Maximum Likelihood Approaches to Variance Component Estimation and to Related Problems

Abstract

Several recent developments promise to increase greatly the popularity of maximum likelihood (ML) as a technique for estimating variance components. Patterson and Thompson (Biometrika, Vol. 58, December 1971, pp. 545-554) proposed a restricted maximum likelihood (REML) approach which takes into account the loss in degrees of freedom resulting from estimating fixed effects. Miller (Technical Report No. 12, Department of Statistics, Stanford University, 1973) developed a realistic asymptotic theory for ML estimators of variance components. There are many iterative algorithms that can be considered for computing ML or REML estimates of variance components. Some were developed specifically for the variance component problem and related problems. Others are general nonlinear optimization procedures. The computations on each iteration of these algorithms are those associated with computing estimates of fixed and random effects for given values of the variance components. MINQUE's of variance components can be computed from one iteration of the REML version of Anderson's (Annals of Statistics, Vol. 1, January 1973, pp. 135-141) iterative procedures.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jun 01, 1975
Accession Number
ADA014244

Entities

People

  • David A. Harville

Organizations

  • Air Force Research Laboratory

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Algorithms
  • Applied Mathematics
  • Data Science
  • Estimators
  • Factor Analysis
  • Information Science
  • Mathematical Filters
  • Mathematics
  • New York
  • Random Variables
  • Statistical Algorithms
  • Statistical Analysis
  • Statistical Inference
  • Stochastic Processes
  • Surveys
  • Theorems
  • United States

Fields of Study

  • Mathematics

Readers

  • Statistical inference.