Applying Sphericity Corrections to Simple Main Effects Tests
Hi,
I have a repeated measures ANOVA with a significant interaction. Mauchly's test indicates that sphericity is violated for this comparison, so I applied the Greenhouse-Geisser correction to the omnibus test of the interaction.
However, I noticed that the GG correction did not apply to the simple main effects comparison. Is applying it as simple as multiplying the degrees of freedom by the GG epsilon value from the Mauchly's test output?
I also noticed that the simple main effects table only includes one df value and not the error df value. Should I also include the error df when reporting the simple main effect, or is this intentional?
The simple main effects are for a factor with two levels at each level of a factor with three levels.
Appreciate any help, thank you.
Comments
I'll pass this on to the JASP team member with the necessay expertise; sorry for the tardy reply.
EJ
Hi @Jadran ,
Yes, you can take the correction factor and apply it to the df in the simple effects table, and with some extra calculations also recompute the F statistic and p-value: the Sum of squares will stay the same, but the mean square will change (since you divide by a different df).
Ideally you'd also need to report the error df, so it would indeed be nice for JASP to also provide those (as well as maybe automatically apply the sphericity corrections to the simple main effects).
Cheers,
Johnny