On the power-divergence statistic in sparse multinomial models requiring parameter estimation
It will be shown that the power-divergence family of goodness-of-fit statistics for completely specified parameters and nuisance parameter, under the sparseness assumption, have the same asymptotic normal distribution under a sequence of local alternatives. Hence, these tests have such low efficiency that they can not distinguish between the known and unknown parameters. Although it seems unlikely that the test statistics with and without estimated nuisance parameters have the same asymptotic behavior, there is a simple intuitive explanation for this phenomenon. The case in which the number of nuisance parameters tends to infinity is addressed.
Year of publication: |
1992
|
---|---|
Authors: | Tiwari, Ram C. ; Wells, Martin T. |
Published in: |
Statistics & Probability Letters. - Elsevier, ISSN 0167-7152. - Vol. 13.1992, 1, p. 57-60
|
Publisher: |
Elsevier |
Keywords: | Nuisance parameter power-divergence family sparseness assumption |
Saved in:
Online Resource
Saved in favorites
Similar items by person
-
Large sample theory of spacings statistics for tests of fit for the composite hypothesis
Wells, Martin T., (1993)
-
Li, Gang, (1996)
-
Inference for Shift Functions in the Two-Sample Problem With Right-Censored Data: With Applications
Lu, Henry H.S., (1994)
- More ...