It's not clear to me that the relationship between two X variables is
of any importance, unless you are concerned about collinearity
reducing the efficiency of your estimates, measurement error, or
perhaps specification error in your linear model. In the absence of
these concerns, if X1 affects X2 and both X1 and X2 affect Y, the
regression of Y on X1 and X2 captures the partial effect of X1 and X2
regardless of the X2=f(X1) effect.
As for evaluating structure in the residuals, here is a cheap example:
webuse grunfeld, clear
xtreg mval inv, fe
predict yhat
predict ehat, ue
sc ehat yhat || qfit ehat yhat
I think there was an article in the Stata Journal on this latter topic
(I don't have it; perhaps someone else can summarize it), and -help
regress_postestimation- suggests a variety of post-regression
diagnostic plots.
On Dec 5, 2007 5:50 AM, valentina p <[email protected]> wrote:
> Good morning,
>
> I am using a fixed effect model.
>
> How can I express the interaction between the regressors?
>
> I mean: bewteen my regressors there are immiration and natural growth. Natural growth (from descriptive analysis I made) has the same trend as immigration(t-1), so I suppose that immigration strongly influences natural growth. How can I express this relation?
>
> And apart from this, any suggestion on how to analyse the residuals in a fixed effect model like xtreg y x1 x2..., fe?
*
* 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/
