First-order Markov (transition) model for ordinal data

[mgx]

I'm currently exploring the possibility of using a transition model to analyze my ordinal longitudinal data, as an alternative to both marginal models and random-effects models. I'm specifically interested in finding SAS tutorial (papers) that cover this topic. While I'm aware of the tutorial available for binary data (https://support.sas.com/kb/24/494.html), I'm wondering if there are similar resources available for ordinal data analysis.

[sbxkoenk · Accepted solution]

See here :

Usage Note 22871: Types of logistic (or logit) models that can be fit using SAS®

https://support.sas.com/kb/22/871.html

--> In the paragraph on "Transition models for discrete state space stochastic processes", it is said that :

Discrete time Markov chains can be represented as log-linear models.

So I would google the internet with search terms :

• SAS Modeling Ordinal Data with Log-linear Models
• SAS ordinal log-linear models
• SAS Log-Linear Model for Ordinal Data

Here's an example for a log-linear model on count data :
SAS/STAT User's Guide
The GENMOD Procedure
Example 51.7 Log-Linear Model for Count Data
https://go.documentation.sas.com/doc/en/pgmsascdc/9.4_3.5/statug/statug_genmod_examples07.htm

Most of the info you will find is using PROC GENMOD but ...

PROC CATMOD (for categorical data modeling) can also be used for log-linear modeling.

Koen