From: Bayesian additional evidence for decision making under small sample uncertainty
Frequentist inference significant | Question: The analysis provides evidence of an effect. How robust would this evidence be given additional data? BAE output: BAE provides the range of results that would yield non-credibility from a hypothetical follow-up study. If the tipping point is close to the null, then the evidence provided by the current analysis is strong. Conclusion: If the range beyond the tipping point is considered plausible, then the current evidence is not strong enough. |
Frequentist inference non-significant | Question: The analysis does not provide strong evidence of an effect. Is it worthwhile to obtain more data for further study? BAE output: BAE provides the range of results that would need to be observed in a follow-up study in order to yield credible evidence of an effect. If the tipping point is close to the null, then the evidence provided by the current analysis is strong. Conclusion: If the range beyond the tipping point is considered plausible, then we can declare the observed data to be useful, and might perform a follow-up study. |