Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Matching in STATA


From   Steven Samuels <[email protected]>
To   [email protected]
Subject   Re: st: Matching in STATA
Date   Mon, 23 Jun 2008 09:41:05 -0400

Complete or 'full" matching on the propensity score and covariates has recovered experimental differences that other approaches missed (Diamond and Sekhoe, 2005; Ho et al., 2007). Ho et al. recommend matching generally as a first stage in any analysis. Lunceford and Davidian (2004) derive a control function approach to combining regression and "inverse probability of treatment weighting" which is "doubly-robust": it is consistent if either the model for the propensity score OR the regression model is correct.

-Steve

Reference:

Diamond, A and JS Sekhon. 2005. Genetic matching for estimating causal
effects: a general multivariate matching method for achieving balance in observa-
tional studies. Working paper, Travers Department of Political Science, UC Berkeley.

Ho, D, K Imai, G King, and E Stuart (2007) Matching as nonparametric preprocessing for reducing model dependence in parametric causal inference. Political Analysis, Vol. 15: 199-236.

Lunceford, JK, and M Davidian (2004) Stratification and weighting via the propensity score in estimation of causal treatment effects: a comparative study. Statistics in Medicine, 15:2937-60.




On Jun 23, 2008, at 7:10 AM, Maarten buis wrote:


--- Henry <[email protected]> wrote:
Salah, I am using secondary data from a primary care database. This
contains  patient details -medical, therapy and demoraphic
information. In this case, i guess i will have to do some matching.
Not all patients in the database have the outcome of interest and
thus in order to use classical logistic regression models to
evaluate risk of outcome with exposure allowing for possible
conounders , we need to find controls as close to the cases as much
as possible.
Matching techniques and `classical logistic regression' both control
for meausured confounders, and both require that the model is correctly
specified. I never understood the advantage matching/reweighting
techniques over multiple regression techniques. I got the distinct
impression that in many applied studies the matching/reweighting
technique is believed to solve much more problems than is actually the
case, though I probably have to reread (Nichols 2007) when I have time.

-- Maarten

Austin Nichols (2007) Causal inference with observational data, The
Stata Journal, 7(4): 507--541.
http://www.stata-journal.com/article.html?article=st0136




-----------------------------------------
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands

visiting address:
Buitenveldertselaan 3 (Metropolitan), room Z434

+31 20 5986715

http://home.fsw.vu.nl/m.buis/
-----------------------------------------


__________________________________________________________
Sent from Yahoo! Mail.
A Smarter Email http://uk.docs.yahoo.com/nowyoucan.html
*
* 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/
*
*   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/



© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index