Transportation Attitudes Over Time: a Longitudinal Approach

ABSTRACT - Research into consumer transportation modal choice attitudes is receiving increased emphasis. This paper compares public modal choice criteria reflected at the peak of the "energy crisis" with those obtained one year later. Despite a relaxing of gasoline lines, no significant changes in ecological variables were noted, and other travel criteria remained constant. Public transportation remains perceptually inferior to private, and resistance to several mass transit subsidy plans may be increasing.

Citation:

Mark I. Alpert and Linda L. Golden (1978) ,"Transportation Attitudes Over Time: a Longitudinal Approach", in NA - Advances in Consumer Research Volume 05, eds. Kent Hunt, Ann Abor, MI : Association for Consumer Research, Pages: 194-200.

Advances in Consumer Research Volume 5, 1978      Pages 194-200

TRANSPORTATION ATTITUDES OVER TIME: A LONGITUDINAL APPROACH

Mark I. Alpert, The University of Texas at Austin

Linda L. Golden, The University of Texas at Austin

[We gratefully acknowledge the support of the Department of Transportation program of University Research, Contract DOT-OS-30093.]

ABSTRACT -

Research into consumer transportation modal choice attitudes is receiving increased emphasis. This paper compares public modal choice criteria reflected at the peak of the "energy crisis" with those obtained one year later. Despite a relaxing of gasoline lines, no significant changes in ecological variables were noted, and other travel criteria remained constant. Public transportation remains perceptually inferior to private, and resistance to several mass transit subsidy plans may be increasing.

Marketing and Consumer Psychology have recently begun to find application in attempts to model and promote mass transportation usage and public support (Blattberg and Stivers, 1970; Business Week, 1974; Hille and Von Cube, 1963; Hille and Martin, 1967; Mundy, Cravens and Woodruff, 1974; Sheth, 1975). Many of these studies have stressed the need to identify attitudes relevant to modal choice decisions and then tailor a marketing mix toward target markets of potential riders (and, occasionally, toward non-riding supporters of the mass transit system). However, little has been done to test the reliability of the scales that have been used to measure these attitudes toward transportation and its features. Techniques that have been developed elsewhere have been employed in this problem area, thinking that validity and reliability can be generalized across products. This may be true, but it would be useful to know if the attitudes that have been measured are relatively stable over time. Modifying an existing transportation system to build in benefits revealed as important deficiencies at one point in time might take so long that by the time the political and system problems have been negotiated (at great expense-e.g., BART), public attitudes may have changed and anticipated support may not develop. Therefore, we need to know whether the attitudes revealed in one study are still salient when measured again, say, a year later. If so, planning may be more secure. If not, one might extrapolate from longitudinal changes what will be the trends in attitudes and relative importance of modal choice criteria, so that these changes in the "ideal system" may be properly anticipated and programmed.

In the spring of 1974, when most of this study's first year's data were collected, the "energy crisis" was first being widely perceived by the public, with lengthened gasoline station lines, rapidly rising prices for gasoline, and increased rhetoric about the Arab oil embargo, self-sufficiency, and related issues. By the time Spring of 1975 had come, gas lines were a (temporary?) thing of the past, and talk of gasoline rationing was not heard amongst the general public. Accordingly, it seemed appropriate to explore whether the general public in the survey area might still seek the same configuration of transportation features, including the relatively high determinance of energy savings and low pollution per passenger. Comparisons could also be made between the relative desirability of various funding proposals for public transportation, given changes in economic and political circumstances of these two years. Changes in perceived images of private autos, buses, and the differences between the two, could also be monitored to see whether public transit was perceived as "gaining" significantly, due to changes in public attitudes transit improvements, or some combination.

Clearly, studying trends in attitudes and transit priorities over a two year period in one study area does not constitute a sufficient data base for generalizing about long-term trends. However, it was felt that some insight into sensitivities to "the energy crisis," and possible changes in a variety of criteria for modal choice might be gained through examining a community's responses to the same questions one year "after the crisis". To the extent that results can be generalized to the broader United States culture, trends (or stability) in this area might be indicative of broader, transportation-relevant societal trends as well.

METHOD

Sample and Data Collection

The findings reported in this paper are drawn from questionnaires administered to two separate samples of adults in Austin, Texas, a medium-sized (population about 300,000) southwestern city.

Like most cities built after the automobile, there is a relatively small central city surrounded by generally low density housing and decentralized centers for shopping and major industry. The city has grown very rapidly in the last ten years, and most of this resembles the kind of urban sprawl perhaps exemplified by Los Angeles. However, there remains a central employment-shopping attraction because of the centrally located University of Texas campus (42,000 students), and state capital complex. In addition, there has been a recent surge of downtown renewal and an attempt to establish evening theater and other entertainment.

Similar to other decentralized cities, travel patterns reflect the dominance of the car. According to the city's own surveys over 97% of trips are taken in private automobile. The bus service is marginal in areas of higher usage-lower income to downtown, but service is generally sparse throughout the rest of the city, with routes along major arteries, but relatively infrequent headways especially in off peak times.

During April through June, 1974, 252 usable responses were obtained from adults who were contacted in a stratified by census-tract (quotas proportional to population) area random sample of city households. Interviewers enumerated households within each census tract, with starting points determined by the researchers selecting random blocks within tracts and random corners and walking directions within blocks. Every third household was approached, with provisions for call-backs, staggered interviewing hours, and alternately selected male and female respondents (18 years and over). Due to the length and complexity of the survey instrument, interviewers were essential in insuring cooperation and providing clarifications to questions. To increase the speed and candor of responses, respondents filled out their own questionnaires, except in those households where translation to Spanish necessitated a more active role by the bilingual interviewers.

Data were collected in the same manner one year later, except that the sample size was trimmed by about one-third (159 usable responses), and certain sections of the questionnaire (not relevant to this analysis) were modified. While a longitudinal study ideally involves obtaining data from the same persons during two or more different time periods, this approach was varied for this study. We decided to apply similar criteria for sample selection in both years, but not to attempt to interview the same households. [Walking directions from starting points were arbitrarily shifted ninety degrees to minimize the chance that the same persons might be contacted two years in a row (none were).] It was felt that persons who were willing to respond a second time to our questionnaire, after having spent 45 minutes doing so a year earlier, were likely to be more positively biased toward public transportation than were those who merely completed the process one time. A lower response rate was also likely, and this would increase survey costs while lowering reliability. Inability to contact people who had moved since the first interview would also bias results, since original respondents were contacted door-to-door, and not from a year-old list. The study, thus, is longitudinal in the sense that similarly selected samples are used to generalize parameters for the survey area in 1974 versus 1975. Given the care that was taken to choose samples representing the community adult population, it is argued that differences between the mean responses for the two survey years would represent community attitude changes over time, provided these differences were significantly greater than those that could be allowed due to random sampling fluctuations. (Some demographic differences between the two samples were observed, but as will be discussed, these had minimal impact on the key comparisons between respondents for the two survey years. )

Overview of Questionnaire and Data Analysis

Part One of the questionnaire first obtained information concerning respondents' traveling frequency for trips to work (or school, if students) and mode usually selected. Next, subjects were asked to assume they were choosing a transportation mode for trips to work or school, and to evaluate 27 modal attributes (e.g., economy, convenience, energy use per passenger...) in two distinct ways. Initially, they were to indicate the relative importance of each attribute on a five-point scale ranging from "no importance" (scored as 1) to "extremely important" (scored as 5). Then, they indicated how much difference they perceived among various transportation modes in terms of each attribute listed. Five-point rating scales were again used, ranging from "no differences'' (1) to "extreme differences" (5). Scale positions had been previously found to be approximately equal intervals (Myers and Warner, 1968).

Scales for these importance and difference perceptions were multiplied together for each respective attribute and respondent, to obtain a measure of the "determinance" (Alpert, 1971; Myers and Alpert, 1968) of that attribute. Potential determinance scores for each subject and attribute thus ranged from 1 (no importance, no differences) to 25 (extremely important, extreme differences).

This combining procedure is based on the assumption that the relative weight of a specific attribute in determining whether or not a particular transportation mode is selected is a function of the combined effect of the importance of the attribute to travelers and the amount of perceived variation among alternative modes, in terms of that attribute. For example, avoiding traffic congestion was perceived as important, but it probably lacks determinance because many modes (in the survey area) are perceived as equally free from (or subject to) traffic congestion. Accordingly, perceptions of this attribute of local transportation modes probably do not influence modal choices nearly as much as, say, dependability, which has both importance and perceived variation among modes.

After providing these importance and difference perceptions, respondents next rated a personal car along five-point semantic differentials (e.g., Economical:______:_____:______:_____:_____:Expensive) to indicate its suitability for these commuter trips in terms of each of the above 27 attributes. This format was also used to secure ratings of a bus's attributes for the same trip purpose.

Respondents indicated their attitudes toward financing public transportation, as well as transportation's role in city planning, pollution, and so forth. This section also contained a "would you use city mass transit if improved..." question which was used to identify nonusers of public transportation who would be likely switchers to an improved system. Comparisons between this segment and the rest of the sample(s) have been reported elsewhere, and provide insight into segmentation strategies for public transportation marketing. Two additional questionnaire sections obtained media and demographic data to assist in this segmentation. In this paper the demographic data are used to compare the two years' samples, to see if demographic differences across samples might influence on the divergence (or lack thereof) of findings for the different survey period.

RESULTS

Table 1 presents a descending ranking of the determinance scores of the 27 characteristics of transportation modes for work/school trips, as rated by the 1974 sample. The "z-values" represent the comparison of the mean determinance rating for each attribute with the mean for all attributes, adjusting for the standard deviation of these ratings, and the number of persons rating each attribute. This is not a strict statistical test, since the true universe mean and sigma are unknown, but it provides a reasonable cut-off for "how high is high." The right-hand column summarizes the results of comparing the perceived images of cars vs. buses for commuter trips, in terms of attributes such as economy, dependability, and the like. (Table 4, below, gives mean scores for each).

One can note that for the ten attributes which were significantly high in determining modal choices, buses were viewed as superior in six and cars in only four. Table 4 provides statistical details and mean image profiles for these comparisons, which were analyzed using analysis-of-variance, with repeated measures (bus vs. car) for each dependent variable (Veldman, 1967). While a more precise quantification of the utility model underlying modal choices is being estimated by currently ongoing conjoint measurement research by the authors, Tables 1 and 3 may be interpreted to show that cars had sufficiently large perceived superiorities along highly determinant attributes (such as convenience and dependability) that more than offset the perceived superiorities (typically smaller in magnitude) of buses in features less determinant of modal choice. Further, since this set of features was drawn from traits generally considered to be positive benefits in transportation modes, one might argue that virtually all of them are desired, albeit only ten, were seen as exceptionally desirable. Across the set of 27 features, cars were seen as superior in 16, buses in nine, and two features were non-differentiating.

TABLE 1

DETERMINANCE SCORES AND MODAL COMPARISONS FOR 1974 ADULT SAMPLE (WORK/SCHOOL)

TABLE 2

DISCRIMINATION BETWEEN DETERMINANCE SCORES 1974 VERSUS 1975

While specific policy recommendations cannot be made without directly analyzing the determinant attributes for potential switchers to public transit (rather than the general public), this longitudinal study sought to compare general community attitudes and criteria for modal choice. Overall changes would be important indicators of general community trends, independent of their importance to various sub-segments of transportation interest. Table 2 shows a remarkable degree of similarity between the profile of determinance scores derived during the two years. Observation of the means for each attribute for both years (averaging the product of importance x perceived differences for each attribute, within each sample), shows almost identical statistics for both years. Attempting to discriminate Year One versus Year Two respondents on the basis of these 27 variables would be futile, since the Wilks Lambda statistic evaluated by the linear discriminant analysis model has an estimated 97 percent probability of being due to chance or sampling fluctuations. In other words, one could not assert that the general profile of criteria for modal choice changed form 1974 to 1975 without taking a 97 percent chance of being incorrect. Furthermore, not one of the attributes was rated as significantly more or less determinant in 1975 than in 1974, even though at the .05 level of significance one would expect between one or two such fluctuations due to chance. Of specific interest is the fact, shown in Tables 2 and 4, that energy usage and pollution remain important criteria (and perceived advantages of public transportation), one year after the temporary peak in the "energy crisis." Freedom from repairs and parking problems may be gaining, but not significantly so, and these kinds of variations have to be considered due to sample fluctuations. Should any trends develop over a longer time span, changes in determinance of various features may prove relatively favorable or unfavorable to public transportation. At this point, the relative modal choice criteria in this community seem stable add retain the mix of attributes in which public transportation was initially seen as superior in some traits and inferior in others. Next, let us examine whether changes in the relative ability of these modes to provide these features changed during this one year period.

Tables 3 to 6 provide considerable detail regarding relative images of cars versus buses during both years, as well as changes in car image and changes in bus image over time. Examining these data, one would have to conclude much the same things as were said above about criteria for modal choice. Not only were the determinance scores stable, but people's perceptions of the relative ability of buses versus cars in supplying these attributes were essentially stable during this time period. The 1974 mean profiles of car versus bus shown in Table 3 indicates patterns of relative superiority for cars in convenience, privacy, dependability and the like, and relative superiority of buses in avoiding parking problems and repairs, as well as ecological advantages. The 1975 mean profiles of car versus bus shown in Table 4 show that the same basic patterns of pluses and minuses were noted one year later. In general, where there was a low probability of obtaining sample means for car versus bus images due to changes for a particular attribute in 1974 (right-hand column in Table 3), there was always a low probability of attributing the perceived gap between the two modes as rated in 1975, in terms of the same attribute (right column in Table 4). In other words, where significant differences were found between the two modes' characteristics in one year, these tended to be observed in the next year.

TABLE 3

1974 ADULTS CAR VERSUS BUS

The reason for this is implied by the relative stability of car and bus perceptions during this time frame. For example, a discriminant analysis of the 1974 versus 1975 perceptions of car attributes for commuting indicated that there was a 67% probability that the differences obtained between profiles of mean scores were due to change. Thus, one could not conclude that car image changed significantly during this time period without taking more than an acceptable risk of being mistaken (the type-I error probability would be far more than the usual .05 level). The profiles of mean scores for car images in 1974 versus 1975 are almost identical. (The mean scores for cars and buses overtime maybe deserved in different form in Tables 3 and 4. Due to lack of significant differences and space limitations tabular presentation of the discriminant analyses are not presented here.)

There was even more stability in the perceived image of buses as a commuter mode in this area. The Wilks Lambda statistic for overall discriminability of the 27 attribute ratings for 1974 versus 1975 was again nowhere near statistically significant (alpha = .85). For both the car and bus images, given the lack of overall significance between profiles for the two years, and given no more than three attributes that appeared to change significantly at the .05 level, (with 2.7 expected changes out of 54 comparisons, due to chance), it would be unwise to attempt to attribute any meaning to either of the "perceived changes" for either mode. Perhaps as conditions in the environment change more dramatically, and as more people begin to utilize public transportation, changes in the relative utility of the two major modes might be reflected in their perceived images. During the 1974 to 1975 time frame in this major southwestern city, no significant changes in perceptions can be reliably reported.

The last two tables show the only instance of real variation between the data for the two survey years, although here, again, the practical significance of these differences for transit planning purposes is quite marginal. In Table 5, we note that there were found significant changes, from 1974 to 1975, in mean desirability scores for seven of the ten evaluated financing alternatives for public transportation. All of these shifts were in the direction of lower desirability for various subsidy plans. However, the relative ordering of these alternatives was virtually unchanged from 1974 to 1975, with a Spearman rank correlation coefficient of .976 (significant at beyond the .01 level). Moreover, the frequently proposed subsidy from the "highway trust fund" is still relatively favorably received, provided the subsidy is a relatively minor part of the non-rider burden (one or two cents tax per gallon.) "No-fare" plans for riders continued to lack popular support, only "more so" than before, and attitudes toward electric bill subsidies are the most negatively perceived financing mechanism. The increasing resistance to property and electric bill tax subsidies may be partially due to increased unemployment in the study area (and in the U.S.) from 1974 to 1975, along with a dramatic increase in electric utility bills.

Table 6 suggests that the generally more conservative approach to tax and other subsidies for public transportation might also be partially due to differences in the demographic composition of the second year sample. Compared to the respondents from 1974, the 1975 group was significantly less female (50 percent versus 62 percent), older (mean age about 37 versus 35.7), longer residents in the city (mean about 6 months longer), and less educated (by about one-half year of formal education). Most of this difference is probably due to tighter controls over the representativeness of the sample, since the 1975 group is somewhat more representative of the universe. However, it should be noted that although statistically significant, most of these differences are slight, and apparently had impact more on the financing attitude profiles than on the modal choice criteria and mode images. To check this, we correlated demographic variables with the other survey questions, with particular attention given to the correlations with age, education, sex and time in the city, as large correlations with these variables might have confounded the changes(or counteracted what would have been changed, where none were reported). The results indicate that little effect can be attributed to demographic variations between the two samples. The highest correlation between any of these four variables and criteria for modal choice showed that more educated people sought more convenience than did less educated people. Thus, convenience should have become less determinant in 1975, given a less educated sample (and it did drop slightly, but not significantly). However, the shared variation between education and need for convenience was less than 8 percent (r=.28), and may have been partially offset by the negative correlation between convenience and time in city (r=-.136), and the slightly longer average time in the city for the Year Two group. The vast majority of correlations between demographic variables (where sampling differences were found), modal choice criteria, and modal perceptions were not statistically significant, and where so, involved between 3 percent to 5 percent shared variance between demographic variations and choice criteria. Thus, for practical purposes, one could take either 1974 or 1975 data as representative of the average resident's criteria in modal choice. (The exception would be that non-response bias affects both groups and probably overstates the receptivity to public transit and its funding, but this effect is probably constant throughout the time period.)

TABLE 4

1975 ADULTS CAR VERSUS BUS

TABLE 5

DISCRIMINANT ANALYSIS ON FINANCING ATTITUDES 1974 VERSUS 1975

TABLE 6

DISCRIMINANT ANALYSIS ON DEMOGRAPHIC PROFILES 1974 VERSUS 1975

Inspecting the correlations between these four demographics and financing attitudes also showed some slight correlations, although again most were not significant. For example, the correlation between the willingness to pay one or two cents per gallon from the gasoline tax as a public transit subsidy (which both years' data indicate is the preferred subsidy method, if there is going to be one) correlated .038 with sex, .043 with age, - .108 with education, and .147 with time in the city. With such small correlations, little impact on the mean attitude towards this method can be attributed to demographic fluctuations in the samples. However, the less educated and longer-in-city 1975 group might have raised the mean slightly (indicating slightly less favorable attitude than before). Similarly, the slightly stronger correlations between these demographic variables and attitudes toward substantial use of the highway trust fund versus riders paying the entire cost, shows a slightly greater effect of more conservative demographics in 1975 influencing some of the slight shift in this direction regarding financing alternatives. However, again, the highest demographic correlate is .248, or about 6 percent shared variation between "time in city" and resistance to using the highway trust fund for public transit. The overall attitude toward this mechanism was still neutral to positive, and whatever impact these small correlations with demographic variations might have had on some mean score shifts between 1974 and 1975 was minimal. The relative preferences for financing alternatives were virtually unchanged, and modal choice criteria and perceived mode images were similarly stable.

CONCLUSIONS

In spite of fluctuations in the general perceived importances of the inflation, the economy, and "energy crisis" from Spring, 1974 to Spring, 1975 (Opinion Research Corporation, 1974; 1975), there seems to have been little movement in the basic determinant attributes of modal choice. Further, the general image gaps between private versus public transportation modes in this study area remained roughly constant over the interval. Given the rather large disparity between cars versus buses for the average respondent, who has considerable discretion, it is not surprising that significant modal switching did not occur. Our data indicated a modest increase in car-pooling activities and slightly bigger car purchases. Given the current environmental conditions, and perceived benefits of private versus public transportation (especially in dependability, convenience, and flexibility) for this survey area, relatively major changes will be needed either in the perceived attributes of public transportation (or perhaps over time in the relative determinance of attributes in which public transportation is already seen as superior) for any amount of modal switching to occur. At the margins some potential switchers may be converted by treating them as a distinct market segment. The general procedure would involve those persons now using private transportation who indicate a high likelihood that they would switch to an unproved public transportation system. An approach describing a procedure for the identification of potential switches and their determinants of modal choice, plus appropriate promotional strategies aimed through media to which these persons were shown to be differentially exposed, is discussed in Alpert and Davies (1975).

Aside from this relatively small group as likely switchers, the average respondent is still rather far from altering his/her life-style to the extent that conversion to public transportation implies at present. Implementing changes in the attributes shown as determinants of modal choice in both years is more likely to bring about shifts in travel patterns and modal choices. Many of these changes may be expensive, but our data suggest a public willingness for tapping the highway trust fund, particularly if increased utility in public transportation is produced, in terms of attributes sought by the travelers (and to some extent lacking in private transportation).

If further research indicates the same kind of stability in modal choice criteria as was shown here, improvements in the system characteristics will become even more important for generating behavioral changes. If public desires are stable, the system must become more responsive; the data presented here indicate that criteria are not changing in any significant way so far. This simplifies the planning process, although it may also suggest promotional campaigns to alter criteria. At present, the available public transportation system is not seen as sufficiently competitive to private alternatives. In this study area, the gaps were not closed from 1974 to 1975. Additional research in other areas should be done to check for generality. In addition, it is possible that trends may be observed in further research that can monitor the public's modal choice criteria and their evaluation of alternative modes' abilities to meet them, as both programmed and unplanned changes occur in the relevant transportation system and its environment.

REFERENCES

Mark I. Alpert, "Identification of Determinant Attributes: A Comparison of Methods," Journal of Marketing Research, 8 (May 1971), 184-91.

Mark I. Alpert and Shane Davies, "A Decision Sciences Approach to the Marketing of Public Transportation," presented at AIDS Sixth Annual Meetings (November, 1974).

Robert C. Blattberg and Samuel R. Stivers "A Statistical Evaluation of Transit Promotion," Journal of Marketing Research, 7 (August 1970), 293-99.

"A New Federal Stab at Aiding Mass Transit," Business Week, January 16, 1974, 52-53.

Thomas Golob, Ricardo Dobson and Jagdish N. Sheth, "Perceived Attribute Importance in Public and Private Transportation," American Institute for Decision Sciences Proceedings, 5 (November 1973), 7-10.

Phillip Kotler, Marketing Management, 3rd ed. (Englewood Cliffs, NJ: Prentice-Hall, 1976).

D. M. Hill and H. G. Von Cube "Development of a Model for Forecasting Travel Mode Choice in Urban Areas, "Highway Research Record 38 (1963), 78-96.

Stanley J. Hille, F. T. Paine, A. N. Nash and G. A. Brunner "Consumer Transportation Attitude in Baltimore and Philadelphia," Transportation Journal, (Summer, 1968), 30-47.

Stanley J. Hille, and T. K. Martin "Consumer Preferences in Transportation," Highway Research Board Record (1967), 6-43.

R. A. Mundy, D. W. Cravens and R. B. Woodruff "Potential for Marketing Management Applications in Public Transportation Planning," American Marketing Association Proceedings (1974), 274-279.

James H. Myers and Mark I. Alpert "Determinant Buying Attitudes: Meaning and Measurement," Journal of Marketing 32 (October, 1968), 13-20.

James H. Myers and W. Gregory Warner, "Semantic Properties of Selected Evaluation Adjectives," Journal of Marketing Research, 4 (November, 1968), 426-428.

Opinion Research Corporation (Gallop Poll), 1974, 1975.

Jagdish N. Sheth, "A Psychological Model of Travel Mode Selection," Association for Consumer Research Proceedings, (1975), 425-30.

Donald J. Veldman, Fortran Programming for the Behavioral Sciences, New York: Holt, Rinehart and Winston, (1967).

----------------------------------------