Operationalizing land use diversity at varying geographic scales and its connection to mode choice :Evidence from Portland, Oregon

Although substantial consideration has been given to analyzing the relationship between land use diversity and travel behavior, the selection of the most suitable geographic scale for operationalizing these measures has received considerably less attention in the research. General consensus favors an examination of the complex relationship between travel behavior and such built environment measures explained at a finer spatial scale. The reasons for supporting a more disaggregate neighborhood scale include statistical advantages, such as a minimization of the modifiable areal unit problem, and more applied intentions, such as a preference for site-specific design measures seen as responsive to urban policy. Complementing this decision about how best to define the geographic extent of the built environment is the determination of which built envi-ronment measures are significantly related to travel mode choice. Of these measures, an increased diversity of land uses has often been linked with an individual's heightened likelihood for using transit, bicycling, and walking. This research advances the knowledge of which land use diversity measures best predict mode choice and explores the proper geographic scale for operationalizing these indicators. Seven diversity indexes represented at four geographic scales encompassing the origins and destinations of discretionary trips in the Portland, Oregon, metro-politan region were examined with a series of multinomial logit models. This study, which introduced several indexes previously unrecognized in transportation research, suggests common diversity measures, and the most disaggregate spatial scale may not always best represent the link between land use diversity and nonautomotive travel.