A second-order investigation of asymptotic ancillarity (Q1087257)

The paper deals with approximate ancillarity as discussed by \textit{B. Efron} and \textit{D. V. Hinkley} [Biometrika 65, 457-487 (1978; Zbl 0401.62002)]. In the multivariate i.i.d. case the second-order Edgeworth expansion of the MLE given a normalized version of the second derivative of the log-likelihood at its maximum is derived, which is a generalization of the one derived by \textit{S.-I. Amari} [ibid. 69, 1-17 (1982; Zbl 0484.62046)]. It is shown that the Fisher information lost by reducing the data to the MLE is recovered by the conditioning, and it is sketched how the loss of information relates to the deficiency as defined by LeCam. Finally, the author investigates some properties of three test statistics, proving a conjecture by Efron and Hinkley (same as above) concerning the conditional null-distribution of the LRT statistic, and establishing a kind of superiority of the observed Fisher information over the expected one as estimate of the inverse variance of the MLE.