Paper 1, Section II, J

Principles of Statistics
Part II, 2010

The distribution of a random variable XX is obtained from the binomial distribution B(n;Π)\mathcal{B}(n ; \Pi) by conditioning on X>0X>0; here Π(0,1)\Pi \in(0,1) is an unknown probability parameter and nn is known. Show that the distributions of XX form an exponential family and identify the natural sufficient statistic TT, natural parameter Φ\Phi, and cumulant function k(ϕ)k(\phi). Using general properties of the cumulant function, compute the mean and variance of XX when Π=π\Pi=\pi. Write down an equation for the maximum likelihood estimate Π^\widehat{\Pi}of Π\Pi and explain why, when Π=π\Pi=\pi, the distribution of Π^\widehat{\Pi}is approximately normal N(π,π(1π)/n)\mathcal{N}(\pi, \pi(1-\pi) / n) for large nn.

Suppose we observe X=1X=1. It is suggested that, since the condition X>0X>0 is then automatically satisfied, general principles of inference require that the inference to be drawn should be the same as if the distribution of XX had been B(n;Π)\mathcal{B}(n ; \Pi) and we had observed X=1X=1. Comment briefly on this suggestion.