Term of Award
Doctor of Public Health in Biostatistics (Dr.P.H.)
Document Type and Release Option
Dissertation (restricted to Georgia Southern)
Copyright Statement / License for Reuse
This work is licensed under a Creative Commons Attribution 4.0 License.
College of Public Health
Committee Member 1
Committee Member 2
Introduction: Matching could be defined as “any method that aims to equate (or “balance”) the distribution of covariates in the treated and control groups”. It could entail 1:1 matching, weighting or sub classification (Stuart, 2010).
Objectives: The objectives of this study are: (1) To compare several matching methods with the aim of choosing the best method for matching groups in causal analysis. (2) To apply the results of the study to real world data to estimate causal effect.
Methods: For this study a dataset S was simulated which is representative of data on hypertensive patients enrolled in a randomized, double blind, placebo (P) controlled clinical trial of a dose of a drug D. The effect measure of interest is the mean difference. Matchit was used for the simulations study. Matchit works in conjunction with the R programming language statistical software.
Results: Looking at the different matching methods side by side from both the one sample simulation and the simulation with 500 different samples, the Genetic Matching method appeared to be the best matching method as it is the matching method that produces dataset that satisfied all the conditions for normality as specified by the normality plots of the histogram, the Shapiro-Wilks’s test and the quantile-quantile plots.
Conclusion: The study showed that the Genetic Matching Method was better than the Nearest Neighbor, Optimal, Coarsened Exact and the Mahalanobis Matching Methods
Okoro, Uche, "Assessment of Matching Methods for Causal Analysis" (2019). Electronic Theses and Dissertations. 2023.
Research Data and Supplementary Material