Lecturer: Dr. Erin M. Buchanan
Missouri State University
Summer 2016
Lecture materials and assignments available at statisticsofdoom.com.
https://statisticsofdoom.com/page/advanced-statistics/
I forgot about power! If you wanted to use GPower you could use:
F test - linear multiple regression R2 increase from zero
Hit determine - put in your expected R2 for the overall model or how big the mediating effect should be, calculate and transfer to main
Alpha = .05
Power = .80
Predictors = 3 x, m1, m2
I found that I would need 37 participants. Power for mediating models can be calculated in more complex ways, but assuming that M1 and M2 will be predicting a lot of Y, while X will not in that last model is a good or easy way to do it.