test for homogeneity of regression for stepdown F in doubly multivariate ANOVA

[Suzanne Decker]

I am working on a dataset that has data on client variables across two
assessment types and four time points under 3 experimental conditions.
With the help of a statistician, our team ran a doubly multivariate ANOVA
to examine changes in a group of client variables across time, across
assessment types, and across experimental conditions.  We obtained the
omnibus results as well as both univariate and stepdown F tests.

Tabachnick & Fidell's text (5th ed) suggests reporting the stepdown
because the DVs are correlated. The text also says I should test for
homogeneity of regression, and gives an example of the syntax needed for a
multivariate test. However, does anyone know how the setup would be done
for a doubly multivariate test like the one we used?  I have tried
adapting the syntax provided - no luck.
Thanks,
Suzanne

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Ryan]

Suzanne,

What procedure did you use to fit this "doubly multivariate ANOVA"? I ask because you could  fit such a model via the MIXED procedure to answer the specific research questions of interest via the TEST subcommand, while accurately parameterizing the residual variance-covariance matrix of the model.

Ryan

On Tue, Dec 6, 2011 at 1:46 PM, Suzanne Decker <[hidden email]> wrote:

I am working on a dataset that has data on client variables across two
assessment types and four time points under 3 experimental conditions.
With the help of a statistician, our team ran a doubly multivariate ANOVA
to examine changes in a group of client variables across time, across
assessment types, and across experimental conditions.  We obtained the
omnibus results as well as both univariate and stepdown F tests.

Tabachnick & Fidell's text (5th ed) suggests reporting the stepdown
because the DVs are correlated. The text also says I should test for
homogeneity of regression, and gives an example of the syntax needed for a
multivariate test. However, does anyone know how the setup would be done
for a doubly multivariate test like the one we used?  I have tried
adapting the syntax provided - no luck.
Thanks,
Suzanne

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD