Consider a model with scalar parameter . Prove that, if you draw from the prior,

draw y| from the data model, then perform Bayesian inference for given y, that

there is a 50% probability that your 50% interval for contains the true value.