ON THE RELATIONSHIP OF INFORMATION AND (TELE)COMMUNICATION SYSTEMS WITH ACTIVITY PARTICIPATION AND TRAVEL

Open Access
Author:
Kim, Tae-Gyu
Graduate Program:
Civil Engineering
Degree:
Doctor of Philosophy
Document Type:
Dissertation
Date of Defense:
April 28, 2004
Committee Members:
  • Konstadinos G Goulias, Committee Chair
  • Martin T Pietrucha, Committee Member
  • Ageliki Elefteriadou, Committee Member
  • William L Harkness, Committee Member
Keywords:
  • ICT
  • Transportation
  • Activity and Travel Behavior
  • Panel Analysis
  • Multilevel Modeling
  • Structural Equation Modeling
  • Latent Class Cluster Analysis
Abstract:
During the last decade we have experienced the rapid advance and growing popularity of information and communications technology (ICT). It began to alter the way in which people conduct their everyday affairs and also the way in which businesses are conducted. Under these circumstances, our traditional concept of accessibility can no longer be valid. In fact, through ICT people can get virtual accessibility to a rapidly growing range of activities without the more traditional spatial and temporal limitations and constraints. Consequently, people have more flexibility to arrange their schedules, and eventually change their activity and travel patterns. These substantial impacts of ICT motivate the need for research on the present and future impacts of telecommunication on activity and travel behavior. With the Puget Sound Transportation Panel (PSTP) data, especially the data collected in Wave7 (1997) and Wave 9 (2000), this study attempts to examine a variety of aspects of the relationships between ICT and activity and travel behaviors within comprehensive conceptual model systems examining correlation patterns. Several models have been developed, each of which focuses more on a specific aspect of the relationships. First, technology choice models were developed to identify user groups for each ICT device. Multivariate multilevel categorical data models were used to account for a strong behavioral correlation among households and within household members as well as a high degree of people heterogeneity in technology adoption. The model revealed that decisions of ICT ownership and usage are most likely determined by joint decisions among members of the same household and there is heterogeneity in each type of ICT ownership and use among different situations. Second, a comparative analysis of three different model systems, including a set of single-equation regression models, seeming unrelated regression models (SUR Model), and multivariate multilevel models, was conducted to examine the effect of ICT on activity and travel duration. Although the three different models produce very similar coefficient values, multivariate models are more advantageous over single-equation models. The multivariate models account for the correlation of the error terms across equations, and as a result, they produce smaller standard errors of the coefficient estimates than those from single-equation models. Between the two multivariate models, the multilevel models are superior because they consider the hierarchy of level in the data and provide more information by estimating variance-covariance matrices and correlations for each of the multiple levels, offering additional insight on behavioral heterogeneity at each level. Third, the joint models using a structural equation modeling technique are formulated for daily time allocation to various activities (subsistence, maintenance, and leisure) and travel, and mode frequency (driving alone, shared ride, transit, bike, walk, and others) as a function of cross-sectional and longitudinal information on personal and household socio-demographics and telecommunication technology ownership and usage. In this way, the impacts of information and communication technologies on daily time allocation to various activities and travel, and on modal split are assessed. At the same time, the complex relationships among different activities and travel time use indicators and their daily frequencies are explored. In addition, using longitudinal information on ICT, it is possible to check whether or not the changes in ICT ownership and usage have symmetric effects on time use for activity and travel behavior. From this model, it was found that the ¡°technology¡± effect depends on the location and type of technology and that the majority of social, economic, and ICT changes have asymmetric effects on behavior. Fourth, dynamic analysis of time use and frequency of activity and travel between Wave 7 and Wave 9 is conducted using latent class clustering and structural equation modeling. A fixed time budget is explicitly taken in account in these models, and existence of very strong substitutional and complementary relationships in time use among different activities and travel are examined. In addition, as a new approach to account for state dependence (past activity and travel behavior effects) in the model, relatively homogenous daily activity and travel behavior patterns in Wave 7 are first identified through latent class cluster analysis. Then time use for a specific activity (in-home activity, out-of-home subsistence activity, and out-of home non-subsistence activity) and traveling, and frequency of episodes by activity in Wave 9 are modeled as a function of cross-sectional and longitudinal information as well as activity and travel patterns in Wave 7. The model results showed that out-of-home activity duration and travel time have very large substitutional effects on in-home activity duration, while out-of-home activity duration has complementary effects on travel time. The model results also confirmed the existence of strong habit persistence in activity engagement and time use even in a relatively long period of time.