Re: st: Understanding Factor variables - is order significant ?

["Michael N. Mitchell" <Michael.Norman.Mitchell@gmail.com>]

Dear Jesper

  This is not really a "Stata" issue, but is an issue regarding the coding of dummy 
variables. Let's take an even simpler case, suppose you have a variable named "female" 
that is coded 1 if a someone is a female, and 0 if they are a male. But, instead, you 
change your mind and include the coefficient for a variable named "male" that is coded 1 
if you are a male, and 0 otherwise. The coefficient will change (in the case of a linear 
model, it will be of opposite sign) but note the p value will remain the same as it still 
remains a test of the difference between males and females.

  Extend that idea to your interaction... Suppose you flip the coding of your "ra" and 
"dm" variables. Note that the test of the interaction, the p value, will remain the same 
(assuming both are dummy variables). The coefficients of "ra" and "dm" will change as 
well, due to the change in coding. The details get more complicated, but are explained in 
section 3.5 of http://www.ats.ucla.edu/stat/stata/webbooks/reg/chapter3/statareg3.htm . It 
is explained using the old "xi" terminology, but the issues still are the same.

  I hope that helps,

Best luck,

Michael N. Mitchell
Data Management Using Stata      - http://www.stata.com/bookstore/dmus.html
A Visual Guide to Stata Graphics - http://www.stata.com/bookstore/vgsg.html
Stata tidbit of the week         - http://www.MichaelNormanMitchell.com



On 2010-05-25 3.22 PM, Jesper Lindhardsen wrote:
> Dear Statalisters,
>
> I am having a hard time understanding why 2 regression models that
> differ only by the "order" of the included factor variables yield
> different results???
> I can't (or am too slow to) find the answer in the documentation, but I
> think it is related to the parsing of the baselevel specifiers (see
> model 1 legend = _b[0o.ra#0b.dm] ???).
>
> Here are the 2 commands and resulting output - as you can see I've only
> changed b1.ra#b0.dm to b0.dm#b1.ra. Output has been edited, but only
> left out if identical between models.
>
> (System: Stata 11/MP for windows, born 10 feb 2010)
>
> 1)
> poisson _d b1.ra#b0.dm i.alder_k sex if ex==0, e(risk_tid) irr
> coeflegend
> 			
> _d          IRR       Legend
> 			
> ra#dm
> 0 0     1.487748  _b[0o.ra#0b.dm]
> 0 1     1.968017  _b[0.ra#1.dm]
> 1 1     2.787839  _b[1b.ra#1.dm]
>
> alder_k
> 1     6.176815  _b[1.alder_k]
> 2     18.09798  _b[2.alder_k]
>
> sex    2.070646  _b[sex]
> risk_tid  (exposure)
> 			
> 2)
>
> poisson _d b0.dm#b1.ra i.alder_k sex  if ex==0, e(risk_tid) irr
> coeflegend
>
> 			
> _d         IRR         Legend
> 			
> dm#ra
> 0 0     .5935912  _b[0b.dm#0.ra]
> 1 0     1.169963  _b[1.dm#0.ra]
> 1 1      1.65762  _b[1.dm#1b.ra]
>
> alder_k
> 1     6.171095  _b[1.alder_k]
> 2     18.07456  _b[2.alder_k]
>
> sex    2.072329  _b[sex]
> risk_tid  (exposure)
> 			
> Hope its not too elementary.....
> 		
> Thanks you all for your contributions to statalist, it's a really
> valuable source of information for me.
> Regards,
>
>
> Jesper Lindhardsen
> MD, Ph.d. student
> Department of Cardiovascular Research
> Copenhagen University Hospital, Gentofte
> Denmark
>
>
> *
> *   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/