Estimating the effect of land use and transportation planning on travel patterns: Three problems in controlling for residential self-selection

The common understanding of "residential self-selection" generally found in research on the effects of the built environment on travel is in error in three main ways. First, scholars have generally failed to recognize that the built environment may have different effects on travel for different households. Second, controlling for residential self-selection is not necessarily relevant to the predictive questions that controlled estimates are meant to inform. Third, in controlling for preferences and sorting, the literature has failed to account for the composition of the population and its consequences for housing demand. These problems may significantly influence the validity and usefulness of the research.