Partial proportional odds model PreviousNext
Mplus Discussion > Categorical Data Modeling >
 Xiaojing Yan posted on Tuesday, March 20, 2007 - 9:23 am
Dear All,

I am conducting analysis with categorical outcomes.

It has been suggested that partial proportional odds model is appropriate as it can relax the assumption for those variables which violates the assumptions. I am just wondering whether it is possible to use 'partial proportional odds model’in Mplus?

Many thanks.
 Bengt O. Muthen posted on Tuesday, March 20, 2007 - 4:48 pm
I have not studied partial prop odds modeling, but I wouldn't be surprised if it could be done via multinomial logistic regression adding parameter constraints using Model Constraint. If you can't figure it out, please send me a reference (Hedeker?) and I will try to find time to do this.
 Xiaojing Yan posted on Wednesday, March 21, 2007 - 11:32 am
Thank you so much for this and here is the reference.

Richard Williams, 2006.
"Generalized ordered logit/partial proportional odds models for ordinal dependent variables," Stata Journal, StataCorp LP, vol. 6(1), pages 58-82, March.

Many thanks.
 Bengt O. Muthen posted on Wednesday, March 21, 2007 - 7:02 pm
I can only get the abstract for this Stata publication - does it refer to other published work (such as Hedeker's)?
 Xiaojing Yan posted on Thursday, March 22, 2007 - 3:21 am
One of the references is by Peterson and Harrell (1990)** on partial proportional odds models. Thank you so much.

**Peterson, B. & Harrell, F. E. 1990. Partial Proportional Odds Models For Ordinal Response Variables. Applied Statistics-Journal Of The Royal Statistical Society Series C, 39(2): 205-217.

 Bengt O. Muthen posted on Thursday, March 22, 2007 - 4:51 pm
We will let you know how this is done in Mplus. Stay tuned.
 Xiaojing Yan posted on Friday, March 23, 2007 - 3:20 am
Thank you very much. I also got confirmation from the author, Richard William, whether the model is as same as that in Hedeker's**.

**Hedeker, D. & Mermelstein, R. J. 1998. A Multilevel Thresholds of Change Model for Analysis of Stages of Change Data. Multivariate Behavioral Research, 33(4): 427.

He replied 'I've just quickly skimmed the article and we certainly seem to be talking about similar things, although I haven't read it closely enough to say if what we are doing is identical.'

Looking forward to hearing from you.
 Bengt O. Muthen posted on Friday, March 23, 2007 - 2:05 pm
Well, it turns out that our tricks for doing this won't work out, so we have to wait until a later version of Mplus implements it.
 Xiaojing Yan posted on Monday, March 26, 2007 - 3:06 am
Thank you very much for your efforts and help though. I am looking forward to the next version.
Many thanks.
 Mads Meier Jæger posted on Wednesday, November 26, 2008 - 1:41 am
Dear Xiaojing

I'm pretty sure the gllamm ado for Stata can do partial proportional odds models in a mixed model framework. Also, the regoprob ado should work.

 Anonymous posted on Thursday, November 08, 2012 - 8:26 am

I would be interested to know if the recent version(s) of Mplus are able to fit partial proportional odds models: I wish to model an ordinal outcome with fewer assumptions than the ordered logistic model but with more than the multinomial model.
 Bengt O. Muthen posted on Thursday, November 08, 2012 - 11:24 am
We don't have that model yet, but it is high on our to-do list.
Back to top
Add Your Message Here
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Options: Enable HTML code in message
Automatically activate URLs in message