On a conservative Bayesian method of inferring extinction