Skip to main navigation Skip to search Skip to main content

Improved estimation for multiple means with heterogeneous variances

  • Edwin J. Green
  • , William E. Strawderman
  • , Ralph L. Amateis
  • , Gregory A. Reams

Research output: Contribution to journal › Article › peer-review

Abstract

Two new estimators are presented for use in situations where simultaneous estimation of more than two sample means is required and alternative, possibly biased information is available. The new estimators are modifications to an older estimator by Green and Strawderman. The latter estimator assumed homogeneous variances, whereas the new ones are designed for the more usual case of heterogeneous variances among the sample means. In simulation experiments, the new estimators yielded superior performance to that of the ordinary sample mean vector (X). Surprisingly, the estimator designed to dominate X under precision-weighted loss was apparently the best estimator, even under nonprecision-weighted loss. Use of the new estimators should allow foresters to achieve considerable savings in estimation precision.

Original languageEnglish (US)
Pages (from-to)1-6
Number of pages6
JournalForest Science
Volume51
Issue number1
StatePublished - Feb 2005

ASJC Scopus subject areas

  • Forestry
  • Ecology
  • Ecological Modeling

Keywords

  • Empirical Bayes
  • Forest inventory

Fingerprint

Dive into the research topics of 'Improved estimation for multiple means with heterogeneous variances'. Together they form a unique fingerprint.

Cite this