# How to interpret the SPSS output for MANOVA?

I am struggling with the interpretation of my MANOVA in SPSS. I have two DV and one IV (age), in the MANOVA table Wilk's lambda is .053, thus not significant at a alpha level of .05. When I then look at the "tests of between subjects effects" table, the effect of age is significant for one of the DV (p <.05).

How should I report and interpret this? Should I ignore the between subjects table as the MANOVA table is not significant?

The Wilks lambda is a test of the multivariate significance.

The test of between subjects effects provides two tests of univariate significance.

Are you interested in the multivariate significance - if so, look at the lambda. If you're not, you shouldn't have done MANOVA, and you can ignore it and look at the univariate significance for each variable separately.

Ideally, you would know this before you look at the p-values. Looking at the p-value and then deciding you don't like the result, and therefore you should do a different test makes your p-values kind of nonsensical.

• Thanks! i did want to do an MANOVA, not an anova, i was just unsure in how far you have to include this into your results (aka my results section in my paper), whether it is enough to report on the manova main effect and then ignore whatever is found in the anova. Commented Jul 23, 2013 at 20:19
• i had to investigate this as my supervisor asked me to check whether i should (if significant) later include age as a covariate in my model. however as i said it is not significant in the manova, only in the anova and now i'm confused as to what to do with this.. does this mean age has an effect and should it be a covariate then? any help would be appreciated! Commented Jul 23, 2013 at 20:21

There is a difference between ANOVA and MANOVA. Anova is employed when there is randomization of subjects. When in your study you have 2or3independent variables on a dependent variable. MANOVA is used when you have 2or3independent variables on 2 or more dependents variables. This is why it is multivariate while ANOVA is univariate.

• I am not sure this adds much to the original answer, can you clarify? Commented Oct 13, 2016 at 17:09