Means Model With Additional Information: Estimation and Analysis

Oscar O. Melo MARTÍNEZ[1]

Luis A. López PÉREZ1

Sandra E. Melo MARTÍNEZ[2]

§    ABSTRACT: In this paper, original theoretical developments have been carried out to estimate parameters and residual variance as functions of initial estimates, for models of cell means of N‑ways of classification with fixed effects when we have additional information into the cells. We have proven the properties of those new estimators such as unbiasedness and minimum variance. In the adjustment of these new parameters we do not require knowledge of initial information. Only for upgrading must we use the initial estimates of the parameters and the residual variance. This situation is usually present in industrial and biological experimentation. Finally, we work on the application of a factorial design in which we apply the obtained results.

§    KEYWORDS: Unbalanced designs; models of cell means; additional information; connectedness.



[1]Department of Statistics, Universidad Nacional de Colombia, Bogotá, D.C., Colombia. E-mail: oomelom@unal.edu.co / lalopezp@unal.edu.co

[2]Department of Agronomy, Universidad Nacional de Colombia, Bogotá, D.C., Colombia. E-mail: semelom@unal.edu.co