Deprecating Statistical Significance: Toward Better Science




April 26, 2019


On Friday April 26, 2019 I gave an invited talk to the Department of Speech-Language-Hearing Sciences in their Pro-Sem series. These are really neat as they are organized by the graduate students. This is the third one of these I have given over the years and each is a treat.

In this talk I spoke about the ASA’s recent call for the deprecation of “statistical significance” and all its related variants (e.g., “significantly different,” “p < 0.05”, and “nonsignificant”, asterisks in a table). Citing articles in the recent volume of The American Statistician and drawing on the historical conflagration that is NHST I laid out why scientists and statisticians have an issue with “statistical significance” and also proposed some ideas for moving forward based on those proposed by Ronald L. Wasserstein, Allen L. Schirm & Nicole A. Lazar.

The slides for the talk are available here.