#statstab #353 The Abuse of Power; The Pervasive Fallacy of Power Calculations for Data Analysis

Thoughts: An seminal paper on "post hoc" power calculations.

#power #QRPs #NHST #posthoc #samplesize #effectsize

https://www.tandfonline.com/doi/abs/10.1198/000313001300339897

#statstab #330 Encourage Playing with Data and Discourage Questionable Reporting Practices

Thoughts: What are and aren't "Questionable Research Practices"? Where is the "grey area"? Interesting opinion piece.

#QRPs #exploratory #EDA #posthoc #phacking

https://link.springer.com/article/10.1007/s11336-015-9445-1

Der #PostHoc-Fehlschluss: Zufällige Ereignisse werden fälschlicherweise als Ursache und Wirkung verknüpft.
Beispiel:
☂️ „Ich habe den Regenschirm mitgenommen – und es hat nicht geregnet! Der Schirm hat den Regen abgehalten!“
🐈 „Ich habe eine schwarze Katze gesehen – danach hatte ich Pech!“
Nur weil zwei Dinge nacheinander passieren, heißt das nicht, dass eines das andere verursacht hat.

👉 Tipp: Prüfe, ob es wirklich eine Ursache-Wirkung-Beziehung gibt oder nur eine zufällige Reihenfolge!

Tutorial paper illustrating how to develop planned contrasts (as an alternative to #PostHoc comparisons) in #Rstats, including empirical examples with reproducible data:
https://journals.sagepub.com/doi/full/10.1177/25152459241293110

#StudyDesign #RCT

#statstab #189 Post Hoc Power: Not Empowering, Just Misleading

Thoughts: "the observed power is a 1:1 function of the P value" If you need a ref for why "post hoc" power is nonsense (paywalled).

#power #posthoc #QRPs #errorcontrol #poweranalysis

https://www.sciencedirect.com/science/article/pii/S0022480420305023

#statstab #176 Pairwise test after a (omnibus) χ2 test of equality of proportions w/ R

Thoughts: I've often had this complaint with many packages where all they offer is the overall dif, not per two cells.

#chi2 #posthoc #r #rstats

https://stats.stackexchange.com/questions/111986/post-hoc-chi2-test-with-r

#statstab #162 Post-hoc Contrasts by {rcompanion}

Thoughts: A good tutorial should always be shared. This goes step-by-step and provides options. Including plots.

#stats #emmeans #posthoc #ttest #nhst #r #dataviz #models

https://rcompanion.org/rcompanion/h_01.html

#statstab #71 {emmeans} contrasts are not (always) like t-tests

Thoughts: t-tests use parts of your data, emmeans use your fitted model (inc. dfs). (g-computation is even more confusing). Important to know what these all do.

#stats #posthoc #nhst

https://stats.stackexchange.com/questions/369619/how-are-the-degrees-of-freedom-in-the-emmeans-package-calculated-r