I have a question about model specification for a Mixed Model
analysis.
It is a repeated measures design and I have two binary variables,
assignment (placebo vs active med) and age (under 48 and over 48).
I have a model with several interaction terms and want to predict the
marginal means for time by assign by age which I have done.
I do not use the EMM feature because this does not make allowance for the
covariate terms in my model.
So the question arises -
Is there any reason I should specify the binary variables of assignments
and age only as factors, or is it meaningful to enter them as
covariates?
The difference in specification doesn't affect the predicted values for
means but it does affect the table of fixed effects.
Since different F values and df are obtained by these two methods I am
presented with very different interpretations of the effects.
Any guidance would be appreciated.
Thanks .
Roy
Roy Money M.S.
Biostatistician
Treatment Research Program
Connecticut Mental Health Center
Yale University
tel (203) 974-7048
fax (203) 974-7057
email
[hidden email]
Please be aware that email communication can
be intercepted in transmission or misdirected. Please consider
communicating any sensitive information by telephone, fax or mail.
The information contained in this message may be privileged and
confidential. If you are NOT the intended recipient, please notify
the sender immediately with a copy to [hidden email] and destroy
this message.