Equating Nonequivalent Groups Using Propensity Scores: Model Misspecification and Sensitivity Analysis

2021 ◽  
pp. 187-197
Author(s):  
Gabriel Wallin ◽  
Marie Wiberg
Keyword(s):  
Sensitivity Analysis ◽  
Propensity Scores ◽  
Model Misspecification ◽  
Nonequivalent Groups

scholarly journals Causal Subclassification Tree Algorithm and Robust Causal Effect Estimation via Subclassification

2020 ◽  
Vol 10 (1) ◽  
pp. 40
Author(s):  
Tomoshige Nakamura ◽  
Mihoko Minami
Keyword(s):  
Propensity Score ◽  
Propensity Scores ◽  
Causal Effect ◽  
Model Misspecification ◽  
Robust Estimator ◽  
Propensity Score Model ◽  
Confounding Variables ◽  
Score Model ◽  
Effect Estimation ◽  
Weighting Methods

In observational studies, the existence of confounding variables should be attended to, and propensity score weighting methods are often used to eliminate their e ects. Although many causal estimators have been proposed based on propensity scores, these estimators generally assume that the propensity scores are properly estimated. However, researchers have found that even a slight misspecification of the propensity score model can result in a bias of estimated treatment effects. Model misspecification problems may occur in practice, and hence, using a robust estimator for causal effect is recommended. One such estimator is a subclassification estimator. Wang, Zhang, Richardson, & Zhou (2020) presented the conditions necessary for subclassification estimators to have $\sqrt{N}$-consistency and to be asymptotically well-defined and suggested an idea how to construct subclasses.

Modern Algorithms for Matching in Observational Studies

2020 ◽  
Vol 7 (1) ◽  
pp. 143-176 ◽  
Author(s):  
Paul R. Rosenbaum
Keyword(s):  
Sensitivity Analysis ◽  
Propensity Score ◽  
Observational Studies ◽  
Propensity Scores ◽  
Variable Number ◽  
Exact Matching ◽  
Covariate Balance ◽  
Software Packages ◽  
Effective Use ◽  
Entire Number

Using a small example as an illustration, this article reviews multivariate matching from the perspective of a working scientist who wishes to make effective use of available methods. The several goals of multivariate matching are discussed. Matching tools are reviewed, including propensity scores, covariate distances, fine balance, and related methods such as near-fine and refined balance, exact and near-exact matching, tactics addressing missing covariate values, the entire number, and checks of covariate balance. Matching structures are described, such as matching with a variable number of controls, full matching, subset matching and risk-set matching. Software packages in R are described. A brief review is given of the theory underlying propensity scores and the associated sensitivity analysis concerning an unobserved covariate omitted from the propensity score.

scholarly journals The effects of traffic-related air pollutants on chronic obstructive pulmonary disease in the community-based general population

2021 ◽  
Vol 22 (1) ◽  
Author(s):  
Hui-Tsung Hsu ◽  
Chih-Da Wu ◽  
Mu-Chi Chung ◽  
Te-Chun Shen ◽  
Ting-Ju Lai ◽  
...  
Keyword(s):  
Chronic Obstructive Pulmonary Disease ◽  
Sensitivity Analysis ◽  
Pulmonary Disease ◽  
Air Pollutants ◽  
Propensity Scores ◽  
Traffic Pollution ◽  
Chronic Obstructive ◽  
Obstructive Pulmonary Disease ◽  
Frequency Matching ◽  
The Impact

Abstract Background Previous studies have shown inconsistent results regarding the impact of traffic pollution on the prevalence of chronic obstructive pulmonary disease (COPD). Therefore, using frequency matching and propensity scores, we explored the association between traffic pollution and COPD in a cohort of 8284 residents in a major agricultural county in Taiwan. Methods All subjects completed a structured questionnaire interview and health checkups. Subjects with COPD were identified using Taiwan National Health Insurance Research Databases. A hybrid kriging/LUR model was used to identify levels of traffic-related air pollutants (PM2.5 and O3). Multiple logistic regression models were used to calculate the prevalence ratios (PRs) of COPD and evaluate the role played by traffic-related indices between air pollutants and COPD. The distributed lag nonlinear model was applied in the analysis; we excluded current or ever smokers to perform the sensitivity analysis. Results Increased PRs of COPD per SD increment of PM2.5 were 1.10 (95% CI 1.05–1.15) and 1.25 (95% CI 1.13–1.40) in the population with age and sex matching as well as propensity-score matching, respectively. The results of the sensitivity analysis were similar between the single and two pollutant models. PM2.5 concentrations were significantly associated with traffic flow including sedans, buses, and trucks (p < 0.01). The higher road area and the higher PM2.5 concentrations near the subject’s residence correlated with a greater risk of developing COPD (p for interaction < 0.01). Conclusions Our results suggest that long-term exposure to traffic-related air pollution may be positively associated with the prevalence of COPD. Graphical abstract

Advances in Quasi-experimental Methods

2018 ◽  
Author(s):  
Peter Miksza ◽  
Kenneth Elpus
Keyword(s):  
Education Research ◽  
Music Education ◽  
Experimental Research ◽  
Propensity Scores ◽  
Regression Discontinuity ◽  
Randomized Experiment ◽  
Random Assignment ◽  
Nonequivalent Groups ◽  
Regression Discontinuity Designs ◽  
Quasi Experimental

Although the primacy of the randomized experiment is often thought of as sacrosanct, in education research—and in music education research in particular—random assignment is often unachievable, unethical, or undesirable for one or more of many potential reasons. Methodologists have developed quasi-experimental research methods that attempt to achieve results that approximate the highly trustworthy results obtained from a randomized experiment. This chapter details two newer methods for quasi-experimental research that have become quite common in the broader field of education but have not yet become frequently employed within the field of music education research. These two methods are regression discontinuity designs and the use of propensity scores for the equating of nonequivalent groups.

scholarly journals Kernel Equating Using Propensity Scores for Nonequivalent Groups

2019 ◽  
Vol 44 (4) ◽  
pp. 390-414 ◽  
Author(s):  
Gabriel Wallin ◽  
Marie Wiberg
Keyword(s):  
Propensity Score ◽  
Test Scores ◽  
Propensity Scores ◽  
Mean Squared Error ◽  
Squared Error ◽  
Admissions Test ◽  
Nonequivalent Groups ◽  
Kernel Equating ◽  
Common Items ◽  
Practical Implications

When equating two test forms, the equated scores will be biased if the test groups differ in ability. To adjust for the ability imbalance between nonequivalent groups, a set of common items is often used. When no common items are available, it has been suggested to use covariates correlated with the test scores instead. In this article, we reduce the covariates to a propensity score and equate the test forms with respect to this score. The propensity score is incorporated within the kernel equating framework using poststratification and chained equating. The methods are evaluated using real college admissions test data and through a simulation study. The results show that propensity scores give an increased equating precision in comparison with the equivalent groups design and a smaller mean squared error than by using the covariates directly. Practical implications are also discussed.

Propensity score-based methods for causal inference in observational studies with non-binary treatments

2020 ◽  
Vol 29 (3) ◽  
pp. 709-727
Author(s):  
Shandong Zhao ◽  
David A van Dyk ◽  
Kosuke Imai
Keyword(s):  
Propensity Score ◽  
Response Function ◽  
Dose Response ◽  
Propensity Scores ◽  
Model Misspecification ◽  
Average Treatment Effect ◽  
Treatment Regimes ◽  
Propensity Score Methods ◽  
Empirical Performance ◽  
Dose Response Function

Propensity score methods are a part of the standard toolkit for applied researchers who wish to ascertain causal effects from observational data. While they were originally developed for binary treatments, several researchers have proposed generalizations of the propensity score methodology for non-binary treatment regimes. Such extensions have widened the applicability of propensity score methods and are indeed becoming increasingly popular themselves. In this article, we closely examine two methods that generalize propensity scores in this direction, namely, the propensity function (PF), and the generalized propensity score (GPS), along with two extensions of the GPS that aim to improve its robustness. We compare the assumptions, theoretical properties, and empirical performance of these methods. On a theoretical level, the GPS and its extensions are advantageous in that they are designed to estimate the full dose response function rather than the average treatment effect that is estimated with the PF. We compare GPS with a new PF method, both of which estimate the dose response function. We illustrate our findings and proposals through simulation studies, including one based on an empirical study about the effect of smoking on healthcare costs. While our proposed PF-based estimator preforms well, we generally advise caution in that all available methods can be biased by model misspecification and extrapolation.

Comparative effectiveness of adjuvant chemotherapy (AC) versus observation in patients with ≥ pT3 and/or pN+ bladder cancer (BCa).

2015 ◽  
Vol 33 (7_suppl) ◽  
pp. 292-292 ◽  
Author(s):  
Matt D. Galsky ◽  
Kristian Stensland ◽  
Erin L. Moshier ◽  
John Sfakianos ◽  
Russell Bailey McBride ◽  
...  
Keyword(s):  
Sensitivity Analysis ◽  
Overall Survival ◽  
Propensity Score ◽  
Survival Data ◽  
Comparative Effectiveness ◽  
Propensity Scores ◽  
Proportional Hazards ◽  
Private Insurance ◽  
Cox Proportional Hazards ◽  
The Impact

292 Background: Though Level I evidence supports the use of neoadjuvant chemotherapy (NAC) in BCa, the data supporting AC has been mixed. Experience suggests an adequately powered trial to definitively assess the role of AC is unlikely to be completed. Alternative approaches to evidence development are necessary. Methods: Patients who underwent cystectomy for ≥pT3 and/or pN+ M0 BCa were identified from the National Cancer Database. Patients who received NAC and/or diagnosed after 2006 (most recent year with survival data) were excluded. Logistic regression was used to calculate propensity scores representing the predicted probabilities of assignment to AC versus observation based on: age, demographics, year of diagnosis, pT stage, margin status, lymph node density, distance to hospital, hospital cystectomy volume, and hospital type and location. A propensity score-matched cohort of AC versus observation (1:2) patients was created. Stratified Cox proportional hazards regression was used to estimate the hazard ratio (HR) for overall survival for the matched sample while propensity score adjusted and inverse probability of treatment weighted proportional hazards models were used to estimate adjusted HR for the full sample. A sensitivity analysis examined the impact of comorbidities. Results: 3,294 patients undergoing cystectomy alone and 937 patients undergoing cystectomy plus AC met inclusion criteria.Patients receiving AC were significantly more likely to: be younger, have more lymph nodes examined and involved, have higher pT stage, have positive margins, reside in the Northeast and closer to the hospital, and have private insurance. AC was associated with improved overall survival (Table). The results were robust to sensitivity analysis for comorbidities. Conclusions: AC was associated with improved survival in patients with ≥pT3 and/or pN+ BCa in this large comparative effectiveness analysis. [Table: see text]

Sign in / Sign up

Export Citation Format

Share Document