
150 Ordered Regression Models
in a specic region of the probability distribution. When a noninformative prior is used
together with a large sample, the results from Bayesian and frequentist models tend to be
very similar.
In deriving the posterior, the product of the likelihood and the prior may be a commonly
recognizable probability distribution function. For example, if one uses a conjugate prior,
which is a prior that yields a posterior from the same distributional family, then the compu-
tation can be made quite straightforward and results are easily interpretable using sufcient
statistics (Gelman etal. 2014). However, not every empirica ...