Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
| From | Maria Sokolova <maria.sokolova@graduateinstitute.ch> |
| To | statalist@hsphsun2.harvard.edu |
| Subject | st: Warning on mlogit |
| Date | Wed, 10 Jul 2013 18:43:04 +0200 |
Dear all, Please give some advice on the following. All regression worked for the full sample of several years. When I restrict the sample for 3 years only, the "Warning: variance matrix is nonsymmetric or highly singular" appears. I read the thread that was here several years ago (http://www.stata.com/statalist/archive/2008-01/msg00980.html), but do not really understand how it is applicable for me.... Mlogit has three choices and the clusters are for year-product (also tried month-product, same thing). Choices are about 30% 55% 15% for a given year. the point is, that even the mlogit on the constant shows this warning... the simplest (shortest) example is mlogit pt dtime*, b(2) robust cluster(cluster1) pt - three choices unordered dtime - monthly dummies.. What can I do about this Warning? Thanks again for your help, Best, Maria -- - - - - - Maria Sokolova PhD student International Economics The Graduate Institute of International and Development Studies maria.sokolova@graduateinstitute.ch * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/faqs/resources/statalist-faq/ * http://www.ats.ucla.edu/stat/stata/