Transit-oriented developments with the goal of reducing dependency on the automobile have been observed nationwide. The implicit hypothesis behind these developments is that where one lives would affect his or her consequent activity and travel behavior. Many questions remain despite decades of research in residential location choice and activity and travel behavior. In residential location choice literature, previous studies have modeled residential location choices as static ones, with no memory of the past. In addition, few studies have modeled the spatial correlation between alternative locations, although they may exist in reality. In activity and travel behavior literature, most studies implicitly assume that behavioral changes are instantaneous, despite the recognition that a behavioral change in response to an external trigger may take time to occur, due to various types of constraints one may face. This dissertation will introduce a residential relocation model that addresses spatial correlation and accounts for the influence of the prior residential location on one’s current preference. Furthermore, I will examine how a change in the built environment (spurred by a change in residence) can affect the response lag generated by a significant change in the time people allocate to discretionary activities and travel.

I developed a generalized extreme value model for residential location choice to account for spatial correlation. I considered the influence of prior location in the model by assuming that the preferences toward various attributes in the current location choice (represented by coefficients in the utility function) are functions of the characteristics of the current location and the past location. Furthermore, the model also considers lifecycle interaction with the prior location influence. To study how a change in the residential location may trigger activity and travel behavioral changes, I constructed a survival model to investigate the response lag in changes to the time allocated to discretionary activities and travel. I addressed three issues in the survival model: data censoring, partial observations, and multi-event.

The empirical results confirmed that past housing experiences influence current location choices. In particular, I found that people have become more tolerant to long commute distances and place more value on attributes like retail, open space, and business opportunities after previous heavy exposure. I also found that lifecycle could interact with the influence of past housing experience. In addition, spatial correlation is confirmed. The results from the response lag study suggested that temporal constraints, as well as family and social obligations were significant contributing factors to the length of the response lag, while the built environment was not.

This dissertation contributes to the existing literature by recognizing the dynamics inherent in human behavior, developing methodologies to model real-world phenomena, uncovering underlying behavioral mechanisms, and lending insights into how these factors can influence policy decisions. In particular, it contributes to the residential location choice literature by empirically validating the existence of past housing experience in current location choices and demonstrating the interactive effect between past housing experience and lifecycle. In travel behavior analysis, this study is one of the first, if not the first, to investigate the response lag of a significant change in time use in response to a residential relocation. Methodologically, the dissertation contributes to the literature by developing a generalized extreme value model to account for spatial correlation in location choices and a survival model handling left-censoring, partial observation, and multi-event in the response lag study.