Análise do uso de testes de comparações múltiplas em estudos publicados em periódicos de alto impacto - Melhoramento genético de plantas

Abstract

The statistics applied to genetic improvement and agricultural experiments is configured as a key tool to achieve effective results, enabling the estimation of experimental error, in addition to validating the importance of analytical contrasts. With that, the objective was to identify the correct application of tests of comparison of averages in studies focused on the genetic improvement of plants deposited in high impact journals, favoring reliable results in research that uses this type of analysis. To achieve the objectives, the methodology was given by a bibliographic research of a transversal approach through publications of articles in high impact journals between the years 2016 and 2021, which used in their methodologies the test of comparison of means, being used as descriptors: statistics, genetic improvement and data interpretation. After collection, 60 articles were selected, being submitted to a first classification, according to the use or not of some type of test to compare means. Those who used this procedure were later grouped according to their application, being classified according to the criteria employed by Bertoldo et al. (2008), in: i) adequate; ii) partially adequate and iii) inadequate. With this, it can be concluded that the main aspects that directly affect the reliability of the researcher's interpretation of results are: i) assumptions that do not take into account different statistical tests, ii) preliminary knowledge about the type of factors and iii) Inconsistencies in the selection of statistical tests that may lead to incomplete and/or inappropriate conclusions.