BS2 Statistical Inference - Lecture 13
The maximized likelihood ratio test
This lecture is concerned with large sample tests in the multi-parameter case, including the case of a composite hypothesis. We define the MLRT and sketch the proof in the case of a simple hypothesis that the MLRT has an approximate chi-square distribution with degrees of freedom calculated as the difference in dimension between model and hypothesis.
We also indicate how other large sample tests such as Wald's test and the chi-square test can be seen as approximations to the MLRT based on using Taylor's theorem.
Overheads
Links
Last updated: Friday, 26 November 2004 10:41. Steffen L. Lauritzen