Re: st: Epidemiological Confounder Selection

[Jay Kaufman <Jay_Kaufman@unc.edu>]

Salah Mahmud wrote:
> 
> Does anyone know of a Stata routine to perform confounder selection
> using a selected change-in-estimate criterion and backwards deletion?

I believe that the user-written routine you cite (epiconf.ado) does 
indeed use a change in estimate approach, not a testing based approach. 
The epiconf.ado output gives p-values in the table, but the variables
are selected for dropping based on the change in the effect parameter.