Thank you! Would you consider this to be the "observed"/post-hoc power of these studies?
@dsquintana
3 жыл бұрын
I think this is better described as how likely it would be to observe a significant effect, given the study design and a range of effect sizes. The concept of observed power is problematic daniellakens.blogspot.com/2014/12/observed-power-and-what-to-do-if-your.html
@arthuralbuquerque6471
3 жыл бұрын
@@dsquintana thanks for the quick reply! Exactly, I was about to refer to that blog post from Lakens. I am starting to learn power analyses, so sorry for the questions, but I didn't quite understand what you said. How does this package test a "range of effect sizes" if the dataset you presented has a single effect size per study?
@dsquintana
3 жыл бұрын
Unless you’re looking at your entire population there’s no ‘true’ effect size. So, this package calculates power based on a range of possible effect sizes. For reference, power is also calculated for the observed effects size from the meta-analysis. You might find this other video useful kzitem.info/news/bejne/raqcnHyghpd1nX4
@quanguc4806
2 жыл бұрын
hey. please share the sample data with the code which it will useful for data understanding.
@dsquintana
2 жыл бұрын
Thanks for watching! I posted the link to the script in the video description, but here it is in case you missed it www.dsquintana.blog/meta-analysis-power-plot/
Пікірлер: 10