Showing papers by "University of Jordan published in 1969"
TL;DR: In this article, the analysis of variance tables for the randomized block and the generalized randomized block designs is shown to provide different degrees of freedom for the estimation of the error variance than does the random block design.
Abstract: require an artificial division into 6 blocks. A glance at thle structure of the Analysis of Variance tables for the randomized block and the generalized randomized block designs reveals that, when there is no block-treatmenit interaction effect, the generalized randomized blockl design provides mnore degrees of freedom for the estimation of the error variance than does the randomized block design. Furthermnore, if there is a block-treatinent interaction effect and blocks are fixed e-ffects, the randomized block design, unlike the generalized randoimized block design, does not provide a test of the hypothesis that there are no differences among the treatments. Because the artificial introduction of blocks to accommodate the requirements of the randomized block design often results in less sensitive -tests or no tests at all for de-tecting treatinent differences, I commnend thle generalized randomized block design for miiore widlespread consideration in -the planning of experiments.