Combining Propensity Score Matching with Additional Adjustments for Prognostic Covariates

By Donald B. Rubin, Neal Thomas. 2000

783 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/eb8f6ac4aceb33cab42cab81d497a01b84181cef

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Statistics and Probability

Citations in the corpus

Why Propensity Scores Should Not Be Used for Matching (2019, background): “In Section 3, we show that PSM is blind to an important source of information in observational studies because it approximates a completely randomized rather than a more informative and powerful, fully blocked experiment (Rubin and Thomas, 2000).”
Optimal Full Matching and Related Designs via Network Flows (2006, background): “2), inference conditional on matched sets (Cox and Hinkley 2000; Hosmer and Lemeshow 2000; Hansen 2004a), or inference from a random effects model (Raudenbush and Bryk 2002; Agresti 2002, chap 12; Smith 1997); and the second reduces inferential precision.”
The effects of public support schemes on small and medium enterprises (2015, positive): “For these reasons it was proposed in literature to combine these two methods, in particular when there are some covariates for which particularly close matches are required (Rubin and Thomas, 2000).”
Using Propensity Scores to Help Design Observational Studies: Application to the Tobacco Litigation (2001, methodology): “The use of the metric matching after propensity score matching is the observational study equivalent of blocking in a randomized experiment; see Rosenbaum and Rubin (1985) for an example and Rubin and Thomas (2000) for further explanation.”
Improving propensity score weighting using machine learning (2010, methodology): “In our simulations, the outcome Y is fully determined by the observed covariates W i and the treatment A , which is not what we normally expect to see in practice but is a common strategy for assessing bias in propensity score settings [34, 35].”
An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies (2011, background): “Propensity score matching can be combined with additional matching on prognostic factors or regression adjustment (Imbens, 2004; Rubin & Thomas, 2000).”
Teaching Statistical Inference for Causal Effects in Experiments and Observational Studies (2004, background): “In the advanced courses, some attention is paid to theoretical results, such as in Rubin and Thomas (2000).”
Results of multivariable logistic regression, propensity matching, propensity adjustment, and propensity-based weighting under conditions of nonuniform effect. (2006, methodology): “Several adjustment methods incorporating the estimated propensity score have been proposed, including matching (8, 9), regression adjustment (1, 10), and weighting (11–15).”
A Survey on Causal Inference (2020, methodology): “eatment effects [11]. 3.3.1 Distance Metric. Various distances have been adopted to compare the closeness between units [43], such as the widely used Euclidean distance [110] and Mahalanobis distance [113]. Meanwhile, many matching methods A Survey on Causal Inference • 15 develop their own distance”
Affinely invariant matching methods with discriminant mixtures of proportional ellipsoidally symmetric distributions (2006, background): “It is also natural, as in [16, 17, 18, 19], to compare the results of the matching to random sampling done in an affinely invariant way, such as randomly sampling from the original treated and control groups, thereby sampling from each component in proportion to its fraction in the population [the α”