Statistics as Part of Scientific Reasoning in Plant Sciences: Overlooked Issues and Recommended Solutions

摘要

New research tools for modern life sciences are emerging every few years and being implemented to reveal the underlying mechanisms of scientific questions of interest.However,statistical practice in modern life sciences has barely moved forward and probably has regressed.In many cases,the decisions from biological experiments are predominantly made by using significant P values involving inappropriate statistical analyses and standards,and positive,favored results are preferentially reported (Fanelli,2012;Head et al.,2015;Demidenko,2016;Yong,2017).This practice increases the likelihood of false-positive statistically significant findings (Dumas-Mallet et al.,2017).A series of publications have raised concerns about this practice and called for the use of proper statistics in life science research (Halsey et al.,2015;Demidenko,2016;Weissgerber et al.,2016;Benjamin et al.,2018).