Causality and self-selection

In academic literature, residential self-selection is often pointed out as a source of error casting doubt about the existence or magnitude of influences of built environment characteristics on travel behavior. However, the causal mechanisms through which built environments influence travel exist regardless of any residential self-selection. The very phenomenon of travel-based residential self-selection presupposes the existence of such causal influences. Moreover, choice of residential location is influenced by a wide range of criteria, and often preferences other than those rooted in travel attitudes play the dominant role. Studies in three Nordic countries and China show only small differences in effect estimates between models with or without residential preference variables. There is also increasing evidence that residential preferences do not only influence residential location but are themselves influenced by habits and experiences from living at a specific location. Travel attitudes will then be irrelevant control variables increasing, instead of reducing, bias.