This dissertation investigates the factors that influence an individual's residential choice. The role that residential choice plays in other individual decisions is also investigated, with an emphasis placed on understanding the importance of land use configuration on individual travel demand. To achieve this, a conceptual model of residential choice/preference was developed that was a comprehensive reflection of those relationships supported by the literature and by informed judgment. The complexity of this model is seen in the many interdependent relationships that involve residential choice. For example, in choosing a place to live, a household may evaluate a dwelling unit and/or neighborhood according to how it fits along several interrelated dimensions, such as: housing type, neighborhood type, distance to work, distance to shopping and other household-related activities, type of mode to work and car ownership.
The measurement of neighborhood type as a continuous variable through factor analysis was another important part of this work. A two-factor disaggregate solution representing traditional and suburban neighborhood dimensions was used in this study. This approach allowed for a single area to possess attributes of both types of neighborhoods, and allowed individuals within the same area to face different neighborhood characteristics - a flexibility amply justified by the empirical results. Further, it has the statistical advantage of producing continuous measures of endogenous variables, a trait that is desirable in both regression and structural equation models.
The data analyzed in this study came from 852 individuals from five neighborhoods in the San Francisco Bay Area. Information such as trip records, life style preferences, and attitudes towards urban transportation, housing and the environment, were incorporated with household demographic and socio-economic data to perform multivariate statistical analyses of an individual's residential choice. Specifically, three different sets of models were estimated: 1) a binary model of residential choice (adjusted ? = 0.52), where residential choice alternatives include suburb (=1) and traditional (=0), 2) single-equation regression models of elements in the conceptual model (with adjusted R2 values ranging from 0.39 for residential choice = traditional to 0.02 for travel demand = daily walk/bike miles rate, and 3) structural equation models (with good model fit indices such as the relative fit index = 0.913)
A particularly noteworthy finding supported by all of the models referred to above is that attitude and lifestyle variables play the greatest role in explaining residential choice and travel demand. Results suggest that the association commonly observed between neighborhood type and travel patterns is not one of the direct causality, but due to correlations of each of those variables with others. In particular, is believed that when attitudinal, lifestyle, and sociodemographic variables are accounted for, neighborhood type has very little influence on travel behavior.
