![]() I'm not a huge fan of using the linear probability model, although it can be done in this case, as you only have categorical explanatory variables. Gpower: calculate power of multiple regression analysis with between groupsįirst, we have to think clearly about what tests we're going to conduct. How to calculate power (or sample size) for a multiple comparison experiment? Just to show I did due diligence, I found these old posts that weren't detailed enough:ĪNCOVA vs multiple regression the same: so why different power analysis results? Is it because there is no consensus on how to do it well? Do academics in practice ignore these nuances? I'm not necessarily trying to get the more rigorous statistical answer, but just to understand what social scientists find as "good enough" in practice. Why is that? Like sociologically, why are there so few practical guides or cookbooks on power analysis special extensions? There are a million on simple two-sample t-test, but none on multiple treatments or interactions. I've seen some theoretical papers on this, but no simple practical guide. The programs I'm familiar with are GPower, PowerUp, and Stata's -power- command. Instead, I will run 3 regressions for each permutation, not just a full ANOVA (ANOVA doesn't tell me WHICH mean is largest). 05/3 right? The best solution I can find is a power calculation formula for ANOVA, but ANOVA isn't exactly the type of analysis I will do. ![]() Three treatments, where I want to compare them all together (so there would be 3 comparisons in total). ![]() ![]() Is the answer as simple as doing a power analysis for T1 vs Control, then using the same sample size for T1 on T2? I can't get a straight answer stating that anywhere (maybe it's too "obvious" for most people, but not me)
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |