Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | "Michael N. Mitchell" <Michael.Norman.Mitchell@gmail.com> |
To | statalist@hsphsun2.harvard.edu |
Subject | Re: st: ## Interaction syntax |
Date | Tue, 11 Jan 2011 15:41:27 -0800 |
Dear RobinThat is very peculiar, because indeed the two examples you provided should be identical. The only thing I can think of that could cause an issue is if you had ill behaved data, and somehow these two techniques were dealing with the ill behaved data differently. (An example would be multi-collinearity or empty cells). Are you able to provide examples of output where these two techniques diverge (showing the model that converged). Perhaps the model that converged might give a clue?
Best regards, Michael N. Mitchell Data Management Using Stata - http://www.stata.com/bookstore/dmus.html A Visual Guide to Stata Graphics - http://www.stata.com/bookstore/vgsg.html Stata tidbit of the week - http://www.MichaelNormanMitchell.com On 2011-01-11 3.28 PM, Robin Jeffries wrote:
I was under the impression that for a categorical variable 'wave' with 3 levels (0, 1, 2) and another binary indicator variable 'group' (0,1) then the following statements are the same: 1) xtlogit i.wave##i.group (other covar) 2) xtlogit wave1 wave2 group wg1 wg2 (other covar) where wave1, wave2, wg1, wg2 are the manually created indicators and interactions for wave and wave*group respsectivly For most outcomes I am using this with, they produce the exact same results. However there have been some instances where using the second method results in a model that won't converge, but the first will. Is there an explanation for this? Thanks, Robin Jeffries * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/
* * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/