On Non-parametric Testing, the Uniform Behaviour of the t-test, and Related Problems (Q4677113)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 2169673
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | On Non-parametric Testing, the Uniform Behaviour of the t-test, and Related Problems |
scientific article; zbMATH DE number 2169673 |
Statements
On Non-parametric Testing, the Uniform Behaviour of the t-test, and Related Problems (English)
0 references
20 May 2005
0 references
asymptotic maximum
0 references
confidence intervals
0 references
large sample theory
0 references
0.7850606441497803
0 references
0.7770794630050659
0 references
0.7766817808151245
0 references
The author deals with some problems in nonparametric hypothesis testing. First, he revisits a result of Bahadur and Savage, who proved that it is impossible to construct valid tests of the value of the mean \(\mu(F)\) of a distribution \(F\) on the real line, if the model of possible distributions \(F\) is suitably large. This result is extended to other testing problems, such as testing whether or not the mean is rational, testing goodness-of-fit, and equivalence testing. Next, the author shows that the size of the \(t\)-test is one even if we restrict attention to distributions supported on a compact set. In addition, the size is bounded away from 1 for the family of symmetric distributions, but still the convergence to the level is not uniform. By imposing a weak condition on the family, the uniform convergence of the rejection probability under \(F\) to the nominal level is obtained, where uniformity is with respect to distributions \(F\) in a large nonparametric family. The author also shows that the \(t\)-test possesses a local asymptotic maximum property over a fairly large collection of distributions.
0 references