s-news
[Top] [All Lists]

Re: Propensity score models

To: "Barlas, Suna" <suna_barlas@merck.com>
Subject: Re: Propensity score models
From: Frank E Harrell Jr <fharrell@virginia.edu>
Date: Thu, 24 Apr 2003 16:17:10 -0400
Cc: s-news@lists.biostat.wustl.edu
In-reply-to: <2C23DE2983BE034CB1CB90DB6B813FD6026A6522@uswpmx11.merck.com>
Organization: University of Virginia
References: <2C23DE2983BE034CB1CB90DB6B813FD6026A6522@uswpmx11.merck.com>
On Thu, 24 Apr 2003 15:32:24 -0400
"Barlas, Suna" <suna_barlas@merck.com> wrote:

> I also have had the problem that Frank is mentioning below and I do agree
> that there are better ways to model "probability of being on treatment". I
> have played with some non-parametric and semi-parametric models and have not
> concluded that there is a best way. It really depends on the data. Sometimes
> very simple parametric models work pretty well and other times it is better
> to go non-parametric at the first stage and semiparametric at the second
> stage. 
> 
> Best,
> 
> Suna.
> 

Just to be clear: the logistic model is a great way to model the propensity for 
treatment.  The problem I alluded to was in how you adjust for the propensity 
score once you've estimated it.  -Frank

----
Frank E Harrell Jr              Prof. of Biostatistics & Statistics
Div. of Biostatistics & Epidem. Dept. of Health Evaluation Sciences
U. Virginia School of Medicine  http://hesweb1.med.virginia.edu/biostat

<Prev in Thread] Current Thread [Next in Thread>