2015
DOI: 10.56645/jmde.v11i25.431
|View full text |Cite
|
Sign up to set email alerts
|

Propensity Scores: A Practical Introduction Using R

Abstract: Background: This paper provides an introduction to propensity scores for evaluation practitioners.  Purpose: The purpose of this paper is to provide the reader with a conceptual and practical introduction to propensity scores, matching using propensity scores, and its implementation using statistical R program/software. Setting: Not applicable Intervention: Not applicable Research Design: Not applicable   Data Collection and Analysis: Not applicable Findings: In this demonstration paper, we describe the contex… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
2
1

Citation Types

0
18
0

Year Published

2018
2018
2023
2023

Publication Types

Select...
7

Relationship

0
7

Authors

Journals

citations
Cited by 94 publications
(18 citation statements)
references
References 29 publications
0
18
0
Order By: Relevance
“…28 Specifically, the propensity score matching implemented in this study reduces the dependence of causal inferences on statistical modeling assumptions, which may not always be justified in an observational study. 29 However, propensity scores are sensitive to the choice of observed covariates and modeling techniques, both of which are arbitrary. 30 In addition, the use of matching has reduced the number of patients by approximately 25% in both categories, which may lead to bias.…”
Section: Discussionmentioning
confidence: 99%
“…28 Specifically, the propensity score matching implemented in this study reduces the dependence of causal inferences on statistical modeling assumptions, which may not always be justified in an observational study. 29 However, propensity scores are sensitive to the choice of observed covariates and modeling techniques, both of which are arbitrary. 30 In addition, the use of matching has reduced the number of patients by approximately 25% in both categories, which may lead to bias.…”
Section: Discussionmentioning
confidence: 99%
“…Most of the Greedy matching methods such as nearest neighbor, reduces sample size to half, whereas the IPTW and optimal full matching retain the entire dataset, which was valuable for our study where race and ethnicity samples were smaller. We used the Olmos & Bovindassmy procedure (2015) 25 to perform PSW in R. We first created a selection model to estimate the propensity scores. The selection model, a logistic regression model, was used to estimate the selection bias on the model, where we included demographic variables that are often associated with the development of incident cognitive impairment.…”
Section: Methodsmentioning
confidence: 99%
“…Median OS and TTNTD were estimated via the Kaplan-Meier method 8 and compared using log-rank tests 9 . To account for differences between treatment sequencing groups, a Cox proportional hazards model 10 stratified by propensity score deciles 11,12 was used for the primary comparative effectiveness analyses. Propensity scores for treatment sequence arms were constructed based on random forest out-of-bag predictions 13 for all patients using baseline ECOG PS, smoking status, age at metastatic diagnosis, sex, albumin, and CA 19-9.…”
Section: Methodsmentioning
confidence: 99%