Assumption of Normality in MANOVA

[Yoshi]

Hi,

When you conduct a MANOVA with several dependent variables, does the distribution of each DV have to be normal?

Or, is it the distribution of all the DVS combined that has to be normal?

Thank you,

[Dason]

Manova assumes that the error terms follow a multivariate normal distribution. You can't necessarily check this just by looking at the dependent variables themselves. You really do need to look at the residuals to make any assessment about the multivariate normality.