Rosana Ferrero

ANCOVA is a valid tool under well-controlled circumstances (prior baseline, relevant covariate, parallel slopes). In unbalanced real-world data, with multiple time points, missingness, and heterogeneous effects, it becomes less adequate.

In its classical form it does not handle repeated measures with more than two levels (otherwise you move to RM-ANCOVA with even stricter assumptions).

A common mistake: using ANCOVA to “correct” post-test values by the pre-test without considering interactions (violation of homogeneity of slopes).

Yes, that must be taken into account. But ANCOVA is by no means free of assumptions; we have:
-Linearity between the covariate and the dependent variable.
-Homogeneity of regression slopes.
-The covariate must be measured prior to the experimental manipulation and not affected by the treatment.

How to Analyze Pre-Post Data There are two ways to analyze pre-post data: repeated measures or ANCOVA. How you decide between them depends on one key attribute.

I believe both procedures are useful but not equivalent: the appropriate one clearly depends on the specific scientific question you want to answer.

www.theanalysisfactor.com/pre-post-dat...

I don’t know, I’ll have to take a deeper look at pwrss, but the first thing I found are post hoc power analyses, which in practice make little sense and are highly questioned.

What advantage do you see compared to Pwr and Webpower?“​​​​​​​​​​​​​​​

buff.ly/4fLF8XV
buff.ly/4i6QFmh

Science Isn’t Broken If you follow the headlines, your confidence in science may have taken a hit lately. Peer review? More like self-review. An investigation in November uncovered a scam in which researchers were rubber-...

fivethirtyeight.com/features/sci...

¡Enhorabuena Clara!

Trying to compile a list of people with active pedagogic interests in teaching statistics go.bsky.app/Qg6YSq6

thank you for making this pack!👏