Mixture prior distributions are much used in statistical applications, such as clinical trials, especially to avoid prior-data conflicts. We explicitly prove that the effective sample size (ESS) of a mixture prior rarely exceeds the ESS of any individual mixture component density of the prior.
