A Generalized Model of Activity Space

This article introduces the concept of a generalized activity space to bridge area-based and activity-based representations of geographic context. We argue that microscale space-time paths fail to account for contextual determinants of behavior, because they emphasize "contacts" over "contexts," a problem that could be solved, in part, by using a broader "generalized" representation of geographic context. This article develops the idea of a generalized activity space and empirically tests the viability of the concept. Support for the viability of the idea is identified through analysis of 34,500 trips by 7,550 individuals in Atlanta. We find that demographic characteristics and residential location jointly shape a person's geographic context. Through a series of hypothesis tests, we find evidence that these location-demographic groupings are generalizable; that is, people with similar socioeconomic backgrounds and residential locations exhibit similar generalized activity spaces. Residential location, by itself, however, is not an effective descriptor of the configuration of a person's context. We argue that generalized activity spaces have potential to inform study of how the environment influences behavior by allowing a more robust consideration of interplay between socioeconomic characteristics and the use of space.