Incomplete panels: A comparative study of alternative estimators for the unbalanced one-way error component regression model

Abstract

This paper considers a one-way error component regression model with unbalanced data and investigates by means of Monte Carlo experiments the performance of ANOVA, MLE, and MIVQUE type estimators of the variance components. Some of the basic findings are the following: (1) For the regression coefficients, the computationally simple ANOVA methods perform reasonably well when compared with the computationally involved MLE and MIVQUE methods. (2) MLE and MIVQUE perform better than the ANOVA methods in the estimation of the individual variance component, especially for severely unbalanced patterns and large variance component ratio. However, for the remainder variance component, there is nothing much to choose among these methods. (3) Better estimates of the variance components do not necessarily imply better estimates of the regression coefficients. (4) Making the data balanced, by dropping observations, worsens the performance of these estimators when compared to those from the entire unbalanced data.