Transportation Mode Choice in the Internet Classroom
Stella Rose Foster
Dr. Lee Gremillion
Researchers have often attempted to quantify transportation mode choice for different characteristics like age, urban density or gender. We were interested in the characteristics a specific population of Internet-education consumers would display, since that specific population seems as yet not to have been described in the literature. We reviewed other studies, composed and tested various possible questions on qualitative and quantitative polling instruments, deployed an ex-post, pre-experimental study and tabulated results. While the sample size was too small to derive inference to the wider population or universe, the survey was successful in indicating where a more controlled, random sampling might probe for potential hypotheses about correlations between demographic characteristics and transportation mode choice.
Our research was modeled on Gatersleben and Uzzell (2007), an "exploratory" study based on "self-reports by commuters" which attempted to identify affective perceptions regarding transit mode choice, to identify characteristics for future exploration. Researchers administered a questionnaire with 11 sections, taking an average 45 minutes to complete, using 5-point Likert agreement scales to measure perception affect regarding numerous areas of travel mode choice. Researchers administered 389 surveys to professors and staff at the University of Surrey, and give descriptive statistics of the sample who returned them. They report the choice modes as well as some corresponding and potentially correlated factors like distance from home to work, whether they found their commute "boring," "depressing" or "relaxing" (422) or not. They then examined variance between various data items using Chi-squared, T- and F-test hypothesis testing, and regression analysis and reported results for various co-occuring determinant and dependent variables, which they arranged in discriminant analysis plots. They condition their results by disclaiming the method could not generalize to the universe because the sampling was not random but administrative; nonetheless they found several useful areas for further research, and the result that some commuters may enjoy using other modes more than the single-occupant car.
Zacharias (2002) performed a more utilitarian survey in Shanghai attempting to describe actual patterns of modal use in two samples before and after treatment by lowering fares for some buses and changing traffic patterns in order to encourage bicyclists to use transit (313). Zacharias (2002) surveyed two different populations encountered on the street "randomly" (314), although this seems more like administrative sampling in a limited area than true random sampling, but the author never describes the method again before giving results that bicycle and car use diminished over the test area but those commuters were still reluctant to use buses (Zacharias, 2002, p. 309), and then drawing recommendations for future policy using those results. Ibrahim (2005) performed a similar study in Singapore, restricted specifically to shopping transportation mode consumption.
The literature review provoked several research questions but the question that interested us was slightly different, to identify attitudes about alternative transportation. Nonetheless the reviewed literature supported using an ex post facto, "pre-experimental" (Black, 1999, p. 70) correlational study as per Black (1999), p. 64, "asking about the nature of relationship and whether it exists, with no pretense of establishing causality," with the intent to identify factors we could then hypothesis-test for correlation with a more refined instrument. Like Zacharias (2002), we recognized that this research would measure subject perceptions of various alternatives if they did not actually use those (Black 1999, p. 36), but our objective was to rule out extraneous factors in order to achieve internal validity (Black 1999, p. 57) such that future treatment would deliver robust inference where we could claim the change in the independent variable (X) explained the change in the dependent (Y) variable. This study corresponded to Black's design "E1: One group observed" (1999, p. 70), administered to a small group using a quantitative survey instrument followed by qualitative interviews. Since the purpose was to "determine the nature of a relationship and not whether there is a difference" before or after treatment (Black 1999, p. 64), there would be no control or between-groups comparisons, so no null and alternative hypotheses were tested but rather pre-experimental questions attempting to identify correlational relationships between variables.
Development of instruments
Research required at least eight questions returning qualitative data and two demographic data questions on the quantitative instrument, to identify attitudes about alternative transportation which we would then triangulate with qualitative, open-ended interview questions. This author assembled a list of questions several students suggested for both sections, which was returned to the other students for review. We questioned qualitative item 5, "What keeps you from purchasing an alternative transportation," as potentially leading, but kept the item because this was the core question we were trying to identify, if the subject had purchased alternative transportation, they could simply indicate such, and so phrasing it as "If you have not bought alternative transportation,..." seemed redundant for that reason, so we left it as it was. In fact two subjects responded exactly this way. We did not create score cards to rank survey appropriateness but took lack of objection by the student review team as consent the instruments were adequate.
We discussed content validity and considered several different options for the individual words "alternative," "diesel" vs. "bio-diesel," or the definition of the automobile as the "traditional" mode with the other options being considered alternatives to that dominant mode choice. While time constraints did not allow us to make scoring instruments or field-test specific content items for different geographic dialect; different age interpretations of 'alternative' for example, or education levels, we arrived at the final choices through a diversity of these characteristics that we hoped would represent the usage and understanding of the typical sample subject. All reviewers felt face validity was adequate because the survey seemed readable, clear and not prohibitively complex.
Several questions were repeated in both the qualitative and quantitative questionnaires. Demographic questions of gender and residential density attempted to probe for correlation to identify if males or females were more likely to use alternative modes of transportation, and whether urban travelers were more likely to employ alternatives to the traditional automotive option. Likewise, asking what reasons people use transportation for was intended to identify whether alternative modes were employed for different uses than the car, on both surveys. While age could also have been a controlling factor, if younger drivers were more likely to use alternatives, we only found it necessary to ask that question on the quantitative instrument rather than both, so as to use the space that question would have taken on the qualitative survey asking if they thought the government would ever mandate such options.
Administration of instruments
We sent the survey out by email to twenty other class participants, with the expectation that this would return a large enough sample to make inferences to a wider population. Since the Internet classroom lacks the physical constraints of location, these consumers should represent a wide diversity of age, gender and geographic location. This diversity, while not constituting a truly random sample of the total universe because the consumers self-selected their participation in the class, would however represent the group of Internet users who are pursuing higher education, which is probably a meaningful group in absolute size and as share of the total population of transportation consumers.
Unfortunately not as many of the surveys were returned as we had hoped in the time we had to complete the experiment, so we did not obtain enough quantitative data from which to draw any inferential or even useful descriptive correlations. Four students returned the qualitative interviews, and those results are summarized in Table 1. We found the beginnings of correlation between cost and mode choice, with all respondents indicating the perceived expense of alternatives to gasoline powered vehicles as a barrier to such mode choice. There were no correlation between residential density or gender and mode choice as we expected. More results may skew this result differently. All the respondents indicated they used gas transportation to get to work, which may interest transportation providers seeking to target alternatives to the gas powered automobile.
2/3 "no," with reasons; 1 n/a, 1 declined to answer
2/4 rural; 1/4 urban, 1/4 suburban
3/4 female; 1/4 male
1. Do you live in an urban or suburban area?
We constructed two ex-post, pre-experimental survey instruments aiming to triangulate and cross-check "attitudes to alternative transportation." We employed a number of different question formats and overlapping content areas in order to test for correlations between age, gender, and urban density, among other attributes. The results yielded a number of potential areas for further study, with sample hypotheses for testing including for example, "The likelihood of alternative transportation consumption is…