Estimation of noncentrality parameters (Q4201518)

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 370641
Language Label Description Also known as
default for all languages
No label defined
    English
    Estimation of noncentrality parameters
    scientific article; zbMATH DE number 370641

      Statements

      Estimation of noncentrality parameters (English)
      0 references
      25 August 1993
      0 references
      noncentral \(F\)
      0 references
      Stein identity
      0 references
      noncentral chi-square
      0 references
      unbiased estimator
      0 references
      noncentrality parameter
      0 references
      quadratic loss
      0 references
      bounds
      0 references
      Monte Carlo study
      0 references
      linear regression models
      0 references
      elliptically contoured distribution
      0 references
      Stein effect
      0 references
      Since the naive improvement \(\delta^ +_ 0X=\max\{X-p,0\}\) (of the unbiased estimator \(\delta^ +_ 0(X)=X-p\) of the noncentrality parameter \(\Lambda\) of the \(\chi^ 2\)-distribution) is under the quadratic loss \((\delta-\Lambda)^ 2\) inadmissible, the authors proposed the estimator \(\delta(\varphi)=X-\varphi(X)\), which is a generalization of the earlier used estimator \(\delta^{PR}(X)=X-p+a/X\). Conditions (in fact the bounds) for \(\varphi(X)\) are given under which \(\delta(\varphi)\) dominates \(\delta_ 0(X)\), and results of a Monte Carlo study are presented (for \(\delta_ 0,\delta^ +_ 0\) and for special choices of \(\varphi)\). A similar improvement is achieved (in the framework of linear regression models) for the estimator of the value \(\beta^ TX^ TX\beta/\sigma^ 2\) under the assumption that the fluctuations in the model follow some elliptically contoured distribution (e.c.d.). Two possibilities are considered: (i) the type of the e.c.d. is known, (ii) the type is not specified. The result in the case (ii) represents a robustified modification of the result for the case (i), and the method is a version of the Stein effect.
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references