Is accessibility a control variable? Is it to be controlled for?

Over the past several decades, academics, practitioners, and policymakers have largely favored using the concept of transportation accessibility. The most common basis of the concept is Hansen's 1959 definition, the "potential of opportunities for interaction." Since then, most studies have comprehensively used this accessibility concept, along with other similar notions. This study began suspecting that, despite the concept's predominance in transportation planning, its evaluation and application have been conducted erroneously. In their regression-type analyses, many researchers have used accessibility as an independent variable based on the assumption that accessibility is controllable. The assumption originates from studies that accessibility shapes land use. As their theoretical ground, Hansen's study has been often referred to and by revisiting the study, we attempted to differently interpret his accessibility equation and to extend its application, and accordingly, we identified three possibilities. The first possibility is that land use can shape accessibility, not only that accessibility shapes land use. Second, while most previous studies used accessibility as an independent variable, such an application may not be generalizable. Last, according to the former two possibilities, the land use–transportation accessibility relationship may be reciprocal, so the potential for the endogeneity bias—specifically, simultaneity—needs to be considered. [ABSTRACT FROM AUTHOR]