Inference about the Number of Resident Bacterial Strains when Sampling from Plate Colonies

摘要

This paper deals with making inferences about the number of resident bacterial strains in mice gut based on the number of strains of bacteria found in a random sample of colonies derived from plated faecal material. It is shown that for what seems to be a natural way to test the hypothesis that there is only one resident bacterial strain against the alternative that there is more than one resident strain the probability of a Type II error may be unacceptably large. It is also shown that given the sample contains only one bacterial strain the probability that the animal has only one resident strain converges to unity at a geometric rate as the sample size increases.