Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Theocharis Kromydas <theocharis.kromydas@strath.ac.uk> |
To | "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu> |
Subject | st: gllamm identification of levels in a repeated cross-sectional data |
Date | Thu, 28 Nov 2013 11:44:46 +0000 |
Dear Statalisters I want to use gllamm to perform a logit model. My dependent var is employment status (0 unemployed 1 employed) and my independent are educational level, experience and year and some others. So it is like that Emp=b0+b1 edlev+b2exp+b3year + b4 country b5 edlev*year + b6 edlev*country+ e It is a pooled cross-sectional data from two years (2004 and 2010) and respondents are not the same. First I want to run a single level model and then a two-level one taking country as level 2. For level 2 I want to include random effects both in terms of the intercept and in terms of slope for edlev and year. Then to compare the two models Moreover I want also to include a contextual variable of unemployment rate (U) for both 2004 and 2010 and its cross-level interactions with edlev, time and country and then compare with model 1 and 2 I was reading a lot of stuff for gllamm but I can't get its syntax very well. First of all when I am creating different interactions variable for each of the countries I analyse, I am getting a response that "variables have been dropped, can't continue" So, for the sake of proceeding with this and see some results I left just one country interaction in the mode, just to do some tests with gllamm.l I then run some regressions but I am having problems on identifying the levels. My original person identification no is idno but contains same values for both years (i.e. I can find idno=1 for 2004 and 2010 but it's not the same person). When I am putting idno after prefix i (i.e. gllamm Emp edlev exp year country edlev*year edlev*country1, i(idno) link(logit) family(binomial) pweight (wt) adapt) the program identifies idno as 2nd level variable What I want is to include persons as level 1 and countries as level 2, but once I put both inside the relevant parentheses then I am getting a three level model. So I guess something is wrong with idno variable and I don't know how to fix it. I would appreciate any comment you have on all these as I am very confused at the moment. I am more used in xtmelogit command and I found the procedure quite straightforward but because it doesn't allow any weights I had to use gllamm which, honestly, I can't really get it that well. Thanks, Harry * * 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/