E-value makes statistical testing more flexible and robust

E-value

Researcher Peter Grünwald (Centrum Wiskunde & Informatica/Leiden University) has demonstrated this. Traditional p-values, developed by Neyman and Pearson, are increasingly criticised for their limited applicability and susceptibility to misinterpretation. This contributes to the replication crisis in science. E-values offer an alternative here: they enable interim adjustments to the research and prevent confusion between p-values and significance levels.

Statistical methods

This discovery opens the door to more reliable statistical methods. Grünwald previously received an ERC Advanced Grant to continue his research into e-value: an important step towards more accurate and flexible scientific analyses.