The asymptotic non-null distribution of theF-static for testing a partial null-hypothesis in a randomized PBIB design withm associate classes under the Neyman model

Abstract

As the final results of a series of our investigations [5]–[8], [10], [11] we present in this article a probability distribution—a non-centralF-distribution—which is asymptotically equivalent in the sense of type (M) d to the power function of theF-statistic for testing a partial null-hypothesis in the analysis of variance of a randomized PBIB design withm associate classes under the Neyman model which is a linear model taking both technical and unit errors into account. Thus this seems to be the final answer for which we have been after from the very beginning of our investigation.