Phys.org

Smarter hypothesis testing with statistics: How e-values can improve scientific research

During his Ph.D. research, mathematician Tyron Lardy worked on a new approach to hypothesis testing. Instead of the traditional p-value, he uses so-called e-values. These turn out to be more ...

A/B Testing: Why a 0.002 difference between p-values of 0.049 and 0.051 separates 'significant' from 'insignificant.' There is no scientific basis for that line.

You conducted an A/B test. The difference in conversion rate came out with a p-value of 0.049. It is 'significant.' The decision is finalized as 'adopt Plan B.'The following week, you ran another A/B ...
newswise

Time to abandon null hypothesis significance testing? Moving beyond the default approach to statistical analysis and reporting

Newswise — Researchers from Northwestern University, University of Pennsylvania, and University of Colorado published a new Journal of Marketing study that proposes abandoning null hypothesis ...