RE: st: estadd margins with multiple outcome models

[Cameron McIntosh <cnm100@hotmail.com>]

Hi again Arka,
I just took a look at my response and realized that you'd probably want to stick with your probit model if you were trying to relax the parallel slopes assumption, and not move to a logit model, as the references I gave you suggested. I guess I'm so used to seeing logits everywhere that my brain was on auto-pilot for that one. :) Anyway, please take a look instead at:
Pfarr, C., Schmid, A., & Schneider, U. (2011). Estimating Ordered Categorical Variables Using Panel Data: A Generalised Ordered Probit Model with an Autofit Procedure. Journal of Economics and Econometrics, 54(1), 7-23.http://www.eeri.eu/documents/jee/JEE_2011_01_02.pdf
Pfarr, C., Schmid, A., & Schneider, U. (2010). REGOPROB2: Stata module to estimate random effects generalized ordered probit models (update).  Statistical Software Components, Boston College Department of Economics.http://ideas.repec.org/c/boc/bocode/s457153.html
Stewart, M.B. (2004). Semi-nonparametric estimation of extended ordered probit models. The Stata Journal, 4(1), 27-…39.http://www.stata-journal.com/sjpdf.html?articlenum=st0056
I would be interested to see the difference between the approaches... it would be a good way to test the tenability of the parallel slopes assumption for sex. Pfarr et al.'s program will tell you if you need to relax it or not... do you have some other covariates/controls too?
Best,
Cam  
> From: cnm100@hotmail.com
> To: statalist@hsphsun2.harvard.edu
> Subject: RE: st: estadd margins with multiple outcome models
> Date: Sun, 11 Sep 2011 21:45:38 -0400
> 
> Hi Arka,
> Before you proceed further, have you checked to see whether your data meet the proportional odds assumption? If not, you may want to consider some other options:
> Kjellsson, M.C., Zingmark, P.-H., Jonsson, E.N., & Karlsson, M.O. (2008). Comparison of proportional and differential odds models for mixed-effects analysis of categorical data. Journal of Pharmacokinetics and Pharmacodynamics, 35(5), 483-501.
> Williams, R. (2006). Generalized ordered logit/partial proportional odds models for ordinal dependent variables. The Stata Journal, 6(1), 58–82.
> Cam
> > From: gabuisi@gmail.com
> > Date: Sun, 11 Sep 2011 18:18:04 -0700
> > Subject: st: estadd margins with multiple outcome models
> > To: statalist@hsphsun2.harvard.edu
> > 
> > Dear Statalisters,
> > 
> > I am using Stata 11 on a 64 bit Windows Vista computer.I am estimating
> > an ordered probit model of the effect of gender on education.My
> > education variable is a categorical variable with three categories 1,2
> > and 3. Sex is a dummy variable which takes the value of 1 if male and
> > 0 otherwise.So I run the following regression:
> > oprobit edu i.sex [pweight=weight],robust
> > 
> > Now what I need is a table with the marginal effects  for each
> > education outcome  with their standard errors. I usually use -esttab-
> > to get tables. This is where I run into difficulty. I run  the
> > folowing after the oprobit command :
> > estadd margins,dydx(sex) predict(outcome(1)
> > estadd margins,dydx(sex) predict(outcome(2)
> > estadd margins,dydx(sex) predict(outcome(3)
> >  But I can not figure out how to store these results by different
> > names so that I can make tables using esttab since estimates store
> > does not seem to work.
> > 
> > Basically I want to replicate the output from the following set of
> > commands by using margins since mfx is now obsolete in Stata 11.
> > 
> > oprobit edu sex [pweight=weight],robust
> > mfx compute,predict(outcome(1))
> > estimates store model1
> > mfx compute,predict(outcome(2))
> > estimates store model2
> > mfx compute,predict(outcome(3))
> > estimates store model3
> > esttab model1 model2 model3,se star(*0.10 **0.05 ***0.01) margin
> > 
> > 
> > I would really appreciate any help in this regard.
> > 
> > Regards,
> > Arka
> > 
> > References: Jann, Ben "Estout:Stata modele to make regression tables"
> > (http://ideas.repec.org/c/boc/bocode/s439301.html)
> > *
> > *   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/
 		 	   		  

*
*   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/