Antonio, Could there be a trend present in your data? That would generate autocorrelations with many lags. Try adding a linear and/or quadratic trend term to your list of predictors. David Greenberg, Sociology Dept., New York U.
----- Original Message -----
From: Antonio Silva <[email protected]>
Date: Tuesday, August 26, 2008 11:13 pm
Subject: st: serial autocorrelation in residuals, what to do?
To: [email protected]
> Hello:
> Today, I followed some of your advice. The arpois program does indeed
> work with the new stata and I downloaded it. Running the model using
> arpois seems to help...but only a very little bit. I have run the
> model with ar(1) all the way through ar(10), and only when we get to
> very large ar numbers does the autocorrelation of residuals begin to
> abate seriously. So I suppose I could do this...but the results are
> unwieldy, as the results have 10 ar coefficients in the printout.
> As for including a lagged dependent variable on the right side of the
> equation, it does help. But not enough to overcome the very serious
> autocorrelation of residuals. Finally, I also ran a nbreg, but it
> reduces to the Poisson, I assume because there is no overdispersion
> present.
> Robust standard errors help too, but still not enough.
> I am trying more, and I keep on learning more, so thanks a lot!
> Antonio
>
> _________________________________________________________________
> Get thousands of games on your PC, your mobile phone, and the web with
> Windows�.
> http://clk.atdmt.com/MRT/go/108588800/direct/01/
> *
> * 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/
*
* 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/
