ABSTRACT
In our analysis we have used the natural (from an analytic viewpoint) estimator for .A. By definition we are not interested in .A, so would be happy if we could use a simpler estimate for it; two immediate candidates are method of moments estimates and one step approximations to the maximum likelihood estimator. If we are in a random effects environment, we could used posterior expectations of the nuisance parameters. We are interested in the efficiency consequences of such methods.
