Public Perception of Mileage-Based User Fees (2016)

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.

19 CHAPTER FOUR LEARNING FROM PUBLIC OPINION SURVEY RESEARCH ON MILEAGE- BASED USER FEES For each survey identified, the researchers were contacted to request additional information not available in published documents, including crosstabs, data sets, and survey ques- tionnaires. Many authors were willing to share additional information beyond what they had formally published. Determining Analysis Criteria For the quantitative meta-analysis of survey findings, numerous criteria hypothesized to be relevant for under- standing public opinion about MBUFs were reviewed. This larger list was narrowed down to those variables for which data existed in enough of the surveys to permit meaningful analysis. The criteria ultimately selected for analysis fall into two categories: 1. Characteristics of the surveys themselves: geographic scope, survey mode, sampling frame, survey sponsor type, year survey was conducted, and how the MBUF question was framed. 2. Characteristics of the survey respondents: gender, age, income, education, race/ethnicity, and political affiliation. Compiling Data Sets for Analysis Three primary challenges arose in compiling the data sets for the meta-analysis. First, the surveys framed questions about MBUFs very differently. Second, the structure and availability of the survey data were inconsistent. Third, the surveys did not provide consistent data about respondent characteristics, and many surveys did not provide this infor- mation in any usable form at all. The first challenge was that questions about MBUFs were phrased very differently across the surveys. Only two cat- egories of questions were identified that were asked with similar enough wording to make a quantitative analysis of the responses feasible: questions asking very generally about support for an MBUF and questions asking about support for replacing the gas tax with an MBUF. Questions related to privacy and fairness were separated out, but they were framed so differently in different surveys that a quantitative analysis of the responses would be meaningless. (For exam- ple, one question might ask whether privacy was a concern, This chapter presents the findings from a meta-analysis of 38 public opinion surveys that, collectively, included 167 unique questions about mileage-based user fees. The first section describes the methods used for finding and analyzing the polls; the following section describes the characteristics of the polls and the main categories of questions focused on MBUFs. Next, an analysis of general support for MBUFs is presented, including a discussion of how support varies by respondents’ demographic and socioeconomic characteristics. Support for questions that focus on replacing the gas tax with an MBUF is analyzed, followed by analyses of questions asking about privacy and fairness related to MBUFs and a brief mention of MBUF-focused questions framed around other topics. The chapter concludes with a summary of key findings. METHODS FOR FINDING AND ANALYZING SURVEY RESEARCH Identifying and Obtaining Surveys for Analysis Many search strategies were used to identify and collect rel- evant surveys and polls. Internet-based public opinion poll archive databases were searched (e.g., Rasmussen Reports, SurveyUSA, and PollingReport.com) as well as research- oriented online databases (e.g., Google Scholar, Web of Science, and ScienceDirect) and the general search engine Google Web. (Additional information regarding these resources is available in Appendix C.) In all the searches, the following key words and phrases (and variations) were used: mileage-based user fee, vehicle miles tax, MBUF, VMT, poll, survey, road usage charge, and public opinion. Finally, individual researchers were contacted for additional information if their poll was referenced in the text or refer- ences section of another report or article. A list of identified surveys was distributed to the mem- bers and affiliates of TRB’s Congestion Pricing Committee, Revenue and Finance Committee, and Mileage-Based User Fee Subcommittee, with a request that members inform the study team about any additional surveys or polls that were missing from the list. This request yielded a few items to add to the list. These search methods produced a total of 38 public opinion polls that included 167 questions on mileage-based user fees.

20 while another question asked how privacy ranked compared with other potential MBUF concerns.) Finally, a large set of questions on other topics was compiled into a table to cre- ate a database of question-wording ideas, but they were not analyzed because the topics were so different. The second challenge was that the structure and avail- ability of the survey data were inconsistent. In some cases the survey authors generously shared a complete raw data set, allowing us to analyze the response data by subgroups of respondents (e.g., men versus women) and to recode vari- ables as necessary for comparison across surveys. However, in most cases the data were available only in the form of topline frequencies or a summary document with a few key statistics from the poll. In the latter cases, it was often impossible to perform analysis according to a respondent characteristic of interest, even if the survey included a ques- tion asking about that characteristic. For example, if a survey asked about household income but did not report the results by those categories and no data set was provided, the study findings could not be analyzed by income group. A third challenge was that surveys used very different methods to collect demographic and socioeconomic infor- mation about respondents. Some surveys collected a wide range of information, while others collected only a few pieces of such information or none at all. In addition, the response options provided were not consistent across sur- veys, so it was often impossible to identify a common set of responses for analysis. For example, many surveys asked people what age category they fell into, but the categories varied among surveys. As Table 3 shows, only 20 surveys at most provided the data required to analyze responses by particular sociodemographic characteristics; in most cases, far fewer surveys provided the needed data. Because of the limited number of surveys providing data by subgroups, it was not possible to test whether certain survey or respondent characteristics are statistically sig- nificantly correlated with support levels. Several statistical approaches were considered, including the test of two pro- portions (comparing each group to a base case for that cate- gory) and the multivariate test of means. However, given the small sample size and relatively small differences in mean support across categories, none of the statistical tests consid- ered would provide reliable results. Thus, while the findings presented here can be considered potentially significant pat- terns, more surveys on MBUFs will be necessary to confirm that the findings are statistically valid. DESCRIPTION OF MILEAGE-BASED USER FEE PUBLIC OPINION POLLS AND QUESTIONS ANALYZED A total of 38 unique public opinion polls were identified that included questions on MBUFs. The polls were pub- lished from 1995 to 2015. Table 4 provides an overview of the polls based on geography, survey mode, sampling base, and type of poll sponsor. Nearly half of the polls have a national focus, while approximately a third focus on the state level. Among the non-national polls, the most frequently surveyed region of the country is the West. A detailed summary of all 38 polls is available in Table A1 in TABLE 3 SUMMARY OF NUMBERS OF SURVEYS WITH COMPLETE DATA ON RESPONDENT SUPPORT, BY DEMOGRAPHIC VARIABLE Variable Variable Subgroupsa Number of Surveys with Complete Data Gender Female 20 Male 20 Age Young (18–24 years; 18–29 years; 18–34 years) 20 Middle (25–54 years; 30–49 years; 30–59 years; 35–54 years) 17 Older (50+ years; 55+ years; 60+ years) 17 Race/Ethnicity White 16 Black 14 Asian 8 Hispanic 10 Other 9 Income Lower (<$40K annually; <$50K annually) 15 Middle ($40–100K annually; $50–100K annually) 15 Higher ($100K+ annually, $110K+ annually) 18 Education High school or less 12 More than high school 11 Political affiliationb Democrat/Liberal 13 Republican/Conservative 13 Independent/Moderate/Other/ Decline to State 13 a Text in parentheses indicates different ranges that were grouped together to form a single subgroup. Efforts were made to create subgroups that were as consistent as possible across surveys in order to increase the sample size while still maintaining data integrity. b Political affiliation was most often presented in the following subgroups: Democrats, Republicans, Independents, and Decline to State. In addition, some surveys used groupings as follows: (1) Democrat, Republican, and Independent; (2) Democrat, Republican, and Other, including Independent; (3) Democrat, Republican, and Other; and (4) Liberal, Conservative, and Moderate.

21 Appendix A, and additional information about every poll follows the table. The majority of polls were conducted by phone (58%), with one-quarter using online survey modes, although in some cases the online surveys use survey panels initially recruited by phone. The vast majority of polls recruited among all adults, while a minority surveyed only registered voters and a few polled only individuals who had participated in an MBUF pilot program. Nearly two-thirds of the polls identified were conducted by either academic organizations (29%) or government agencies (37%). The remaining polls were conducted by industry groups, news outlets, polling firms, and other types of organizations. The peak year for polling on MBUFs was 2011, when 21% of all identified polls were conducted. The most recent 5-year period (2011–2015) saw more than twice as many polls as the previous 5-year period (2006–2010). The 38 polls included 167 unique survey questions about MBUFs (see Table 5). The average number of MBUF- focused questions per poll was four, with 53 as the maximum number of questions and one as the minimum. Twenty-one polls had just one question. TABLE 5 MBUF SURVEY QUESTION THEMES Question Themes Percentage of Survey Questions N General support for an MBUF 20 33 Support for replacing the gas tax with an MBUF 16 27 Privacy 7 11 Fairness 8 14 Other topics 49 82 Note: Total number of survey questions on mileage-based user fees = 167. The content of the questions falls into five main areas: 1. Questions that focus on general support or opposition to mileage-based user fees. These questions assessed support for the general concept of MBUFs. They do not specify whether the MBUF would replace the existing gas tax. 2. Questions that specifically focus on support or oppo- sition to replacing the existing gas tax with a mileage- based user fee. 3. Questions related to privacy. TABLE 4 CHARACTERISTICS OF THE PUBLIC OPINION SURVEYS ANALYZED Category Number of Surveysa Geographic focus Local 3 National 18 Regional 2 State 15 Census regionb Midwest 4 Northeast 1 South 5 West 9 Midwest/Westc 1 Survey mode In-person 1 Mail 1 Multiple modes 4 Online 10 Phone 22 Sampling frame Adults 28 Registered voters 8 MBUF pilot program participants 2 Sponsor type Academic 11 Government agency 14 Industry trade group/firm 3 News organization 2 Polling firm 4 Other (philanthropy, political organi- zation, or think tank) 3 Year data collectedd 1995 1 2006 1 2008 1 2009 3 2010 4 2011 7 2012e 6 2013 4 2014 6 2015 1 Note: Total number of surveys = 38. The poll is still live, but based on reader comments, the vast majority of respondents seem to have participated in 2012. Therefore, this study treats the poll data as coming from 2012. a Numbers within each category may not sum to 100% due to rounding. b For the 20 non-national polls. c One regional poll extended over two census regions. d Data provided for only 34 polls. Four polls either did not indicate the polling year or collected data over multiple years. e One poll administered in 2012 was an online, convenience sample poll conducted with readers of a newspaper article.

22 4. Questions related to fairness. 5. Other types of questions focusing on mileage-based user fees. The thematic areas most commonly tested were general support (20% of questions) and support for replacing the existing gas tax with an MBUF. Approximately half the sur- vey questions fell into the “other” category. The wording and basic response frequencies for each MBUF question can be found in Tables B1 through B5 in Appendix B. Each table focuses on questions from one of the five themes. ANALYSIS OF GENERAL SUPPORT FOR A MILEAGE- BASED USER FEE A total of 33 poll questions asked about general support or oppo- sition to mileage-based user fees without specifying that the fee would replace the existing gas tax (see Table B1 in Appendix B for a list of all questions and a summary of responses). Average support was 24%, with a range from 8% to 50%. Support by Personal Characteristics Table 6 shows how support for general MBUF survey ques- tions varies according to respondents’ personal character- istics. The most striking finding is that there is very little TABLE 6 SUPPORT FOR GENERAL MBUF SURVEY QUESTIONS, BY RESPONDENT CHARACTERISTICS Variable Mean Support (%) Minimum Support (%) Maximum Support (%) Na Overallb 24 8 50 33 Gender Female 26 7 56 25 Male 26 10 45 25 Agec Young 30 8 58 25 Middle 27 8 50 22 Older 27 7 50 22 Race/Ethnicity White 26 11 47 18 Black 28 6 53 16 Asian 45 19 72 11 Hispanic 30 13 54 12 Other 29 8 49 12 Incomed Lower 29 1 54 20 Middle 28 12 52 20 Higher 28 5 49 22 Education High school or less 28 6 52 16 More than high school 26 8 50 17 Political affiliatione Democrat/Liberal 32 15 57 16 Republican/Conservative 21 6 43 16 Independent/Moderate/Decline to State/Other 23 9 44 16 Note: Total number of survey questions focused on general support for mileage-based user fees = 33. a Sample size varies across the variables because some polls did not provide data for all demographic categories or did not provide data in a way that could be coded for analytical purposes. b “Support” included responses in the following categories: strongly support or support; 8–10 on a 10-point Likert scale (1 to 10) with 10 = strongly support; and 7–10 on an 11-point Likert scale (0 to 10) with 10 = strongly support. c “Young” included responses in the following categories: 18–24 years, 18–29 years, and 18–34 years. “Middle” included responses in the following categories: 25–54 years, 34–54 years, 30–49 years, and 30–59 years. “Older” included responses in the following categories: 50+ years, 55+ years, and 60+ years. d “Lower” included responses in the following categories: less than $50,000, less than $40,000. “Middle” included responses in the following categories: $50,000–$100,000 and $40,000–$100,000. “Higher” included responses in the following categories: $100,000+ and $110,000+. e Political affiliation was categorized differently in different surveys, using the following groupings: (1) Democrats, Republicans, Independents, and Decline to State; (2) Democrat, Republican, and Independent; (3) Democrat, Republican, and Other, including Independent; (4) Democrat, Republican, and Other; and (5) Liberal, Conservative, and Moderate.

23 variation in mean support by sociodemographic characteris- tics. Mean support among males and females is identical, and there is virtually no difference on the basis of age. Looking at race and ethnicity, whites express the lowest mean support for MBUFs, but the difference between whites and most other racial or ethnic groups is very small. Asians express the high- est mean support, but only 11 survey questions were included in the analysis; results from such a small sample size should be interpreted with caution. There are virtually no differences in mean support on the basis of income or education. The one exception to the overall lack of variation accord- ing to personal characteristics is that Democrats and liberals are more supportive of MBUFs compared with Republicans, conservatives, independents, moderates, those with other political affiliations, and those who declined to state a politi- cal affiliation. Support by Survey Characteristics Support for the MBUF survey questions was also examined on the basis of a variety of characteristics of the surveys themselves, including the sponsoring agency, mode of sur- vey administration, geography, and sampling frame. Results are shown in Table 7. There is a noticeable difference in mean support for gen- eral MBUF questions depending on the type of entity that sponsored the survey. Specifically, support tends to be high- est when the sponsor is an academic institution or govern- ment agency (27% and 25%, respectively). The lowest mean support level (11%) was found for surveys conducted by a firm or industry trade group, although the sample size here was quite small, so whether this pattern would hold across a larger number of surveys is uncertain. Turning to survey administration mode, random-digit- dial phone surveys and online surveys had very similar support levels. The mail surveys had the lowest mean sup- port, while the single mixed-methods survey showed a dra- matically higher level of support than all the other surveys. An important caveat about this finding is that only a few polls were conducted by mail, online, or mixed methods, so whether this pattern would hold across a larger number of surveys is uncertain. Looking at geography, there are some differences cor- related with Census region and possibly with geographic scale. In terms of Census regions, the highest mean support occurred in the West (37%), while the lowest mean support level occurred in the Northeast and Midwest (both at 17%). Support levels varied little whether the survey was a broad national survey or focused more narrowly on a specific region or state. Support levels were noticeably higher for the one poll conducted at the local level, but one cannot confirm a meaningful correlation based on a single poll. A noticeable difference in mean support was also found depending on the sampling frame. When surveys are broadly directed toward all adults, support levels are higher than support in surveys that focus on registered voters. However, sample size makes this finding uncertain, because only three surveys sampled registered voters. Another area considered for analysis was whether sup- port varied according to whether the survey presented the MBUF as a hypothetical scenario or survey participants were directly involved in a pilot program. However, none of the survey questions framed as general support for MBUFs were part of a pilot program, so that particular analysis was not possible for this theme. Support Over Time Another question explored was whether any potential trends in support exist over time. Figure 1 shows the percentage TABLE 7 SUPPORT FOR GENERAL MBUF SURVEY QUESTIONS, BY SURVEY CHARACTERISTICS Survey Characteristic Mean Support (%) N Sponsor type Academic institution 27 18 Government agency 25 9 Industry/Industry trade group 11 2 Polling firm 15 3 Other 17 1 Census region Midwest 17 5 Northeast 17 1 South 27 2 West 37 7 Midwest/Westa 20 5 Geography National 23 13 Regional 20 5 Local 33 1 State 26 14 Survey administration mode Mail 20 5 Online 26 2 Phone 25 23 Multiple 44 1 Sampling frame Adults 25 30 Registered voters 16 3 Note: Total number of survey questions focused on general support for mileage-based user fees = 33. a One poll with five unique MBUF-related questions extended over two census regions.

24 support for MBUFs by the year the poll was conducted. A trend line is also shown. The figure shows that the data are quite dispersed, with only a very slight upward trend. The correlation coefficient between the two variables is only 0.002, suggesting that there is no trend in support over time. Of course, small sample size hampers our ability to general- ize this finding with confidence. 0 10 20 30 40 50 60 1990 1995 2000 2005 2010 2015 Pe rc en t S up po rt Polling Year FIGURE 1 Percentage of support for MBUF by polling year. Note: Sample size = 28. Five questions on general support for MBUFs are excluded from this figure because the poll extended over multiple years or the information on the year the poll occurred was unavailable. ANALYSIS OF MILEAGE-BASED USER FEE QUESTIONS FOCUSED ON REPLACING THE GAS TAX Mean support for MBUFs was also examined when the survey question specifically asked about replacing the current gas tax with an MBUF (see Table B2 in Appen- dix B for the wording and response frequencies for all these questions). A total of 27 survey questions focused on replacing the gas tax with an MBUF. Four of the ques- tions were from surveys conducted as part of a pilot pro- gram, and those four are excluded from the following analysis unless otherwise noted. These questions were excluded because pilot participants cannot be compared with the general population for various reasons. Because they are in the pilot, participants are much better edu- cated than the general population about how MBUFs function. Also, they may have unusually positive opin- ions because many are paid small amounts of money in exchange for their participation. Mean support across the 23 survey questions that pre- sented an MBUF as a hypothetical scenario was 23%, with a range from 8% to 42%. These values are very similar to support levels for the general MBUF questions discussed in the previous section. Support by Personal Characteristics Table 8 presents results for an analysis of support for replac- ing the gas tax with an MBUF, breaking respondents into groups according to various personal characteristics. As in the analysis of survey questions asking generally about support or opposition to MBUFs, the small sample size pre- vented us from conducting reliable statistical testing, but some trends emerged that could be further explored as more survey data become available. As in the previous analysis, there is little variation on the basis of demographic and socioeconomic characteristics. No strong correlation emerges between mean support for replac- ing the gas tax with MBUFs and gender, income, or race/ ethnicity. Younger adults appear slightly more supportive, as do people with lower levels of formal education. Finally, as with general support for MBUFs in Table 6, Democrats and liberals are more supportive than Republicans, conservatives, independents, moderates, those with other political affilia- tions, and those who declined to state a political affiliation. Support by Survey Characteristics Table 9 presents data on mean support for replacing the gas tax with a mileage-based user fee on the basis of various overall survey characteristics such as sponsor, geography, mode of administration, sampling frame, and whether the MBUF was presented as a hypothetical scenario or as part of a pilot pro- gram. Findings should be interpreted with caution given the small sample sizes; in many cases, only a single survey question representing a particular survey type was available for analysis. Support levels did not vary a great deal by sponsor type. The highest support levels for replacing the gas tax with an MBUF occurred when the sponsor was a polling firm, but this finding is based on only a single question. The next high- est level of support was found when the sponsor was a gov- ernment agency. Lower support levels were found in surveys conducted by academic institutions and news organizations (although the latter is based on only a single question). In terms of geography, support was highest in the West, followed by the South and then the Midwest. Looking at geo- graphic scale, there was little variability depending on whether the survey was conducted at the state or national level. One survey conducted at the regional level had a noticeably lower support level; however, that finding should be interpreted with caution because it is based on a single data point. Looking across survey administration modes, the high- est level of support was found for online surveys, with the lowest support level coming from a single survey that used multiple modes. With a single data point, results may relate more to other characteristics of that particular survey rather than to the mode alone.

25 There was very little variation in support on the basis of whether the surveys sampled all adults or only registered voters, although only two questions were asked of the latter. Support Over Time Figure 2 shows the trend in percentage support for replacing the gas tax with an MBUF according to the year the poll was conducted. As with the previous analysis of general support for MBUFs (see Figure 1), the data points are quite scattered. However, in this case, the trend line is noticeably positive and the correlation coefficient between the two variables is 0.30, suggesting that support for replacing the gas tax with an MBUF is increasing over time. This trend of support increasing over time aligns with social psychology literature that finds message repetition to be a key factor in changing public opinion and attitudes toward an issue. Research suggests that repeated exposure to a par- ticular message can lead to more positive attitudes (Zajonc 1968), although Cacioppo and Petty (1989) suggest that the message repetition theory is nuanced; repetition allows more opportunities to scrutinize the argument, and strong argu- ments lead to more favorable opinions than weak arguments. TABLE 8 SUPPORT FOR REPLACING THE GAS TAX WITH AN MBUF, BY RESPONDENT CHARACTERISTICS Variable Mean Support (%) Minimum Support (%) Maximum Support (%) Na Overallb 23 8 42 23 Gender Female 30 12 37 6 Male 28 14 43 6 Agec Young 32 17 40 6 Middle 26 12 40 6 Older 30 13 41 6 Race/Ethnicity White 21 12 28 3 Black 24 15 33 3 Asian 26 18 35 2 Hispanic 25 15 33 3 Other 21 16 26 2 Incomed Lower 30 14 38 5 Middle 31 12 42 5 Higher 29 13 64 6 Education High school or less 27 16 37 6 More than high school 20 12 28 2 Political affiliatione Democrat/Liberal 26 16 38 3 Republican/ Conservative 19 10 26 3 Independent/Moderate/ Decline to State/Other 21 15 28 3 Note: Total number of survey questions focused on replacing the gas tax with mileage-based user fees using a hypothetical scenario as opposed to a pilot program = 23. Including the four questions from a pilot program increases the overall mean support to 27%, with a range from 8% to 71%. a Sample size varies across the variables because some polls did not provide data for all demographic categories or did not provide data in a way that could be coded for analytical purposes. Key demographic information associated with the four questions from pilot programs was not available, so the data presented here is only from survey questions using a hypothetical scenario. b “Support” included responses in the following categories: strongly support or support; 8–10 on a 10-point Likert scale (1 to 10) with 10 = strongly support; and 7–10 on an 11-point Likert scale (0 to 10) with 10 = strongly support. c “Young” included responses in the following categories: 18–24 years, 18–29 years, and 18–34 years. “Middle” included responses in the following categories: 25–54 year, 34–54 years, 30–49 years, and 30–59 years. “Older” included responses in the following categories: 50+ years, 55+ years, and 60+ years. d “Lower” included responses in the following categories: less than $50,000, less than $40,000. “Middle” included responses in the following categories: $50,000–$100,000 and $40,000–$100,000. “Higher” included responses in the following categories: $100,000+ and $110,000+. e Political affiliation was categorized differently in different surveys, using the following groupings: (1) Democrats, Republicans, Independents, and Decline to State; (2) Democrat, Republican, and Independent; (3) Democrat, Republican, and Other, including Independent; (4) Democrat, Republican, and Other; and (5) Liberal, Conservative, and Moderate. TABLE 9 SUPPORT FOR MBUF SURVEY QUESTIONS FOCUSED ON REPLACING THE GAS TAX, BY SURVEY CHARACTERISTICS Survey Characteristic Mean Support (%) N Sponsor type Academic institution 20 9 Government agency 26 11 News organization 20 1 Polling firm 30 1 Other 23 1 Census region Midwest 17 5 South 27 5 West 35 3 Geography National 21 10 Regional 15 1 Local — 0 State 26 12 Survey administration mode Online 26 14 Phone 19 8 Multiple 13 1 Sampling frame Adults 23 21 Registered voters 21 2 Note: Total number of survey questions focused on replacing the gas tax with mileage-based user fees using a hypothetical scenario = 23.

26 It is important to note, however, that repeated negative mes- saging can lead to decreased support, as shown by Fernandes’ (2013) analysis of negative political messaging. 0 5 10 15 20 25 30 35 40 45 1990 1995 2000 2005 2010 2015 2020 Pe rc en t S up po rt Polling Year FIGURE 2 Percentage of support for replacing the gas tax with an MBUF by polling year. Note: Sample size = 23. Four questions on support for replacing the gas tax with an MBUF were excluded from the analysis because they are drawn from pilot programs instead of presenting the MBUF as a hypothetical scenario. Support for these four questions ranged from 37% to 71%. Support Among Pilot Program Participants for Replacing the Gas Tax As noted earlier, our analysis in this section focused primarily on questions that presented the MBUF as a hypothetical sce- nario (the vast majority of the questions asked about replacing the gas tax). However, four questions of this type were asked of participants in two separate pilot programs. Support among pilot participants ranged from 37% to 71%, with a mean of 51%. In one pilot program, respondents were surveyed three times (at the beginning, middle, and end of the pilot), with the highest support level (71%) at the end of the program. Even at the beginning of the program, 42% of participants sup- ported replacing the gas tax with an MBUF. This is consider- ably higher than the mean support levels among individuals responding to questions posed as hypothetical scenarios. The high level of support at the beginning of the pilot suggests that these participants already had significant interest in MBUFs. In addition, pilot participants are often provided with some type of incentive for participating in the program, which might generate positive feelings. ANALYSIS OF MILEAGE-BASED USER FEE QUESTIONS ABOUT PRIVACY Only four surveys asked questions related to MBUFs and privacy. Among them, they posed a total of 11 questions, six of which came from a single survey. (Table B3 in Appen- dix B presents the specific wording and responses to each of these.) The questions focused on very different aspects of privacy issues, making comparisons across questions almost impossible. However, the questions can be organized into the following general categories: • Questions asking whether collecting mileage data from drivers is an invasion of privacy (two questions, from one survey). • Question asking whether respondents dislike a GPS- based MBUF because the government can monitor their driving patterns (one question). • Questions asking how privacy issues rank compared with other issues as a concern with MBUFs (two ques- tions, from two surveys). • Questions asking whether certain program design fea- tures would reduce privacy concerns (four questions, from three surveys). • Questions about outsiders accessing government-held data on GPS mileage (two questions, from one survey). The responses to the 10 questions for which the survey authors provided response data show that privacy is a seri- ous concern. For seven of the 10 questions, more than half of the respondents indicated that privacy was a concern or that they preferred an MBUF program structure that did not collect information on where they drove. One other ques- tion asked respondents to rank five factors they liked least about MBUFs, and privacy was the factor chosen most often (by 40% of respondents). The last of the 10 questions asked whether a specific program structure designed to reduce pri- vacy concerns would make the MBUF acceptable; only 15% said yes. ANALYSIS OF MILEAGE-BASED USER FEE QUESTIONS ABOUT FAIRNESS Six surveys asked about issues of fairness and equity, with a total of 14 questions on this theme. Table B4 presents the wording and responses to these questions. The surveys probed issues of fairness with MBUFs in a variety of contexts: • Fairness to all drivers: – Is an MBUF a fair way to raise transportation rev- enues? (two questions, from two surveys) – Is an MBUF more or less fair than a gas tax? (four questions, from four surveys; there were two pairs of annual surveys) • Is an MBUF fair because it charges in direct proportion to highway use? (one question) • Fairness to certain classes of drivers: – Is an MBUF fair to people who drive vehicles that use little or no gasoline? (three questions, from two surveys) – Is an MBUF fair to rural drivers? (two questions, from two surveys)

27 – Is an MBUF fair to people who drive long distances for work? (one question) – Is it fair to charge a higher MBUF rate for heavy vehicles, because they cause more wear and tear on roads? (one question) Most respondents did not see MBUFs as a fair way to raise revenue. In response to the two questions that asked this in the most general way, only 33% and 45% of respon- dents said that MBUFs are fair. In response to the four ques- tions asking respondents whether they thought MBUFs were fairer than gas taxes, only 15% to 38% said yes. One survey asked this question both before and after giving respondents information about MBUFs, and the percentage of respon- dents who thought the MBUF was fairer than gas taxes rose 7 percentage points, from 31% to 38%. When respondents were asked how fair they thought MBUFs are to certain classes of drivers, majorities thought they were not fair. For example, in one survey, 79% of respondents believed that an MBUF is unfair to people who live in rural areas because they have to drive long distances, while 73% of respondents thought MBUFs are unfair to peo- ple who drive a lot for work. The second survey that asked if an MBUF is fair to rural drivers did so in the context of comparing an MBUF with a gas tax; in this survey, 57% of respondents believed that the MBUF is “less fair” than the gas tax. With respect to drivers of alternative-fuel or high-effi- ciency gasoline vehicles, somewhat fewer respondents had fairness concerns. From 31% to 56% of respondents believed an MBUF was unfair, depending on the survey. One sur- vey asked two questions on this topic, phrased in slightly different ways, and it is illustrative to note how differently respondents answered each question. Thirty-one percent of respondents agreed with the statement “Fees based on miles traveled are fair because they require drivers of vehicles that use little or no gasoline to also pay their fair share for using roads and bridges.” This is considerably less than the 46% who agreed with the statement “A vehicle-miles-traveled use fee program is one solution that is flexible enough to work with all vehicles so that they pay their fair share for use of the roadway system— high mileage vehicles, gas-electric hybrids, ethanol- and biofuel-powered vehicles, plug-in vehicles and other technologies.” MILEAGE-BASED USER FEE QUESTIONS ON OTHER TOPICS The remaining 82 MBUF questions contained in the surveys (about half the total number of questions) did not relate to any of the key themes. These questions ranged widely from prob- ing respondents’ familiarity with MBUFs to asking whether such a fee system would affect transit use to asking whether fees should vary depending on time of day or type of street. Table B5 presents the exact wording and responses for all these questions, as a resource for future survey researchers. CONCLUSIONS Quantity and Type of Surveys Conducted and Data Availability A key conclusion from this study is that only limited poll- ing data about mileage fees are available to policymakers and researchers. An extensive search using many academic and preferred databases and search engines, plus direct com- munication with knowledgeable members of the transporta- tion finance community, netted only 38 surveys with MBUF questions. This is far smaller than the known number of surveys on other transportation revenue sources such as gas taxes or tolls. By comparison, Agrawal and Nixon (2015) collected 108 surveys asking about gas taxes, and Zmud and Arce (2008) collected 110 surveys asking about tolls. For the 38 surveys with MBUF questions, data were collected on general survey characteristics including geo- graphic scope, survey mode, sampling frame, survey spon- sor type, the year the survey was conducted, and how each MBUF question was framed. In addition, information about survey respondents—such as gender, age, income, educa- tion, race/ethnicity, and political affiliation—was gathered whenever possible. In a large number of cases, however, respondent characteristics were unavailable. The surveys were conducted between 1995 and 2015, with more than half between 2011 and 2015. The majority of sur- veys were conducted by academic organizations or govern- ment agencies, used a random-digit-dialing phone method, and sampled adults at the national or state level. The 38 polls included a total of 167 unique survey ques- tions that focused on MBUFs. The average number of MBUF- focused questions per poll was four, with 53 as the maximum number of questions and one as the minimum. The questions fall into five main categories: (1) questions that focus on gen- eral support or opposition to mileage-based user fees, (2) questions that focus on support or opposition to replacing the existing gas tax with mileage-based user fees, (3) questions related to privacy, (4) questions related to fairness, and (5) other types of questions focusing on mileage-based user fees. The two thematic areas most commonly tested were general support for an MBUF (20% of questions) and support for replacing the gas tax with an MBUF (16% of questions). Support for MBUFs The surveys confirmed what the qualitative research on MBUFs suggested: support for MBUFs rarely reached 50%.

28 When survey questions about support for replacing the gas tax with an MBUF were analyzed to look for correla- tions with survey administration and personal characteristic factors, the patterns that emerged were generally the same as those for the questions on general MBUF support, with two exceptions. First, support for replacing the gas tax with an MBUF appears to be increasing modestly over time. Second, when support levels were considered on the basis of whether the MBUF was presented as a hypothetical scenario or in the context of a pilot program, support was much higher among respondents in a real pilot program. Support was as high as 71% for one survey question administered to program par- ticipants at the end of a pilot program. A key caveat to this finding is that three of the four questions directed toward pilot participants came from the same pilot project, which surveyed participants multiple times. Views on Privacy and Fairness The issues of privacy and fairness were prominent in the qualitative studies, but the surveys asked few questions related to either issue. Therefore, a meta-analysis of the sur- vey questions provides only a tentative sense of how gener- alizable these concerns might be. Also, the surveys provide virtually no details about the specific reasons people worry about fairness and privacy or the types of program design features that might reduce such worries. Only four surveys asked any questions related to MBUFs and privacy. Collectively, the surveys posed a total of 11 questions, six of which came from a single survey. The responses to the questions show that privacy is a serious con- cern. For seven of the 10 questions for which response data were available, more than half of the respondents indicated that privacy was a concern or that they preferred an MBUF program structure that did not collect information on where they drove. One other question asked respondents to rank five factors they liked least about MBUFs, and privacy was the factor chosen most often (by 40% of respondents). Six surveys asked about issues of fairness and equity, with a total of 14 questions on this theme. The questions fell into two categories: (1) Are MBUFs fair to all drivers? and (2) Are MBUFs fair to certain classes of drivers? (The groups explored were rural drivers, people who drive long distances for work, and people who drive vehicles that use little or no gasoline.) Comparing responses across surveys is challenging because the questions are framed so differently, including the fact that some ask if an MBUF is “fair,” some ask if it is “unfair,” and yet others ask respondents to decide whether an MBUF is more or less fair than a gas tax. How- ever, the overall picture that emerges from the surveys is one of concern about fairness. Fewer than half of the respon- dents thought MBUFs were a fair way to raise transporta- tion funds, whether in general or compared with the gas tax. Also, the survey questions asking about fairness to specific Looking across the 33 polls that asked very generally about support for an MBUF, the mean support was only 24%, with a range from 8% to 50%. Only six questions had support near to 40% or higher, and four of these came from an annual series of polls that repeated the same question each year. The analysis of whether support for MBUFs varies by per- sonal characteristics revealed very little variation in mean support by sociodemographic characteristics. For example, there are virtually no differences in mean support based on gender, age, income, or education. The one exception to the striking finding of how little support varied by the personal characteristics examined is that Democrats had higher mean support levels for MBUFs (32%) than Republicans (21%) or independents (23%). Support levels did not differ much according to whether a poll question was framed as general support for MBUFs or, more specifically, as replacing the gas tax. However, there is some limited evidence to suggest that support for replacing the gas tax with an MBUF is increasing over time. Several key findings emerge from the analysis of how the characteristics of the surveys themselves might corre- late with support for the general MBUF survey questions that did not specifically discuss replacing the gas tax. Fac- tors that revealed little clear correlation with support lev- els were survey year, geographic scale of the survey, and survey administration mode. On the other hand, the survey administration factors in the bulleted list below did correlate with support levels, although the number of surveys used for comparison was sometimes very small, so it is unwise to generalize from these results without additional research to confirm them: • Survey sponsor type: Support tended to be highest when the sponsor was an academic institution or gov- ernment agency (27% and 25%, respectively). The lowest mean support level (11%) was found for surveys conducted by industry or industry trade groups. An important caveat about this finding is that the num- ber of industry-sponsored polls was quite small, so whether this pattern would hold across a larger number of surveys is not at all clear. • Census region: Surveys from the West had the high- est mean support levels (37%), while those from the Northeast and Midwest had the lowest mean support (17% each). An important caveat about this finding is that the number of polls from the Northeast and Midwest regions was small. • Sampling frame: Surveys sampling among all adults had higher support levels (25%) than those that focused on registered voters (16%). However, because only three surveys sampled registered voters, it is unclear whether this correlation reflects a real difference in views between registered voters and all adults.

29 • Drivers would have a hard time paying periodic MBUF charges, compared with the relative ease of paying the gas tax in small, frequent increments. • Replacing the gas tax with an MBUF would cause the government to lose a policy tool that incentivizes the purchase of fuel-efficient vehicles. • An MBUF with a congestion pricing component would be unfairly expensive for people with inflexible job schedules. In addition to these concerns, three other issues came up in the qualitative research that were not explored in the surveys: • MBUFs would be desirable because, unlike fuel taxes, MBUFs fairly charge electric vehicles and fuel-effi- cient vehicles for their road use. • People prefer a simple road-charging system rather than a complex one. • People currently see no compelling reason to replace the gas tax with an MBUF. classes of people found no strong support for the claim that MBUFs are fair. Possible Concerns That the Surveys Failed to Explore The qualitative research on public opinion revealed a num- ber of issues that seriously concerned the participants, none of which was addressed by enough surveys to make a meta- analysis possible. These issues included concerns that— • Billing errors would be created because of problems with technology or incompetence on the part of the government agency administering an MBUF program. • Administering an MBUF program would be costly for the government and for drivers (if the latter had to install and maintain on-vehicle equipment). • In-state drivers would inaccurately be charged an MBUF on out-of-state miles. • Out-of-state drivers would not be charged for their travel in an MBUF state.