Wouldn't you want to use -clad- rather than simply -qreg- as the alternative
specification to account for the censoring? -clad- will provide consistent
estimates in the presence of non-normality, heteroskedasticity, and
asymmetric errors. It does, however assume that the errors have a zero
median.
Hope this helps,
Scott
> -----Original Message-----
> From: [email protected] [mailto:owner-
> [email protected]] On Behalf Of Flora Au
> Sent: Friday, May 06, 2005 10:05 AM
> To: [email protected]
> Subject: st: re: tobit
>
> Hi everyone,
>
> Thank you very much for those who reply to my last inquiy, now I have
> another one. I believe I posted this a couple days ago, however no reply
> on it yet. The question was is it any command that would test
> heteroscedacity or omitted variable for tobit model. One paper suggest to
> test the hetero with a hausman test. That is first test the model with the
> tobit then test the model with a -qreg- model. So the command read as
> follow:
>
> quietly tobit eq_5d density degree non_cauc momage infant young mededu
> highedu medinc highinc service, ll(0)
> est store tob
> quietly qreg eq_5d density degree non_cauc momage infant young mededu
> highedu medinc highinc service
> hausman tob .
>
> Does anyone know if this is correct to test hetero? If not, any
> suggestion?
>
> Thank you,
> Flora
>
> *
> * 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/
*
* 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/
