Go back

Researchers put ‘statistical significance’ to the sword

Image: James Lee [CC BY 2.0], via Flickr

Hundreds of researchers have backed a call to end the common practice of reporting experimental results in terms of whether they are statistically significant.

“We […] call for the entire concept of statistical significance to be abandoned,” three researchers wrote in a comment article for the journal Nature on 20 March. Their article was backed by a further 850 researchers, statisticians and other people working in fields that use statistical modelling.

Results should never be described as being significant or insignificant based purely on descriptions of how well they match the experimental hypothesis—the so-called “p values” and confidence intervals, wrote the researchers, led by zoologist Valentin Amrhein of the University of Basel in Switzerland.

This article is only available to Research Professional News subscribers. If you are a subscriber you can read the article in full on researchprofessional.com

Pivot-RP users can log in and view the article via this link.