.

• Autor(a) principal: Bertoldo, Juliano Garcia
• Data de Publicação: 2008
• Outros Autores: Coimbra, Jefferson Luís Meirelles, Guidolin, Altamir Frederico, Mantovani, Adelar, Vale, Naine Martins do
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Biotemas (Online)
• Texto Completo: https://periodicos.ufsc.br/index.php/biotemas/article/view/2175-7925.2008v21n2p145

Diffi culties related to the use of tests of comparison of averages in scientifi carticles. The objective of this work was to verify the researcher’s main diffi culties in the use of means comparison tests, as well as to propose alterna- tives for a better statistical inference. For a correct application of the statis- tical tests it is fundamental for the researcher to know the experimental design, the factors, the variables and the treatments that will constitute the work, because these are decisive in the choice of the statistical test. Two hundred and ninety-two scientifi c papers from plant sciences were revised. These papers were published in the Brazilian Qualis A journal, from 2000 until 2006. The papers were classifi ed by their use of means comparison tests as i) appropriate, ii) partially appropriate, and iii) inappropriate. There were diffi culties involved in choosing the correct statistical procedure in relation to the experimental design used. The papers  in which more than one factor was studied (75%) were classifi ed as inadequate due to their abuse of the means comparison tests. On the other hand, 3% and 22% were  classifi ed as partially appropriate and appropriate, respectively.