The design versus the analysis of observational studies for causal effects: parallels with the design of randomized trials
Проектирование vs анализ наблюдательных исследований для причинных эффектов: параллели с дизайном рандомизированных испытаний
2006-10-30
SCID: 54.1/yfaejmue
Discuss with AI
causal inferencecovariate balancedesign versus analysisobservational study designpropensity score methods
Figures from the paper
Abstract (AI)
For estimating causal effects of treatments, randomized experiments are generally considered the gold standard. Nevertheless, they are often infeasible to conduct for a variety of reasons, such as ethical concerns, excessive expense, or timeliness. Consequently, much of our knowledge of causal effects must come from non-randomized observational studies. This article will advocate the position that observational studies can and should be designed to approximate randomized experiments as closely as possible. In particular, observational studies should be designed using only background information to create subgroups of similar treated and control units, where 'similar' here refers to their distributions of background variables. Of great importance, this activity should be conducted without any access to any outcome data, thereby assuring the objectivity of the design. In many situations, this objective creation of subgroups of similar treated and control units, which are balanced with respect to covariates, can be accomplished using propensity score methods. The theoretical perspective underlying this position will be presented followed by a particular application in the context of the US tobacco litigation. This application uses propensity score methods to create subgroups of treated units (male current smokers) and control units (male never smokers) who are at least as similar with respect to their distributions of observed background characteristics as if they had been randomized. The collection of these subgroups then 'approximate' a randomized block experiment with respect to the observed covariates.
Key Findings
1
Design of observational studies should use only background (covariate) information to create subgroups of similar treated and control units without access to outcome data.
2
Objective creation of covariate-balanced subgroups assures design objectivity and can be accomplished using propensity score methods.
3
Observational studies can and should be designed to approximate randomized experiments as closely as possible to estimate causal effects.
4
Propensity score–based subgroups can make treated and control units as similar in observed covariates as if randomized, approximating a randomized block experiment.
5
The paper demonstrates this approach in a US tobacco litigation application comparing male current smokers to male never smokers using propensity score methods.
Research Object
Non-randomized observational studies for estimating causal effects (designed to approximate randomized experiments)
Research Subject
Designing observational studies—using background covariate information and propensity score methods to create covariate-balanced subgroups of treated and control units without using outcome data so as to approximate randomized trials
Publication Details
Publication Date
2006-10-30
Journal
Publisher
ISSN
Cited by
1081
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest