Maximum likelihood estimation of variance components

In this thesis, the Maximum Likelihood and Restricted Maximum Likelihood methods of estimating variance components are investigated for the one-way model. Expressions for the estimators and their variances are obtained, and algorithms for finding the estimates are tested by means of a Monte Carlo study. The quantitative effects of non-normality on the variability of estimates are discussed. Finally, diagnostic tests for identifying outliers and non-normality are proposed, and illustrated with data concerning soybean plant growth