Unimodality of Likelihood Functions for the Binomial Distribution
Journal of the American Statistical Association · 1983 · Volume 78 · Issue 381
DOI: 10.1080/01621459.1983.10477949
Abstract
A maximum likelihood estimate exists for both parameters of the binomial distribution if and only if the sample mean exceeds the sample variance. The derivative of the log-likelihood function has at most one finite root; consequently the likelihood function is unimodular.
Enable JavaScript for citation tools and additional paper interactions.