© 1981 by Biometrika Trust
Efficient conditional tests for mixture experiments with applications to the birth and branching processes
Department of Mathematical Statistics, La Trobe University Bundoora, Victoria
Conditional tests, which are efficient in the sense of being uniformly most powerful unbiased for any sample size and are also asymptotically efficient with reference to a relevant likelihood in the sense of maximizing asymptotic power at alternatives converging to the null hypothesis, are derived for mixture experiments involving dependent observations. As simple applications efficient tests in birth processes, branching processes, stable additive processes, etc. are derived.
Key Words: Birth process • Branching process • Conditionality principle • Efficient test • Exponential family • Inference for stochastic process • Mixture distribution and process • Nonergodic process