Re: st: Additional covariate kills effect in -reg-, but not in -cnreg-; Which should I trust?

[Maarten buis <maartenbuis@yahoo.co.uk>]

--- On Wed, 1/12/10, Stata Chris wrote:
> I'm regressing unemployment durations on some hypothesized
> determinants thereof.
> 
> To get more statistical power, I want to start with
> least-squares regression rather than a Cox model, but I
> thought that I should at least deal with the fact that many
> durations are right-censored in that the spell has not yet
> ended by the time my panel ends.
>
> So I decided to use -cnreg-. I find that whereas when I use
> -reg- adding year fixed effects "kills" my main effect of
> interest, the same is not true when I use instead use -cnreg-.
> Since I do not fully understand what -cnreg- does differently
> from -reg-, I wonder which of the two results I should trust
> more?

If you decide to use -cnreg- than you should have used -streg-
instead. The ideas behind these two commands are somewhat 
similar, but -streg- is explicitly made for your type of data, 
so much more appropriate. If you have a large proportion of 
censored cases then -regress- is simply not an option.

Hope this helps,
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/