"Severity"is a relatively recent concept in statistical testing introduced by thephilosopher Deborah Mayo. Usually, in significance testing, if a test rejectsthe null hypothesis, this is a strong indication against the null hypothesis,but if a test does not reject the null hypothesis, nothing can be said aboutwhether the null hypothesis is indeed true. However, it is often possible tocompute the probability that a test that indeed did not reject the nullhypothesis would have yielded a less favourable result if the alternativehypothesis holds with specific parameters. The computation is similar to thatof a p-value or of the power of a test. Mayo arguesthat the null hypotheses is "severely confirmed" if the probabilityis high under the alternative that the test would have given a more significantresult.

The project is about computing (and simulating, if necessary, usinga statistical software) severity values for some standard tests such as the Binomial, t-and chi-squared test or some others

我会支付酬劳的