Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Should observational studies be designed to allow lack of balance in covariate distributions across treatment groups?

Should observational studies be designed to allow lack of balance in covariate distributions... I thank Miguel Hernan for the valuable discussions. REFERENCES 1. Rosenbaum PR, Rubin DB. The central role of propensity scores in observational studies for causal effects. Biometrika 1983; 70(1):41–55. 2. Pearl J. Causality: Models, Reasoning and Inference. Cambridge University Press: Cambridge, 2000. 3. Rubin DB. Author’s reply. Statistics in Medicine 2002; 27(14):2741–2742. Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/sim.3532 AUTHOR’S REPLY Should observational studies be designed to allow lack of balance in covariate distributions across treatment groups? First I wish to thank Drs Pearl, Shrier and Sjolander for their interest in my work. The target article being discussed in these letters [1] provided advice on the design of nonrandomized observational studies, including how to achieve balance on observed covariates across treatment groups, similar to the balance that randomized experiments achieve on both observed and unobserved covariates; to be clear, a covariate here is a baseline measurement unaffected by the treatments. The only design advice that I can garner from these letters is that we should ponder the possible existence of unobserved imbalances—OK, and then do nothing about some observed imbalances—not OK. The rationale for this implicit advice is the well-known mathematical fact that marginal independence between two http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Statistics in Medicine Wiley

Should observational studies be designed to allow lack of balance in covariate distributions across treatment groups?

Statistics in Medicine , Volume 28 (9) – Jan 30, 2009

Loading next page...
 
/lp/wiley/should-observational-studies-be-designed-to-allow-lack-of-balance-in-dRnAU53M80

References (17)

Publisher
Wiley
Copyright
Copyright © 2009 Wiley Subscription Services
ISSN
0277-6715
eISSN
1097-0258
DOI
10.1002/sim.3565
Publisher site
See Article on Publisher Site

Abstract

I thank Miguel Hernan for the valuable discussions. REFERENCES 1. Rosenbaum PR, Rubin DB. The central role of propensity scores in observational studies for causal effects. Biometrika 1983; 70(1):41–55. 2. Pearl J. Causality: Models, Reasoning and Inference. Cambridge University Press: Cambridge, 2000. 3. Rubin DB. Author’s reply. Statistics in Medicine 2002; 27(14):2741–2742. Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/sim.3532 AUTHOR’S REPLY Should observational studies be designed to allow lack of balance in covariate distributions across treatment groups? First I wish to thank Drs Pearl, Shrier and Sjolander for their interest in my work. The target article being discussed in these letters [1] provided advice on the design of nonrandomized observational studies, including how to achieve balance on observed covariates across treatment groups, similar to the balance that randomized experiments achieve on both observed and unobserved covariates; to be clear, a covariate here is a baseline measurement unaffected by the treatments. The only design advice that I can garner from these letters is that we should ponder the possible existence of unobserved imbalances—OK, and then do nothing about some observed imbalances—not OK. The rationale for this implicit advice is the well-known mathematical fact that marginal independence between two

Journal

Statistics in MedicineWiley

Published: Jan 30, 2009

There are no references for this article.