Statalist The Stata Listserver


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Partial Relation in Fixed Effects Estimation


From   David Jacobs <jacobs.184@sociology.osu.edu>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Partial Relation in Fixed Effects Estimation
Date   Thu, 09 Mar 2006 13:34:28 -0500

Something like what you want probably can be found using a regression diagnostic called "acprplot" or one of its counterparts (see the chapter on regression diagnostics in the manual or the help files). Generate the dummies for your cases and then put them an OLS model along with your other explanatory variables. Then use this or one of the other diagnostic graphics that show the independent relationship between a particular explanatory variable and the dependent variable.

Dave Jacobs

At 09:42 AM 3/9/2006, you wrote:

I really appreciate the help people provide on this board.

I'm running fixed-effects regressions to estimate the impact of var1 on
growth using the following command.

tsset id year

xtreg growth var1 var2 , fe

var1 turns out to have significant impact on growth in this specification.

I'm interested in generating a scatter plot of growth and var1 conditional
on control variable var2 and the fixed effects (id dummies). I would like
to visually inspect this relationship to see which observations are
driving the results. That is, I want to identify the within-id variation
that's giving me the fixed effect results.

To do this I've done this:

xtreg growth var2, fe
predict growthres, e

xtreg var1, fe
predict var1res, e

reg growthres var1res
twoway scatter growthres var1res

I was hoping to obtain the residual growth conditional on var2 and the id
effects and obtain the residual of var1 conditional only on the id
effects.  Then I wanted to plot the residuals to see which observations
were important for the results.

But when I run the regression of growthres on var1res, I get no
significant results and the scatter plot is basically a cloud. This
doesn't seem consistent with the initial result from running the
regression (at the top).

Would anyone know of a way to do something like this to identify the
within-id variation that's giving me the results?

Thanks very much for your help.

Jason
*
*   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 1996–2020 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index