Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: mean-centering to interpret interaction in treatreg


From   "Shehzad Ali" <sia500@york.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: mean-centering to interpret interaction in treatreg
Date   Thu, 31 Jul 2008 16:36:33 +0100

Hi listers,

I am using a treatreg model with an interaction term. Here is the simplified
model:

treatreg y x1_c x2_c x3 x1_c*x2_c, treat(x2=x1 x3 x4)

Here I have used mean-centred x1 and x2 (called x1_c and x2_c respectively)
for the -outcome- part of the model (dependent var is y) but I used original
values of x2 in the -treat- part because it only allows two values for the
dependent variable, i.e. 0 and 1. Now looking at the result, Stata drops
x2_c and instead uses x2 for the outcome part. I was wondering if I could
still interpret the coefficient on interaction term and x1_c and x2 at mean
levels? Can anyone please suggest whats the best way to approach this model?


Any help would be appreicated.

Many thanks,

Shehzad 


*
*   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/



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