Are there cases where the Hyunh Feldt and Greenhouse Geisser correction F tests are less conservative than the Conservative Growth Curve approach for ANOVA analysis?
The textbook Design of Experiments in R has a problem in which we find that sphericity is violated in a Repeated Measures experiment. The problem has us compare the significance tests performed using the Hyunh Feldt/Greenhouse Geisser corrections with the significance tests using the growth curve analysis method.
It is insinuated the growth curve method is more conservative. But I’ve found instances where this isn’t true. Could someone speak on this a bit and what conditions may contribute to this?
