Résumé : We introduce a behavioural condition, called choice dominance, which (partially) ranks individuals based on their consumption of a certain good. The notion is equivalent to a single crossing restriction on the indifference curves of the individuals. We provide a revealed preference condition, called X-GARP, to test for choice dominance and we incorporate the notion into a household framework to obtain testable restrictions on the change in household demand after an increase in the reservation utility for one of the household members. We apply our fundings to an experimental dataset and a budget survey from a conditional cash transfer programme.