Efficient tests for branching processes
分枝过程的有效检验
This paper is concerned with efficiency in testing for the mean of the offspring distribution of a supercritical Galton-Watson process. It is suggested that the maximum likelihood estimator is generally inefficient for this problem in the sense that asymptotically it is not an optimum discriminator between alternative values of the parameter. Three competing tests are discussed and their efficiency compared via an example.