Sunday, February 19, 2006
st: mean substitution with dummies for adjustment
In our section we frequently run multivariate models (usually simply regressions), and then wish to calculate adjusted values. The model is often something like:
lung function = age sex infection center_2 center_3 center_4
To calculate the adjusted value for each center we substitute the mean for age, sex, infection, then simply sum:
adjusted lung function for center_1= intercept + b1*(mean_age) + b2*(mean sex) + b3*(mean infection) adjusted lung function for center_2= intercept + b1*(mean_age) + b2*(mean sex) + b3*(mean infection) + b4 adjusted lung function for center_3= intercept + b1*(mean_age) + b2*(mean sex) + b3*(mean infection) + b5 adjusted lung function for center_4= intercept + b1*(mean_age) + b2*(mean sex) + b3*(mean infection) + b6
where the centers are coded as 0,1 indicators, center 1 is left out as the referent. The idea is that each center is dealing with the same "average" set of patients, and the center effect is captured by its coefficient.
First, does anyone else do things this way? Do you have a reference? I have spent some time on line and in my stats books and can't come up with anything useful.
Hebe Quinton Dartmouth Hitchcock Medical Center Clinical Research/Department of Medicine 506 Rubin One Medical Center Drive Lebanon, NH 03756
(603) 653-3536 Fax (603) 653-3554
I am Xena, Worrier Princess. * * 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/
Links to this post: