Robustness and power of tests for a null variance ratio (Q3780252)

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 4043036
Language Label Description Also known as
default for all languages
No label defined
    English
    Robustness and power of tests for a null variance ratio
    scientific article; zbMATH DE number 4043036

      Statements

      Robustness and power of tests for a null variance ratio (English)
      0 references
      0 references
      0 references
      1988
      0 references
      null variance ratio
      0 references
      one-way random effects model
      0 references
      asymptotic power calculations
      0 references
      Wald test
      0 references
      asymptotic normality
      0 references
      invariance
      0 references
      Pitman alternative
      0 references
      unbalanced design
      0 references
      one-way ANOVA model II
      0 references
      locally optimum test
      0 references
      F-test
      0 references
      approximate robust
      0 references
      The ususal one-way ANOVA model II, \[ y_{ij}=\mu +\alpha_ i+\epsilon_{ij}\quad (i=1,...,t_ j;\quad j=1,...,n_ i) \] with random \(\alpha_ i\), \(E(\alpha_ i)=0\), \(V(\alpha_ i)=\sigma^ 2_ a\geq 0\), and normally distributed \(\epsilon_ i\) with \(E(\epsilon_ i)=0\), \(V(\epsilon_ i)=\sigma^ 2_ e>0\) is considered. Letting \(\Delta =\sigma^ 2_ a/\sigma^ 2_ e\) we have to test \(H_ 0: \Delta =0\) versus \(H_ 1: \Delta >0\). For the unbalanced case (not all \(n_ i\) equal) it is well-known that the usual F-test for \(H_ 0\) is not distributional robust. A locally optimum test under normal assumption has also the correct asymptotic level of significance for nonnormal data (asymptotic means increasing the number t of classes with bounded but unequal \(n_ i\) \((i=1,...,t)).\) Asymptotic power calculations demonstrate that his locally optimum test is more powerful than the F-test. A simulation experiment indicates that the locally optimum test is approximate robust even in small sample situations for which the F-test is not.
      0 references

      Identifiers