One of the basic principles of experimental design is blocking, which is an important factor in the treatment of the systematic spatial variability that can be found in the edaphic properties where agricultural experiments are conducted. Blocking has a mitigating or suppressing effect on the spatial dependence in the residuals of a model, something desirable in standard linear modeling, specifically in design models. Some computer programs yield a p value associated with the blocking effect in the analysis of variance table that in many cases has been incorrectly used to discard it, and although it may improve some properties of the analysis, it may affect the independence assumption required in several models. Therefore, the present research recommends the use of the H statistic associated with the corrected blocking efficiency to show the role of blocking in modeling with the incorporation of an additional advantage rarely considered related to the suppression or mitigation of spatial dependence. With the use of the Moran index, the spatial dependence of the residuals was studied in a simple factorial design in a completely randomized and blocking field layout, which evidenced the advantages of blocking in the mitigation or suppression of the spatial dependence despite the apparently little or no importance it seems to show in the analysis of variance table.
All Science Journal Classification (ASJC) codes
- Modeling and Simulation
- Computer Science Applications