MANOVA Sig. Interaction but no Sig. simple main effects - How to proceed?

#1
Hi All,

I'm running a study on delay discounting. My predictor variables are induction condition (stress vs. relaxation) and reward responsiveness (high vs. low). The dependent variables were proportion of immediate choices on two laboratory measures of discounting.

I ran a MANOVA and got a significant main interaction, but no significant univariate effects. From the Means and SDs it's not clear to me which direction the results are in, so I have no idea how to interpret the significant multivariate interaction. I've read that you can do a Discriminant Function Analysis, but there's only space for one grouping variable.

I'd really (really really really) appreciate some advice. Thank you.

Amanda
 
#2
In general and using the terminology in your post, a significant interaction implies there is a significant "univariate effect" somewhere. This is because in the presence of an interaction we can talk about the significance of one main effect only at the specific level of another effect.