Re: st: multinomial choice with panel data

["Guillaume Frechette" <gurst1@hotmail.com>]

Hi Guanghui: you should take a look at gllamm which I believe will allow two 
levels of random effects (although I don't know for certain). If you also 
want to have laged dep. var. then you need to deal with the initial 
condition problem. One way to do this is correlated random effects (I 
include two references at the bottom). To read about random effects ordered 
probit/logit take a look at any advance textbook in econometrics. You could 
of course simply estimate an ordered probit/logit and adjust the standard 
errors via something like cluster which is simpler but biased (and 
inconsistent at least for logit). Good luck.

@INBOOK{Heckman:1981a,
  author    = "James J. Heckman",
  chapter   = "The Incidental Parameters Problem and the Problem of Initial 
Conditions in Estimating a Discrete Time-Discrete Data Stochastic Process",
  editor    = "Charles F. Manski and D. McFadden",
  title     = "Structural Analysis of Discrete Data with Econometric 
Applications",
  publisher = "MIT Press",
  address   = "Cambridge, MA",
  year      = "1981",
  pages     = "179-195"}

@ARTICLE{Chamberlain:1980a,
  author  = "Gary Chamberlain",
  title   = "Analysis of Covariance with Qualitative Data",
  journal = "Review of Economic Studies",
  year    = "1980",
  volume  = "47",
  number  = "",
  month   = "",
  pages   = "225-238"}



> From: "Guanghui Li" <guanghui@u.washington.edu>
> Reply-To: statalist@hsphsun2.harvard.edu
> To: <statalist@hsphsun2.harvard.edu>
> Subject: st: multinomial choice with panel data
> Date: Mon, 16 Jun 2003 16:27:00 -0700
>
> Dear all,
>
> I have cross sectional time series data on a group of women. In each 
> period,
> the women choose from 3 choices: 0, 1 or 2.
>
> Which choice the woman chooses in each period depends on a group of
> independent variables and also choices she made in the previous periods.
>
> I want to estimate a multinomial model of the three choices, accounting for
> correlations among the observations from the same women, and also
> correlations among women residing in the same areas due to cluster 
> sampling.
>
> The fixed effect approach does not serve my purpose because the majority of
> the independent variables are time-invariant.
>
> And I am not sure which econometric model to use so that her current period
> choice may depend on choices she made in all previous periods. I could
> construct two new variables which reflect all her previous decisions. But I
> am not sure it is the right way to do.
>
> Thank you in advance for your generous help.
>
> Sincerely,
>
> Guanghui Li
>
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
_________________________________________________________________
Add photos to your messages with MSN 8. Get 2 months FREE*.  
http://join.msn.com/?page=features/featuredemail

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/