ABSTRACT

It occasionally may happen in practical applications that the total number of censored observations is known, but the number in each tail separately is not known. In these samples, we let a and d designate the terminals and let c designate the combined number of censored observations in the two tails. The total sample size is Nand n = N - cis the number of complete observations. The likelihood function of a sample of this type from a Poisson distribution with probability function (13.2.1) is

On taking logarithms, differentiating, and equating to zero, we obtain the estimating equation

d In L = nx _ n + c [ dA A 1

We simplify this equation to obtain

and the estimate A is obtained as the solution.