Journal Article

Penalized balanced sampling

BREIDT, F. J., CHAUVET, G.

2012

Overview

Linear mixed models cover a wide range of statistical methods, which have found many uses in the estimation for complex surveys. The purpose of this work is to consider methods by which linear mixed models may be used at the design stage of a survey to incorporate available auxiliary information. This paper reviews the ideas of balanced sampling and the cube algorithm, and proposes an implementation of the latter by which penalized balanced samples can be selected. Such samples can reduce or eliminate the need for linear mixed model weight adjustments, a result demonstrated theoretically and via simulation. Horvitz-Thompson estimators for such samples will be highly efficient for any responses well approximated by a linear mixed model in the auxiliary information. In Monte Carlo experiments using nonparametric and temporal linear mixed models, the strategy of penalized balanced sampling with Horvitz-Thompson estimation dominates a variety of standard strategies.

• Publisher: Biometrika Trust, University College London,Oxford University Press,Oxford Publishing Limited (England)
• Subject: Algorithms / Applications / Approximation / Biology, psychology, social sciences / Calibration / Combinatorics / Combinatorics. Ordered structures / Cubes / Degrees of freedom / Designs and configurations / Error rates / Estimators / Exact sciences and technology / Experimental design / General topics / Linear regression / Mathematical models / Mathematics / Monte Carlo simulation / Population estimates / Probability and statistics / Random walk / Sampling techniques / Sciences and techniques of general use / Statistical methods / Statistical variance / Statistics / Studies