G*Power 3.1 Tutorial: Factorial ANOVA Power Analysis (Episode 5)
HTML-код
- Опубликовано: 21 авг 2024
- In this episode, I explain how to complete a priori power analyses for a factorial/between-subjects ANOVA.
G*Power 3.1 download: www.psychologi...
Find me on Twitter: / profaswan
Go to my website: swanpsych.com
Twitch streams on psych & related topics: / cogpsychprof
Discuss this video and others on my Discord channel: / discord
![Eminem - Somebody Save Me (feat. Jelly Roll) [Official Music Video]](http://i.ytimg.com/vi/Vwa0HenQMi4/mqdefault.jpg)
The reason the numerator df is decoupled from the # of groups is because you can have different test structures from a set number of groups. Suppose you have a 2 x 4 (8 groups), 1 of your main effect would have 1 df, the second would have 3 df, and the interaction would have 3df as well. However, if your 8 groups form a 2 x 2 x 2 design, each main and interaction would be tested with a 1df numerator.
That makes perfect sense! In the moment I was like "huh" but it's so obvious. Cheers for the input!
@@AlexanderSwan Thank YOU for the useful videos on GPower. Do you know if the GPower team has put out a more elaborate manual discussing power estimation for repeated designs ANOVA. The manual I was able to find online is very sparse. Thanks.
@@ChungkingExpress1996 I do not think there is much anywhere at all, to be honest. The dev team, from my perspective, is merely trying to keep the program updated to work with OSes, looking at the release notes. I it looks like manuals are in the hands of users, kind of like me putting these tutorials together... they aren't even on YT!
Does this mean that the calculated statistical power for a 3way interaction is the same as for a main effect in a 2x2x2 design, but in a 2x4 design you'd have reduced statistical power for the interaction?
"to give this tutorial a little more.... UMMMPGH" xD
By setting your numerator df to be groups - 1, you basically did a one-way design, not a factorial design.They are independent so that you can specify a factorial design with separate power estimates for different effects in the model.
Ok, if I understand you, you can specify the numerator based on which effect you want, such as a main effect or an interaction, correct? Thanks for sharing the info, I appreciate it!
ty bro