Abstract
SUMMARY The role of the sampling mechanism for Bayesian inference on symmetric functions of a finite population is examined under partial design and partial label information. A new definition of partial exchangeability is proposed, together with a corresponding property of the sampling mechanism. These properties provide conditions under which certain symmetries in priors or designs can protect against lack of symmetry in others, in the sense of Scott & Smith (1973).
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have