A positive correlation between two independent variables can lead to estimates that are negatively correlated. But the estimates themselves are not biased by correlations among the predictors. Large standard errors for your estimates are a feature of your data. You won't get rid of them by orthogonalization. Instead you will get estimates for parameters other than the ones you are interested in. David Greenberg, Sociology Department, NYU
----- Original Message -----
From: Erasmo Giambona <[email protected]>
Date: Wednesday, August 20, 2008 5:25 am
Subject: Re: st: Multicollinearity and Orthogonalization
To: [email protected]
> Dear All,
>
> Thanks very much for your helpful hints. One thing remains unclear to
> me. I thought (perhaps wrongly) that one problem with
> multicollinearity is that if y1 and y2 are highly correlated (e.g.,
> 0.9), then their coefficient estimates in regression can get
> "artificially" alternate signs (e.g., + and - or vice versa). To me is
> not clear yet whether you suggest that Stata would not suffer from
> this problem or whether I should orthogonalize in this case.
>
> Any furhter thoughts would be truly appreciated.
> Regards,
> Erasmo
>
>
>
>
> On 8/17/08, SR Millis <[email protected]> wrote:
> > In some case, one way to deal with collinearity between an
> interaction and its terms is to mean center the variables.
> >
> > SR Millis
> > *
> > * 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/
*
* 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/
