www.cmm.bristol.ac.uk/forum

I have been struggling to run multilevel multinomial logit regression on Indian health data to get separate variance for random effect with respect to 2 of the three outcomes (outcome 2 as reference). I am trying to regress BMI categories (underweight, normal and overweight with normal weight as reference category) on socio economic variables. I tried the latest gsem command of stata13 which did not work. gllamm works but produces only one variance. I have not yet tried the expanded version of gllamm. Because gllamm is taking 30 hours to execute just one regression. Before that I thought of taking advantage of your runMLWin module and followed your do file. I am currently having trial version of MLwin in my computer. I am unable to see results in my stata results window. Stata keeps showing me that things are in progress but I do not see any activity except model written in MLwin window. Please help me!!! I have no idea of how MLwin works.

If you are running the trial version of MLwiN then a window will pop up when it starts asking you to continue the trial. You will need to do this in order for MLwiN to fit the model. If you didn't specify the nopause option in runmlwin then MLwiN will pause twice while running, firstly after the model has been specified, and secondly when the model has been fit. After each time it pauses you will need to click the "resume macro" button near the top of the window. Stata will continue to wait until MLwiN has finished running so while the MLwiN window is still open Stata will remain in the state that you describe.