On Tue, Feb 25, 2014 at 2:51 PM, Nathan Pace <n.l.p...@utah.edu> wrote:
> I¹ve used twang to get ATE propensity scores. > > I¹ve done multivariable, case weighted Cox PH models in survey using > svycoxph and in rms using cph with id(cluster) set to get robust estimates. > > The model language is identical. > > The point estimates are identical, but the CIs are considerably wider with > svycoxph estimates. > > There is a note in the svycoxph help page stating the SEs should agree > closely unless the model fits poorly. > > The actual note on the svycoxph help page says "The standard errors agree closely with survfit.coxph for independent sampling when the model fits well, but are larger when the model fits poorly. " That is, the note is for the survival curve rather than the coefficients. It's still surprising that there's a big difference, but I think we need more information. -thomas -- Thomas Lumley Professor of Biostatistics University of Auckland [[alternative HTML version deleted]]
______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.