How do I compare variance and covariance accross groups (MANOVA)?

art

New Member
#1
Hello everyone,
As you'll guess, I'm pretty clueless in statistics, but I do try my best...
So I'm doing a MANOVA, my sample sizes are all over the place, normality is violated and obviously Box's test is significant as well, so I read that Tabachnick and Fidell (2007) say that I need to check variance/covariance: if the largest groups have produced the largest variances/covariances, then probability values are going to be conservative; if the smaller groups have produced the largest variances, then probability is going to be liberal.
My question is: what is meant by variance/covariance? Should I be checking variance by group for each dependent variable (I have 2)? They don't have the same patterns across the two. And what about covariance? Are they talking about SSCP matrices? Surely not, as these matrices are not produced separately for groups within the independent variables. Is there a way to produce these matrices for subgroups?
As you'll probably have guessed, I'm working on SPSS.
Hope someone can help out, please do even if you feel really condescending reading this!
Many thanks in advance.