For some distributions commonly used in reliability analysis the likelihood function has a singularity at the smallest observation. The use of grouped likelihood in Bayesian inference is discussed. It is shown that the product of spacings can be used in place of the likelihood. Similarities with and differences from a standard Bayesian approach are illustrated by means of examples. In particular some computational advantages are pointed out.
|Number of pages||6|
|Journal||Reliability Engineering and System Safety|
|Publication status||Published - 1994|