Syndicate content

Add new comment

Submitted by Ron on
David: It seems to me the real power of this method would be attempt at the knotty question of external validity of RCTs. At present if one were interested in getting at the issue of predicting impact in the general population we do not have a rough and ready quantitative method. With thus though one could essential reweight the general population to "look" like the control population in the study and thus examine how the treatment would fare if scaled up. Of course the other alternative is to use structural models but this seems like a simpler more mechanical way of getting at this.