www.cmm.bristol.ac.uk/forum

Dear all,
We are working on a data set with 1.7m of observations and runmlwin saved our life by substantially reducing the run time. Our data have three-levels, e.g. firm, region, and country, where firm nested on region, while region nested on country. Currently we try to use the Nested Logit regression to control the correlation within region. But it seems that runmlwin does not have a command for the nested logit. Any suggestion on alternative way to get the nested logit estimation?
Many thanks,
Kevin

Dear Kevin,

Thank you for your kind comments regarding runmlwin

You are correct there is no nested logit functionality in MLwiN and therefore runmlwin.

I'm afraid I also do not know what software can estimate multilevel versions of these models.

You could take a look and email the developers of the following for advice?

http://www.limdep.com/

Best wishes

George