercise, and they also rated “I would save money” higher. However, they rated their ability to “improve health and reduce pollution” lower. Although it is not clear why respondents rated the “improve health and reduce pol- lution” lower when the seven transportation alternatives were available, the respondents’ insignificant change in attitude is consistent with the lowered rating for the behavioral belief that they would “improve health and reduce pollution.”

Relationship Between Normative Beliefs and Subjective Norm In the TPB, SN is influenced by the opinions of other people (the normative beliefs) and the extent to which the re- spondent cares about the opinions of others (the motivation to comply). Depending upon the characteristics of the re- spondent and the type of issue being examined, SN can vary in influence. As shown in Table 12-2 and Table 12-3, SN has less influence on intent (to walk and use public transporta- tion more) than ATT. Subjective norm and SCF are similar in influence. Table 10-5 showed the values of the motivation to comply with the desires of family, friends, neighbors, and coworkers. These are summarized again in Table 12-7. As can be seen, family has the most influence, followed by friends, coworkers, and then neighbors, according to the self-stated measures. Another method for looking at the influence of others on the respondents’ SN is to use regression to examine the rela- tionship between the normative beliefs and the subjective norm. The regression coefficients provide statistical measures of the impact of each normative belief on the subjective norm. Table 12-8 shows such a regression for the final subjective norm after the messages and description of alternative trans- portation options. The importance of family’s expectations is shown again in this table. Only family opinion appears to be significantly related to the final SN. The results in Table 12-8 may indicate why respondents’ ratings for SN increased significantly from the initial TPB to the final TPB exercise. Referring back to Table 10-21, there was a significant gain in the normative belief regarding fam- ily opinion, and this appears to have a significant influence on the SN. Relationship Between the Power of Control and Self-Confidence Self-confidence is the third component influencing intent in the TPB. Its influence, as shown in Table 12-2 and Table 12-3, is similar in magnitude to the subjective norm. As shown in Table 12-1, there were significant increases in the average value of self-confidence after respondents were shown pro- transit messages and the seven alternative transportation options. Because self-confidence does have a significant effect on intent, and because our survey panel survey results indicate self-confidence can be improved, it is worthwhile to try to understand the factors that affect self-confidence. 127 Dependent Variable: Final Subjective Norm Independent Variable: Final Normative Beliefs (with the new services available, Coefficient t-Statistic Probability Constant 0.97* 5.1 .0001 my family would be more supportive of my walking more and taking public transportation more. 0.54* 5.1 .0001 my friends would be more supportive of my walking more and taking public transportation more. 0.04 0.3 .7334 my neighbors would be more supportive of my walking more and taking public transportation more. -0.02 -0.2 .8393 my co-workers would be more supportive of my walking more and taking public transportation more. 0.09 1.0 .3252 * indicates significant coefficients at p < .05 R2 = 34%, 501 observations. Table 12-8. Regression for final subjective norm with seven alternative services available. Motivation to Comply with: Mean (SD) My family 5.1 (2.0) My friends 4.2 (1.9) My coworkers 2.8 (1.6) My neighbors 2.5 (1.6) Table 12-7. Mean ratings for final motivation to comply.

128 In the TPB, SN is influenced by the control beliefs and the power of each of those beliefs. Table 10-7 in Chapter 10 showed the mean control beliefs; these are shown again in Table 12-9. As can be seen, the highest rated item had to do with the need to make local trips. This was followed by three items rated second in magnitude: (a) the need for access to a car to make spur-of-the-moment trips, (b) the need for ac- cess to a car to carry heavy things, and (c) the bother of wait- ing for transit and not knowing when it was coming. Concern about being stranded was rated above neutral, at 4.7. The lowest rated beliefs were concern about getting downtown, encountering crime while walking, and dealing with the fare payment system. Regression analysis was used as an alternative method for judging the influence of the control beliefs. Table 12-10 shows a regression using the power of control variables and SCF from the final TPB exercise. The order of the coefficients shown in Table 12-10 is very different from the order of the control beliefs shown in Table 12-9. As can be seen in the regression results, there are only two significant coefficients, with “I worry about being stranded” having the largest mag- nitude. Those that said they would have less concern about being stranded also had a higher rating for SCF. The second significant variable had to do with making trips downtown. The negative coefficient says that respondents who agreed that it would be more difficult to get downtown tended to have lower self-confidence. Although, on the whole, respondents rated their need to get downtown the lowest of all of the control beliefs, their belief in their ability to travel downtown with the seven transportation options was significantly associated with their confidence in their ability to walk and take public transportation more. On the other hand, although they rated the control belief “I need to make local trips” highest, their belief in their ability to make local trips with the seven transportation options did not appear associated with their confidence that they could increase walking and public transportation use. The R2 for the regression shown in Table 12-10 is the low- est for the regressions shown in this chapter and indicates that there are many other factors underlying SCF than identified in this research. Summary This chapter examined the relationships between the direct measures of the TPB and also between the direct measures and the indirect measures, using the data from the Phase 2 Inter- net panel survey. The statistical technique of regression analy- sis was used to analyze the relationships between variables. As shown in prior chapters, SN, SCF, and intent increased significantly between the initial TPB exercise and the final exercise, but ATT did not increase significantly. Regression analysis was used to examine the relationship between the re- spondents’ intentions to increase their use of public trans- portation and walking and their ATT, SN, and SCF. In both the initial and final TPB exercise, intent to increase the use of Belief (Rated on a Seven-Point Scale) Mean (SD) I need to make local trips (to reach destinations such as the library, post office, restaurant, or coffee shop). (not very often to very often) 5.5 (1.6) I need access to a car to make spur of the moment trips. (not very often to very often) 5.1 (1.9) I need access to a car to carry heavy things (not very often to very often) 5.1 (1.8) I find waiting for the bus or train and not knowing when it is coming is a bother. (strongly disagree to strongly agree) 5.1 (1.9) I worry about being stranded if I rely on public transportation and miss the bus or train. (strongly disagree to strongly agree) 4.7 (2.0) I worry about crime or other disturbing behavior on public transportation. (strongly disagree to strongly agree) 4.1 (2.0) I need to travel to other parts of the region. (not very often to very often) 4.1 (2.1) I find dealing with the fare for public transportation is a bother. (strongly disagree to strongly agree) 3.9 (2.0) I worry encountering crime or other disturbing behavior when walking. (strongly disagree to strongly agree) 3.8 (2.0) I need to travel downtown (not very often to very often) 3.4 (2.3) Table 12-9. Mean ratings for final control beliefs.

129 public transportation and walking was most closely related to a respondent’s attitude. Intent was also related to SN and SCF; these were of similar influence to each other, but smaller influence than attitude. Regression analysis was used to examine whether the behavioral beliefs measured in the final TPB exercise were sig- nificantly related to the respondent’s attitude. The most important belief was found to be that with the new services available, “I would rely on public transportation and walking to get me to my destination in a timely way.” The next most important belief was a composite of beliefs about improving health and reducing pollution. While respondents increased their rating of their ability to rely on public transportation and walking, they decreased their rating of improving health and reducing pollution. This result may explain why attitude did not change significantly. Regression analysis was also used to examine the relation- ship between the normative beliefs and the final SN. The most important belief was found to be that “with the new services available, my family would be more supportive of my walk- ing more and taking public transportation more.” The sig- nificant increase in this normative belief corresponds with the positive change in the SN. Finally, regression analysis was used to examine the rela- tionship between the power of control ratings and SCF. The most important power of control statement was “With the new services available, I would have less concern about being lost or stranded by missing the bus or train.” The overall message of this exercise seems to be that to increase transit use and walking requires the following: • The perceived reliability of the system must be improved. • The positive health and environmental impact of walking more and taking public transportation more must be more convincing. • Customers must be convinced that they will not be left stranded. • Families must approve of increased transit use and walking. Dependent Variable: Perceived Behavioral Control (SCF) Independent Variable: Final Power of Control Coefficient t-Statistic Probability Constant 3.13* 8.33 .0001 Have less concern about being stranded 0.30* 4.52 .0001 Feel safer from crime and other disturbing behavior 0.09 1.47 .1425 Paying the fare would be simple 0.09 1.16 .2459 Easy to know the schedule 0.06 0.78 .4384 More difficult to get to the region 0.03 0.53 .5942 Harder to make spur of the moment trips -0.07 -1.28 .2025 Harder to carry heavy things -0.09 -1.57 .1163 More difficult to make local trips -0.05 -0.69 .4882 More difficult to get downtown -0.18* -2.71 .0069 * indicates significant coefficient at p < .05 R2 = 23%, 501 observations Table 12-10. Regression for final self-confidence with seven alternative services available.