FAQ/ENH: observed, post-hoc power: don't make invalid conclusions

[josef-pkt]

Computing observed power, i.e. power at observed effect sizes, have been strongly criticized in recent literature, but mainly because it is used to argue in favor of null or alternative hypothesis.

However, I'd like to include them together with effect sizes, mainly so we get an additional statistic that complements p-value with a measure of type 2 error.
Also, we should be able to use the same functions to compute prospective power and sample size computation at different effect sizes and sample sizes.
(currently I have observed power in anova PR)

O’Keefe, Daniel J. 2007. “Brief Report: Post Hoc Power, Observed Power, A Priori Power, Retrospective Power, Prospective Power, Achieved Power: Sorting Out Appropriate Uses of Statistical Power Analyses.” Communication Methods and Measures 1 (4): 291–99. https://doi.org/10.1080/19312450701641375.

Onwuegbuzie, Anthony J., and Nancy L. Leech. 2004. “Post Hoc Power: A Concept Whose Time Has Come.” Understanding Statistics 3 (4): 201–30. https://doi.org/10.1207/s15328031us0304_1.

There are more articles, but I don't find them anymore at the moment.