Statistical properties of consideration sets

The concept of a consideration set has become a central concept in the study of consumer behavior. This paper shows that the common practice of estimating models using only the set of alternatives deemed to be in the set considered by a consumer will usually result in estimated parameters that are biased due to a sample selection effect. This effect is generic to many consideration set models and can be large in practice. To overcome this problem, models of an antecedent volition process that defines consideration will effectively need to incorporate the selection mechanism used for inclusion of choice alternatives in the consideration set.