On Sun, 20 Apr 2008 16:25:11 -0700 (PDT), beginner1.mat@[EMAIL PROTECTED]
> Can anyone tell me the advantage of running a MANOVA rather than
> multiple ANOVA for each dependent variable? I ran 3 ANOVAs on 3
> different dependent variables and was criticized for this approach,
> but I do not know why. Is this similar to running multiple t-tests
> instead of an ANOVA? Any help, and/or references would be appreciated!
The main *advantage* of running MANOVA is that
only with MANOVA can you detect effects that depend
on (an unexpected) pattern among the outcome variables.
The *criticism* probably arises because of the multiple-test
cir***stance, and controlling by Bonferroni levels leaves
you without the complexity of MANOVA to deal with.
Constructing a composite out of the three outcomes, or
naming one of them as primary, increases statistical power.
This makes use of what you know and want to know, and
is often the preferable approach.
--
Rich Ulrich
http://www.pitt.edu/~wpilib/index.html
|
