367135_en-US

liliconcom

Generalized Logits Models

This is a new one to me.  Customer is asking about "Generalized Logits Models."  I don't know if JMP or JMP Pro will do these kinds of models.

 

As a reference, I found this SAS proc that does what he wants:

https://v8doc.sas.com/sashtml/stat/chap22/sect9.htm 

 

Customer happens to be using that exact data set from the link to see if JMP can do it.

 

It looks like GLM comes close (link functions, etc.), but I'm not familiar enough with it to know if it can be used for this purpose.

 

Anyone else know?  If so, just point me to the proper reference and I'll read up on my own.

 

Thanks!

 

Re: Generalized Logits Models

I 'think' that means nominal logistic regression, as opposed to 'cumulative logits' that used in ordinal logistic regression.

Re: Generalized Logits Models

The Nominal Logistic personality produces the same results as in the example, though the ANOVA tables in Figs 22.9 and 22.10 correspond to the Wald tests under the red triangle, instead of the default Effect Likelihood Ratio Tests. To match the parameter estimates in Fig 22.11, the value orders need to be set in the data table so that the baseline categories are set the same as in the example (i.e., School: 1, 2, 3; Program: Regular, Afternoon; Type: Self, Team, Class). Going with the default alphabetic value ordering will give different parameter estimates but should still yield the same Wald test results.