Type S error for classical and Bayesian single and multiple comparison procedures (Q1424597): Difference between revisions

The authors investigate the conception of Type S (for sign) errors in comparison procedures which appear when one claims ``\(A_1>A_2\) with confidence'' when \(A_2>A_1\) (or vice versa). Type S error rates for classical and Bayesian confidence statements are defined, calculated and compared. The authors outline that the traditional Type 1 error calculations are not suitable for understanding a null hypothesis both in the classical and Bayesian setting, while Type S errors are more relevant in evaluating the long-term frequency properties of statistical inferences about comparison in the case of continuous parameters.