Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: st: AW: Tobit, negative predictesd values


From   Maarten buis <maartenbuis@yahoo.co.uk>
To   statalist@hsphsun2.harvard.edu
Subject   RE: st: AW: Tobit, negative predictesd values
Date   Tue, 10 Nov 2009 17:03:56 +0000 (GMT)

--- On Tue, 10/11/09, Solorzano Mosquera, Jenniffer wrote:
> Actually, I took as 0's those missing
> values of the firms whose answer to the question "Did you
> offered training to your employees this year?" was negative.
> In fact, when the answer was not they didn't have to answer
> anything about intensity. In thaht case, intensity only
> could be larger than 0. But, as Zwick estimated in his
> paper, he tooks the whole sample, those who answered yes and
> no in the offering training question. I still don't
> understand why.

Because you are predicting the latent (= not observed) ideal 
intensity, i.e. how much the employer _wants_ to train his 
empoloyees, not how much (s)he actually offered. Your model
assumes that those employers that did not offer any training
actually wanted to take training away from their employees,
but were not able/allowed to do so, so they ended up offering
nothing. Since you predicted the latent propensity, you
should expect negative intensities for those with observed 0s.

-- Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


      

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index