Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Stas Kolenikov <skolenik@gmail.com> |
To | statalist@hsphsun2.harvard.edu |
Subject | Re: st: on WLS in a survey dataset. |
Date | Fri, 2 Jul 2010 14:50:04 -0500 |
You might have to write your own -ml lf- evaluator program... which should be five or so lines of code: normal density for the continuous case (accounting for the known variance), cdf for the truncated case. Unfortunately, there is no code to steal as -tobit- is essentially a built-in command. On Fri, Jul 2, 2010 at 12:48 PM, G. Dai <dgecon@gmail.com> wrote: > dear all, > > I'm using a survey dataset right now and all things work fine except > the following problem: > > After setting the survey data, i need to do a Tobit analysis since my > dependent variable is truncated some way. > > However, the variance of the error term in the Tobit regression is > known, actually they are heteroscedastical and > > can be estimated. To get the efficient estimator, I try to do the commands: > > ::: svyset raehsamp [pw=wtresp],strata(raestrat) singleunit(certainty) > > ::: svy: Tobit y x [aw=1/variance],ll() ul() > > Obviously, the above wouldn't work since we can't have weight again > after declaring the survey data. So, I'm thinking > > how to incorporate the variance weight into the regression. any > suggestion is appreciated. -- Stas Kolenikov, also found at http://stas.kolenikov.name Small print: I use this email account for mailing lists only. * * 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/