Re: MANOVA design

Bruce,

Thank-you for your reply. I should have given more infomation in my
original post. There are three dependent variables, hence my decision
to use MANOVA.

I guess what I was trying to ask was how to test the null hypothesis
of: "the relationship between the active variables and the dependent
variables is significant for both products". If I include the 2
products in the MANOVA, to make it a 2x2x2(products) from the original
2x2 design, then a non-significant result for the products cell would
be enough to support the null hypothesis that the relationships are the
same for both products.

Brent.



Bruce Weaver wrote:
monopsony wrote:
I have a 2x2 design that was replicated for two different products, to
see if the results would generalise. Is it appropriate to analyse it as
a 2x2x2 design incorporating the additional two product groups, and run
one big MANOVA for both product groups, instead of doing two separate
MANOVAs?


MANOVA = multivariate analysis of variance. Multivariate means there
are 2 or more dependent (or outcome) variables. From your description,
it sounds like you have only one outcome variable. Therefore, I think
you mean ANOVA, not MANOVA.

From what you've said, and assuming the outcome variable is the same
for both products, a 2x2x2 ANOVA sounds preferable to two 2x2 ANOVAs.

--
Bruce Weaver
bweaver@xxxxxxxxxxxx
www.angelfire.com/wv/bwhomedir

.

Relevant Pages

  • Re: sig MANOVA followed by ANOVAs?
    ... > I never understood the point of following up MANOVA with ANOVAs. ... the conservativeness of the classical correction methods. ... > good idea about the relations among the dependent variables and feel ...
    (sci.stat.consult)
  • Re: repeated measures question
    ... Statistica doesn't seem to be the right tool for a very extensive Manova, with several dependent and independent variables. ...
    (sci.stat.math)