Multiple Tests or MANCOVA?

tomc

New Member
#1
I have 7 subscales/dependent variables and I want to find the relationship/effect of 1 two-group qualitative variable, a quantitative covariate, and their interaction on those 7 DVs. I ran 7 hierarchical regressions (with the interaction in block 2), and 7 ANCOVAs to test these and in all cases there was a significant main effect of the numeric variable, a significant interaction, but no significant main effect of the qualitative variable. What I'm concerned with is whether this poses an alpha inflation problem and if I should consider (1) using a correction such as Bonferroni, or (2) running a MANCOVA. When I run the MANCOVA with the 7 DVs, the Box's M test significance is p = .002. Some sources I have seen suggest this is not acceptable, while I have seen sources that say p > .001 is OK.

Should I move ahead with either 7 hierarchical regressions or 7 ANCOVAs, or should I move ahead with the MANCOVA as is or by making some changes? If I should just use 7 hierarchical regressions or 7 ANCOVAs, how do I apply the Bonferroni correction to them?

Thanks in advance.
 
Last edited: