<div dir="ltr">Divulgação da palestra desta semana do Seminário do <a href="http://www.icmc.usp.br/Portal/conteudo/1096/13/interinstitucional-de-pos-graduacao-em-estatistica">Programa Interinstitucional de Pós-graduação em Estatística</a> (PIPGEs ICMC/USP e UFSCar), São Carlos. Palestrante: André Rogatko - Samuel Oschin Comprehensive Cancer Institute, Cedars-Sinai Medical Center. Título: Misuse of Statistical Methods and Reproducibility Problems in Science Resumo: In a gathering convened by the U.S. National Institutes of Health, Nature, and Science, a group of editors representing more than 30 major journals; representatives from funding agencies; and scientific leaders assembled at the American Association for the Advancement of Science’s headquarters in June 2014 discussed principles and guidelines for preclinical biomedical research given serious problems in reproducibility of scientific findings. The guidelines recommend that journals include in their information for authors their policies for statistical analysis and how they review the statistical accuracy of work under consideration. Unfortunately, misuse of statistical methods in data modeling and inferences is frequent and may account for a significant portion of irreproducible results in all areas of science. This paper discusses three cases in inference and data analyses where misuse of statistical methods is highly prevalent: goodness-of-fit test for Hardy-Weinberg equilibrium, dose finding in cancer, and analysis of proteomics data. In addition, a Bayesian test of multiple hypotheses is discussed as a possible solution to effectively control decision errors and increase reproducibility. Local: Sala de Seminários do Departamento de Estatística da UFSCar Data/Horário: Segunda-feira 22/06/2015; 16hs. -- Rafael Izbicki Assistant Professor Department of Statistics Federal University of São Carlos (UFSCar) <a href="http://www.rizbicki.ufscar.br/">http://www.rizbicki.ufscar.br/</a> </div>