P values are only an index to evidence: 20th- vs. 21st-century statistical science

摘要

OVERVIEW COMMENTS We were surprised to see a paper defending P values and significance testing at this time in history. We respectfully disagree with most of what Murtaugh (2014) states. The subject of P values and null hypothesis significance tests is an old one and criticisms by statisticians began in the late 1930s and have been relentless (see Commentaries on Significance Testing for a partial impression of the large literature on the subject [available online]).~2 Oakes (1986) summed it up over 25 years ago, "It is extraordinarily difficult to find a statistician who argues explicitly in favor of the retention of significance tests..."