Estimation of Variance Components in Unbalanced Nested Designs

Motohiro Yamasaki ； Yoshikazu Ojima ； Seiichi Yasui

《品質學報》 17卷6期 (2010/12) Pp. 469-482

並列摘要

To evaluate the precision of measurement method, nested designs are often used for estimating variance components. Variance components are the element of dispersion by a factor conducting dispersion of measurement result. So, it is the useful evaluation of measurement method for us to estimate variance components. In general, its estimator is derived from the sums of squares by ANOVA, and it is defined as the ordinary method in this paper. But it is possible to conduct the alternative sums of squares by ANOVA in the unbalanced design. Therefore, in two-stage nested design, that is one-way ANOVA model, the estimator for variance components by using the sums of squares that conducted overall mean as a simple mean of mean of each level is proposed. It is defined as the proposed method in this paper. Additionally, the expectations and variances of estimators are derived by using the positive semi-definite quadratic form in our study. And, the performance of estimators obtained from two methods is evaluated by their variances. As a result, the performance of the proposed method was superior to that of the ordinary in many conditions.

並列關鍵字

unbalanced nested design ； variance components ； ANOVA ； unbiased estimators

參考文獻

Bainbridge, T. R.(1965).Staggered nested designs for estimating variance components.Industrial Quality Control.22(1),12-20.

Byoung, C. J.,Andre, I. K.,Juneyoung, L.(2008).Comparison of designs for the three-fold nested random model.Journal of Applied Statistics.35(6),701-715.

Calvin, L. D.,Miller, J. D.(1961).A sampling design with incomplete dichotomy.Agronomy Journal.53(5),325-328.

Goldsmith, C. H.,Gaylor, D. W.(1970).Three stage nested designs for estimating variance components'.Technometrics.12(3),487-498.

(ISO 5725-2, 1994, Accuracy (trueness and precision) of measurement methods and results - Part2: Basic method for the determination of repeatability and reproducibility of a standard measurement method, International Organization for Standardization, Geneva, Switzerland.).