Date of Award
Doctor of Philosophy (PhD)
Asad J. Khattak
R. Michael Robinson
Transportation planners have long recognized that it is urgent to integrate emerging spatial analysis with travel behavior studies. A clearer understanding of the spatial interactions among travelers and the complex environment they face has the potential to reap benefits of the ongoing technologies of travel behavior, spatial analysis and Advanced Traveler Information Systems (ATIS).
Considering that spatial patterns have been overlooked in the literature of travel behavior and ATIS, the main objective of this research is to use robust methods of spatial analysis to enhance the understanding of how the associations between traveler decisions, built environment and socio-demographic characteristics are organized spatially. This dissertation takes a significant step towards filling this gap by using innovative spatial data description methods, e.g. geo-imputation, dynamic buffer analysis, spatial statistics to model the travel behavior of both the general population and university students.
This study starts by developing a unique database from extensive behavioral data combined with a variety of spatial measurements, taking advantage ofincreased GIS capabilities. Five different activity-based databases from different regions are used, combined with their related socio-demographic and land use data. Among them are two general population travel surveys from North Carolina, which were conducted in Charlotte and at the Greater Triangle in 2003 and 2006, respectively. The Virginia Add-on for the general population was conducted in 2008, while two waves of the Virginia University Student Travel Survey (USTS) were conducted in 2009 and 2010. The general population and the university students are compared with each other in terms of how they traveled and responded to ATIS.
Issues addressed in this dissertation include two aspects. The first one is how to describe data in space more accurately. When there is a need to know the exact locations of residences (geo-coordinate), but such information is unknown, geo-imputation is used as a fundamental method of assigning synthetic locations randomly to these residences based on available zonal information. After locating the residences by using geo-imputation, dynamic buffer analysis is used to capture locally built environment characteristics around residences, which place emphasis on capturing accessibility.
The second issue is modeling travel behavior in space. Particular emphasis is placed on modeling associations between trip making, trip decision changes and their associated explanatory variables. The general population is compared with the university students who represent an energetic and technology-savvy subgroup of the population. Different spatial scales are used for these two groups: the regional level is used for the general population; the university campus is used as a special trip generator for the university students.
At the regional level, a unique model structure, i.e. Geographically Weighted Regression (GWR), is used to allow associations to change across space, referred to as spatial heterogeneity. Significant spatial heterogeneity is found in the associations between trip-making and built environment, as well as in the model oftravelers' information acquisition behavior and their travel decision adjustments. The spatial heterogeneity in the trip-making models suggests that there is higher spatial variability in favor of the statement that better land use design can help reduce auto trips. It is important to note that these potentially useful insights would have remained uncovered if using a non-spatial model that does not take spatial heterogeneity into account.
At the special trip generator level, when local models don't work well, the university campus is studied as a case which represents a combination of livable environments and a group of people who have different life cycles compared with the general population. Particular spatial analysis is applied to capture the association between trip-making and students' residential proximity to campus. The models confirm there are rings of mobility around the campus. Different from the traditional travel demand model for the general population, this varied level of mobility of students based on their residential proximity of campus is important and must be considered in the students' travel demand model.
"Spatial Analysis of Travel Behavior and Response to Traveler Information"
(2012). Doctor of Philosophy (PhD), Dissertation, Civil/Environmental Engineering, Old Dominion University, DOI: 10.25777/dnv6-st37