Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: RE: RE: interaction term in negative binomial regression


From   "Scott Merryman" <smerryman@kc.rr.com>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: RE: RE: interaction term in negative binomial regression
Date   Fri, 15 Apr 2005 13:29:34 -0500

Sheng,

No.

Though does it matter?  Though it may change the standard error (and
coefficient) of your dummy variable, the marginal effect and the standard
error depend on both the dummy variable and interaction term.  

For example:

. sysuse auto, clear
(1978 Automobile Data)

. gen foreXmpg = foreign*mpg

. center mpg

. gen forXc_mpg = foreign*c_mpg

. reg price mpg foreign foreXmpg, nohead
---------------------------------------------------------------------------
       price |      Coef.   Std. Err.      t    P>|t|     [95% Conf.
Interval]
-------------+-------------------------------------------------------------
         mpg |  -329.2551   74.98545    -4.39   0.000    -478.8088
-179.7013
     foreign |  -13.58741   2634.664    -0.01   0.996    -5268.258
5241.084
    foreXmpg |   78.88826   112.4812     0.70   0.485    -145.4485
303.225
       _cons |   12600.54   1527.888     8.25   0.000     9553.261
15647.81
----------------------------------------------------------------------------

. reg price c_mpg foreign forXc_mpg, nohead
---------------------------------------------------------------------------
       price |      Coef.   Std. Err.      t    P>|t|     [95% Conf.
Interval]
-------------+-------------------------------------------------------------
       c_mpg |  -329.2551   74.98545    -4.39   0.000    -478.8088
-179.7013
     foreign |   1666.519    717.217     2.32   0.023      236.075
3096.963
   forXc_mpg |   78.88826   112.4812     0.70   0.485    -145.4485
303.225
       _cons |   5588.295   369.0945    15.14   0.000     4852.159
6324.431
----------------------------------------------------------------------------

It would be incorrect to say that the effect of car type (the dummy variable
_foreign_) is statistically insignificant in the first regression (with
uncentered interaction term) and statistically significant in the second
regression (with the centered interaction term).   In fact, the marginal
effect and standard error of _foreign_ is the same in both models.


Hope this helps,
Scott



> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-
> statalist@hsphsun2.harvard.edu] On Behalf Of Sheng Wang
> Sent: Friday, April 15, 2005 8:40 AM
> To: statalist@hsphsun2.harvard.edu
> Subject: st: RE: RE: interaction term in negative binomial regression
> 
> 
> Dear Scott:
> 
> This is very helpful. Thank you! Just want to clarify. About the output
> you
> had, was that based mean-centered mpg or not? Because if I don't center my
> continuous variable, I would have some s.e. of above 4 while if I center
> it
> first before running the regression, all s.e. were below 1. Does that make
> a
> difference?
> 
> Thanks again!
> Sheng
> 


*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index