Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Hey Sky <heyskywalker@yahoo.com> |
To | statalist <statalist@hsphsun2.harvard.edu> |
Subject | st: question on gllamm for discrete latent variable with one factor structure |
Date | Sun, 9 May 2010 13:25:41 -0700 (PDT) |
Hey, all I use gllamm to model people's choice among edu, wrkwt (work with training), wrknt (work without training) and home (stay at home), panel data. edu/wrknt/wrkwt/home are dummy variables. the latent variable's one factor stucture, sita=a +c*mu, represents for different choice, there is different parameters, a and c, for these people (Ham and Lalonde 1996). mu has two discrete value which represents there have two types people. that is, for people who choose edu, the structure of the latent variable is sita_edu = a_edu + c_edu*mu for wrknt: sita_wrknt = a_wrknt + c_ wrknt*mu. and the same for the other choices. I have tried the following code, which takes Rabe-Hesketh's model 2 in notes GLLAMM models with discrete latent variables as example. Model 2: Class probabilities depend on sex (vj=[fem]) eq fac: wom cou mar fin gen ris rap eq fem: fem gllamm ab wom cou mar fin gen ris rap, nocons weight(wt) i(id) l(logit) f(binom) eqs(fac) peqs(fem) ip(f) nip(2) but my code does not work eq sita1: edu eq sita2: wrknt eq sita3: wrkwt gllamm choice indep, i(id) ip(f) eq(sita1 sita2 sita3) nrf(4) family(binom) link(mlogit) base(4) trace dots where is the mistake? should I use the option peqs/geqs and why? I do not really understand the option peqs() here. thanks for your answer and any suggestion is appreciated. Nan from Montreal * * 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/