A General Behavioural Model of Random Choice

We develop and study a stochastic eliminative choice procedure, called choice by random checklist (CRC). On the one hand CRC extends the idea of Tversky's Elimination by Aspects (EBA) to explain, for example, stochastically intransitive choices. On the other hand a special case of CRC is equivalent to random utility maximisation (RUM). In this way we bring within a single formal framework two leading random choice models in economics and in psychology, which are otherwise difficult to relate. Unlike EBA, which is defined recursively, we give an explicit formula for the choice probabilites in a CRC. We show that a CRC satisfies Monotonicity (extending a result of Sattath and Tversky [14]). We define new testable properties, Stochastic WARP and the Probabilistic Inequality, that are satisfied by CRC (and so, in particular, by a RUM).