Bootstrap Inference of Matching Estimators for Average Treatment Effects

Otsu, Taisuke; Rai, Yoshiyasu

doi:10.1080/01621459.2016.1231613

Cited by 55 publications

(90 citation statements)

References 26 publications

(22 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, the variance of such an estimator is difficult to obtain because the weights involve the estimated propensity scores, which are estimated from a logistic regression that uses data from all subjects. Ignoring the sampling variability of the estimated propensity score could lead to biased variance estimation, and hence incorrect statistical inference . In this section, we propose a novel approach for variance estimation in the context of propensity score weighting analysis with survival outcomes.…”

Section: Point and Variance Estimationmentioning

confidence: 99%

On the propensity score weighting analysis with survival outcome: Estimands, estimation, and inference

Mao

Yang

et al. 2018

Statistics in Medicine

View full text Add to dashboard Cite

Propensity score analysis is widely used in observational studies to adjust for confounding and estimate the causal effect of a treatment on the outcome. When the outcome is survival time, there are special considerations on the definition of the causal estimand, point, and variance estimation that have not been thoroughly studied in the literature. We investigate propensity score analysis of survival data with a class of weighting methods. We consider the following estimands in the two-sample context: average survival time, restricted average survival time, survival probability, survival quantile, and the marginal hazard ratio. We propose a unified analytic framework to obtain the point and variance estimators. The proposed methodology properly adjusts for the sampling variability in the estimated propensity scores. Extensive simulations show that the point and variance estimators possess desired finite sample properties and demonstrate better numerical performance than some existing weighting and matching methods commonly used in the literature. The proposed methodology is illustrated with data from a breast cancer study.

show abstract

Section: Point and Variance Estimationmentioning

confidence: 99%

On the propensity score weighting analysis with survival outcome: Estimands, estimation, and inference

Mao

Yang

et al. 2018

Statistics in Medicine

View full text Add to dashboard Cite

show abstract

“…In the light of the result that the standard bootstrap is inconsistent for some matching algorithms, recent studies propose modied bootstrap procedures that are consistent even for non-smooth (pair or one-to-many) matching estimators with continuous covariates. For instance, Otsu and Rai (2015) introduce and prove the validity of a weighted bootstrap algorithm for particular classes of pair matching estimators that, however, do not include propensity score matching. Bodory, Camponovo, Huber, and Lechner (2016) generalize the approach of Otsu and Rai (2015) by introducing a wild bootstrap procedure that can also be applied to propensity score matching estimators.…”

Section: Introductionmentioning

confidence: 99%

“…For instance, Otsu and Rai (2015) introduce and prove the validity of a weighted bootstrap algorithm for particular classes of pair matching estimators that, however, do not include propensity score matching. Bodory, Camponovo, Huber, and Lechner (2016) generalize the approach of Otsu and Rai (2015) by introducing a wild bootstrap procedure that can also be applied to propensity score matching estimators. Unlike the standard bootstrap, this wild bootstrap algorithm does not construct bootstrap samples by randomly selecting with replacement from the original sample.…”

Section: Introductionmentioning

confidence: 99%

The Finite Sample Performance of Inference Methods for Propensity Score Matching and Weighting Estimators

Bodory

Camponovo

Huber

et al. 2018

Journal of Business & Economic Statistics

View full text Add to dashboard Cite

This paper investigates the nite sample properties of a range of inference methods for propensity score-based matching and weighting estimators frequently applied to evaluate the average treatment eect on the treated. We analyse both asymptotic approximations and bootstrap methods for computing variances and condence intervals in our simulation designs, which are based on German register data and U.S. survey data. We vary the design w.r.t. treatment selectivity, eect heterogeneity, share of treated, and sample size. The results suggest that in general, theoretically justied bootstrap procedures (i.e. wild bootstrapping for pair matching and standard bootstrapping for`smoother' treatment eect estimators) dominate the asymptotic approximations in terms of coverage rates for both matching and weighting estimators. Most ndings are robust across simulation designs and estimators.

show abstract

“…This is mainly because the bootstrap based on approach (a) cannot preserve the distribution of the numbers of times that the units are used as matches. As a remedy, Otsu and Rai (2016) propose to construct the bootstrap counterparts by resampling based on approach (b) for the matching estimator.…”

Section: Bootstrap Variance Estimationmentioning

confidence: 99%

Combining Multiple Observational Data Sources to Estimate Causal Effects

Yang

Ding

2019

Journal of the American Statistical Association

View full text Add to dashboard Cite

The era of big data has witnessed an increasing availability of multiple data sources for statistical analyses. We consider estimation of causal effects combining big main data with unmeasured confounders and smaller validation data with supplementary information on these confounders. Under the unconfoundedness assumption with completely observed confounders, the smaller validation data allow for constructing consistent estimators for causal effects, but the big main data can only give error-prone estimators in general. However, by leveraging the information in the big main data in a principled way, we can improve the estimation efficiencies yet preserve the consistencies of the initial estimators based solely on the validation data. Our framework applies to asymptotically normal estimators, including the commonlyused regression imputation, weighting, and matching estimators, and does not require a correct specification of the model relating the unmeasured confounders to the observed variables. We also propose appropriate bootstrap procedures, which makes our method straightforward to implement using software routines for existing estimators.

show abstract

Bootstrap Inference of Matching Estimators for Average Treatment Effects

Cited by 55 publications

References 26 publications

On the propensity score weighting analysis with survival outcome: Estimands, estimation, and inference

On the propensity score weighting analysis with survival outcome: Estimands, estimation, and inference

The Finite Sample Performance of Inference Methods for Propensity Score Matching and Weighting Estimators

Combining Multiple Observational Data Sources to Estimate Causal Effects

Contact Info

Product

Resources

About