Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: RE: Problems with ci_marg_mu and ordinal variable

From   Nick Cox <>
To   "''" <>
Subject   st: RE: Problems with ci_marg_mu and ordinal variable
Date   Wed, 18 Jan 2012 17:25:45 +0000

You do not specify where these user-written programs come from. -findit ci_marg_mu- finds precisely one location, SSC. 

You are correct. The program there does not allow the option -above()-. The error message comes from -gllapred-, which it calls, but -ci_marg_mu- offers no way to specify the extra option. 

Last I heard the program author is not a member of Statalist, so if you do not care to fix a clone of the program yourself or do not get a better answer, I recommend contacting the author directly. 


Lilli Banholzer

I am trying to get confidence intervals for my multilevel model, using gllamm, gllapred and finally  the "ci_marg_mu" command.

My model is: 
gllamm demobilize  length_stay education econ_stabilty, i( region_birth_n) link(ologit) adapt eform 

I then use gllapred with the "marginal" und "mu" option to obtain predicted probabilities. Since I am using an ordinal dependent variable, I also specify the "above ()" option. 

gllapred prob0, marginal mu above(0)
gllapred prob1, marginal mu above(1)

And here is where I get stuck. When I try to get the prediction intervals with the "ci_marg_mu" command, STATA tells me: 

ci_marg_mu lower upper
"must specify above() option for ordinal responses"

However, apparently the "above" option is no allowed with the ci_marg_mu command as stata tells me: 

ci_marg_mu lower0 upper0, above(0)
option above() not allowed

What am I doing wrong? How can I specify the above option? I am using Stata 11 for Mac.

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index