Implementation of Bayesian tests pbayes and dbayes for randomized block design in R code.

Abstract

In the experimental statistic it is necessary to determine, after performing an experiment, which treatments differ wich other. In this context, Bayesian tests pbayes and dbayes allow the identification of these differences for data with or without balancing and for valid or not variance analysis hypotheses. The implementation of these tests in the context of completely randomized designs has already been performed in code R. Due to the importance of these tests, their extension to other designs is of great relevance. The purpose of the work is to expand them to a randomized block design. The implementation of the tests in R code for the randomized block design was successful. The programming was validated with three experiments: data with valid variance analysis assumption, balanced heterogeneous data and unbalanced data. The results were satisfactory, presented higher or equivalent sensitivity to traditional tests, evidencing the importance and versatility of pbayes and dbayes tests. Keywords— Analysis of variance, Differences between averages, Experimental statistic, Heterogeneous data, Unbalanced data.