BS2 Further Statistical Inference HT08 - Lecture 11
Saddle-point approximations and asymptotics of the MLE
This lecture discusses the more accurate asymptotics of maximum likelihood estimation. In particular it exploits the Edgeworth expansion of the density of a sum of random variables and tilting to find accurate approximations to the distribution of an MLE conditional on an ancillary statistic. Corresponding material can be found in Young and Smith (2005), sections 9.5, 9.6, and 9.8
Last updated: Wednesday, 02 April 2008 11:30. Steffen L. Lauritzen