The method described so nicely by Francis can be implemented in
SAS PROC NLMIXED or PROC GLIMMIX.
From: Francis Levira <[log in to unmask]>
To: [log in to unmask]
Sent: Wed, November 4, 2009 11:17:30 AM
Subject: Re: Repeated measurement in categorical data
There is an extension of logistic regression that account for the
categorical data with more than two category which is called
*Proportional Odds Logistic regression*. Using this model the probability of
observing a lower response versus higher one is modeled.
Since you have repeated measure data, a modified GEE (Generalize Estimating
Equation) version of analyzing longitudinal data will be used. The important
thing is to have correct assumption about the correlation structure between
your repeated measurement. Marginal or Subject specific model can be fitted.
Hope this will give you a light during your analysis.
On Wed, Nov 4, 2009 at 5:53 PM, PGW <[log in to unmask]> wrote:
> I have a set of data with 3 repeated measurements in categorical data 1, 2,
> 3, or 4. and like to see if there is any increase along the time. I would
> use ANOVA if it was a continuous data.
> It would be very much appreciated if anyone could give me some advice on
> how to tackle categorical data with repeated measurement.
> Thank you very much in advance.
> Kind regards
Center for Statistics