Talk:Most probable number

From WikiProjectMed
Jump to navigation Jump to search

This article says absolutely nothing specific about this method. You can't even guess what this method is after reading this. Michael Hardy (talk) 20:40, 9 September 2008 (UTC)[reply]

A guess is that the underlying model is a Poisson distribution, but the observations are either 0 for no events or 1 for one or more events and that "MPN" might actually refer to a maximum likelihood estimate. This would mean a binomial distribution with parameters related to those of the Poisson process. Unfortunately the reference now included is unlikely to be immediately to hand for most statisticians. Melcombe (talk) 08:54, 10 September 2008 (UTC)[reply]