Abstract
In casual-effect relationship research, similarity of groups being compared in terms of covariates or patient/disease characteristics is critical to ensure fairness of the comparison and unbiasedness for the findings. When dissimilarity is suspected, one can either adjust for imbalance or match the groups according to certain important covariates or characteristics. Regression analysis is commonly used to adjust the imbalance and matching techniques are usually used to match subjects between groups. Diamond and Sekhon [2] proposed a genetic matching algorithm to maximize the covariate balance. We describe the theory and conduct a simulation study to compare the relative performance of propensity score matching. Mahalanobis matching, and Genetic matching. Generally Genetic matching achieves better covariate balance and produces more stable and unbiased treatment effect estimates. We also apply Genetic matching to a clinical study to investigate the treatment effects on rheumatoid arthritis.
Original language | American English |
---|---|
Title of host publication | Proceedings of the International Conference on Bioinformatics and Computational Biology |
State | Published - Jan 1 2011 |
Keywords
- Approximation
- Mahalanobis Matching
- Propensity Score
- Randomized Controlled Clinical Trials
DC Disciplines
- Biostatistics
- Community Health
- Public Health