Jeffreys prior with MCMC returns marginal prob > 1

[merliseclyde]

rexpr: in test-bas-glm.R

test_that("Jeffreys & MCMC", {
 data(Pima.tr, package="MASS")
 pima_BAS <-  bas.glm(type ~ .,
                      data = Pima.tr, method = "MCMC",
                      betaprior = Jeffreys(),
                      family = binomial(),
                      modelprior = tr.beta.binomial(1, 1, 4))
 expect_equal(0, sum(pima_BAS$probne0 > 1))