BS2 Further Statistical Inference HT08 - Lecture 4
More on nuisance parameters
This lecture discusses more complex examples involving nuisance parameters, where there is no clear ancillarity or partial sufficiency present, including the most common case of a sample from a normal distribution with unknown mean and variance.
We consider in particular an example where there are many nuisance parameters and show how straight likelihood methods will lead us astray.
Various modifications of likelihood function such as marginal, conditional, profile, and integrated likelihood is described and discussed.
Recommended reading for this lecture is Section 3.3 (page 54-60) of
Garthwaite, Joliffe and Jones (2002). Statistical Inference. 2nd edition, Oxford University Press,
as well as Sections 8.6 and 9.3 of Young and Smith (2005).
Last updated: Friday, 01 February 2008 11:49. Steffen L. Lauritzen