Modeling ‘Dependence of Relevant Alternatives’ in Consumer Choice: A Synthesis from Disparate Literatures

Abstract

In this paper we are concerned with modeling consumer choice among competing products when spatial location matters to consumers as a product attribute. We review the literature on hospital choice and find many older studies using inappropriate ‘independence of irrelevant alternatives’ (IIA) models, while more recent studies exploit increasingly more sophisticated modeling which affects what we call here ‘dependence of relevant alternatives’ (DRA) formulations. Models which embody the IIA property to estimate probabilities of choice among alternatives do not allow the relative values of choice probabilities to change when new choices become available, which is unrealistic because new choices change substitution patterns among available products. Parameters from IIA models are thus not useful to assess the value to consumers of changes in available products, which has considerable importance to public policy. We show that tractable DRA models exist for situations where location matters, and hope that in explicitly comparing some IIA and DRA models, the value of a spatial approach to consumer choice problems will be highlighted and embraced more broadly in future spatial science and health policy research.