Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: Re: RE: controlling what variables to dropping?


From   "Steve Stillman" <[email protected]>
To   <[email protected]>
Subject   st: RE: Re: RE: controlling what variables to dropping?
Date   Thu, 20 Oct 2005 18:39:11 +1300

Hi Andrea.  Dummy variables are dropped for a number of reasons: 1) 1 dummy of a complete set (or the model constant) needs to be dropped to avoid perfect collinearity, 2) you only have 1 observation for a particular dummy (ie country) , or 3) you have multiple observations within country but there is a high degree of collinearity between the outcome variable and RHS variables across these observations (ie you have limited (or no) variation within countries).

The only way to solve what you view as a problem is to do more diagnostic work with your data.  Countries with only one observation or limited variation across observations will have to be dropped or grouped with other countries.  There is no way to avoid this problem when estimating a fixed effect model - diagnostics are just easier when estimating using a dummy variable approach instead of xtreg, fe.

Cheers,
Steve

-----Original Message-----
From: [email protected]
[mailto:[email protected]]On Behalf Of Andrea
Molinari
Sent: Tuesday, October 18, 2005 3:59 AM
To: [email protected]
Subject: st: Re: RE: controlling what variables to dropping?


Steve,

Thanks for your suggestion. I have included country dummies by hand 
into my (linear) cross-section model. The problem I've got is that some 
of those country dummies get dropped, and I cannot have control over 
the dummies that get dropped.

Any further suggestions will be more than welcomed!


Thanks,
Andrea

----- Original Message ----- 
From: "Steve Stillman" <[email protected]>
To: <[email protected]>
Sent: Saturday, October 15, 2005 2:48 AM
Subject: st: RE: controlling what variables to dropping?


> You can just create the fixed effects by hand (a dummy variable for 
each
pair of countries) and include them in your model (assuming it is a 
linear
model).
>
> Cheers,
> Steve
>
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]]On Behalf Of Andrea
> Molinari
> Sent: Saturday, October 15, 2005 3:59 AM
> To: [email protected]
> Subject: st: controlling what variables to dropping?
>
>
> I am sorry about the confusing subject in my previous email. Here's 
the
one
> I intended to put
>
> Dear Statalisters,
>
> I have a simple question, but that I haven't been able to answer for 
some
> time...
>
> I have a bilateral trade (gravity) model, with pairs of countries as
> individuals.  I want to estimate a model in cross-section that 
includes
> country fixed effects, but have not been able to control the dropping 
of
> some of these effects.
>
> Does anyone know of any command to tell STATA how to prioritise the
dropping
> of collinear variables?
>
> Any suggestions/ideas will be more than welcomed!
>
>  Andrea
>
> Andrea Molinari
> DPhil candidate in Economics
> University of Sussex
> Department of Economics
> Arts Building E - office 517
> Falmer, Brighton BN1 9SN
> United Kingdom
>
> Tel: +44 (0) 1273 - 873538
> Fax: +44 (0) 1273 - 673563
>
> *
> *   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/
>
*
*   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/



© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index