Re: Re: st: What is the effect of centering on marginal effects?

[Christopher Baum <kit.baum@bc.edu>]

<>
On Aug 3, 2012, at 2:33 AM, John wrote:

> I don't believe you are using the right vif.  Should it not be the 
> - -uncentered- option in vif for both regressions below? In this case the 
> vifs are the same.

Agreed, centering in VIF matters (my colleague David B. always recommends normalizing each column of X to unit length)
but that does not cause the VIFs to be equal:

. estat vif, uncentered

    Variable |       VIF       1/VIF  
-------------+----------------------
      length |   1598.70    0.000626
      weight |   1493.44    0.000670
    c.length#|
    c.weight |   1347.44    0.000742
   intercept |   1180.62    0.000847
-------------+----------------------
    Mean VIF |   1405.05



. estat vif, uncentered

    Variable |       VIF       1/VIF  
-------------+----------------------
    c_length |      9.71    0.102969
    c_weight |      9.75    0.102591
  c.c_length#|
  c.c_weight |      1.94    0.516267
   intercept |      1.91    0.522829
-------------+----------------------
    Mean VIF |      5.83

Thanks for the references.

Kit


Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
                             An Introduction to Stata Programming  |   http://www.stata-press.com/books/isp.html
  An Introduction to Modern Econometrics Using Stata  |   http://www.stata-press.com/books/imeus.html


*
*   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/