Attributes and Limited Attention

We consider an attribute-based model of stochastic choice with variable attention to different attributes. We characterize stochastic choice rules with attributes and limited attention (SCRALA). Under SCRALA, the probability with which an alternative (say, x) is chosen is the product of the probabilities with which attention is drawn by the attributes where x is ranked highest and the (weighted) probability with which attention is not drawn by the attributes under which x is not the highest ranked. Our results are characterized by axioms defined on observable choice data and all the attention parameters are uniquely identified.