Revealed Stochastic Choice with Attributes

Many theoretical models of stochastic choice are characterized by availability variation. Instead, most stochastic choice datasets have information on attribute values that vary across decision problems. This paper uses attribute variation to characterize a framework that encompasses existing interpretations of stochastic choice including context dependence, nested choice behavior, and consideration sets. The model has utility indices that depend on attribute values, and is characterized by a monotonicity condition relating probabilities and utility indices. Linear utility indices can be estimated for the model using existing methods without taking a stand on a particular reason why choice is stochastic