Subscribe Bookmark RSS Feed

Generalized estimating equations in JMP

jennifer_n_blak

Community Trekker

Joined:

Jan 25, 2016

Hello!

I have found numerous questions in the forum about JMP's capability to perform generalized estimating equations (GEE), but few have been answered.  Because of this I am starting a new thread.

 

I am conducting a multinomial multiple logistic regression, with outcome measurements that are taken over several years.  Because of the categorical output, repeated measures anova is not an option, as one post suggested.  I've been told I should use a GEE for my analysis.  This is apparently very easy in SPSS, but I can't find it in JMP.  I do not have access to SAS.  

 

A release regarding JMP 11 from four years ago said that GEEs were coming, but I cannot find it in JMP13.

 I am looking for options already present within JMP or within a JMP add-in, not scripting suggestions in JSL.

 

Thanks for your help!

 

JMP 11: https://community.jmp.com/t5/JMP-Blog/JMP-11-statistics-sneak-peek/ba-p/30320

1 ACCEPTED SOLUTION

Accepted Solutions
susan_walsh1

Staff

Joined:

Jul 22, 2014

Solution

JMP does not have the capability to fit generalized estimating equations. The JMP development team is aware of user interest in this functionality.

 

5 REPLIES
billw_jmp

Staff

Joined:

Jul 2, 2014

Jennifer,

I believe what you are calling GEE's is referred to as Generalized Regression in JMP which is found only in the Pro version of JMP.  There are some options for Generalized Linear Model in regular JMP as well.  Both will be found under the Analyze > Fit Model platform.  Be sure to specify your output/Y variable and then you will be able to choose a modeling type under Personality.

 

HTH

Bill

jennifer_n_blak

Community Trekker

Joined:

Jan 25, 2016

Dear Bill,

   Thank you for the reply!  After looking at generalized regression, I don't believe that is what I was attempting to do via a generalized estimating equation.

 

I am essentially trying to perform a multinomial multiple logistic regression on measurements collected over several years.  I have a treatment and control group, with a categorical response measured in 2014, 2015, 2016, 2017 etc.   I am still investigating the glm platform in JMP 13 pro.  I know that these can theoretically be used with repeated measures data, but when I go to fit model and select glm, I do not see options for repeated measures (as is shown in a JMP video on repeated measures in mixed models https://www.jmp.com/content/jmp/en_us/events/ondemand/mastering-jmp/linear-mixed-models-using-jmp-pr...).  Do you have any thoughts on how to access repeated measures capabilities through the glm platform you recommended?

Thanks so much for your thoughts!

Jennifer

 

billw_jmp

Staff

Joined:

Jul 2, 2014

Jennifer,

Regarding repeated measures via the GLM platform.  That is not possible at this point in JMP.  

 

Just for others who might be looking in, GLM is found in regular JMP as well as JMP Pro.  Generalized Regression which is JMP PRO only has a couple more options for analyzing categorical responses than GLM, but is also not built for analyzing auto-correlated data or repeated measures at this point.

 

 

jennifer_n_blak

Community Trekker

Joined:

Jan 25, 2016

That is disappointing to hear.  It is at least good to know that I can stop looking around in JMP for this test.

Thank you for your assistance!

Jennifer

susan_walsh1

Staff

Joined:

Jul 22, 2014

Solution

JMP does not have the capability to fit generalized estimating equations. The JMP development team is aware of user interest in this functionality.