There are 2 major classes of matching methods, according to a content analysis by King and Nielsen (2018). They are:
- Equal Percent Bias Reducing (EPBR)
- Monotonic Imbalance Bounding (MIB)
Mahalanobis Distance Matching (MDM) is an example of EPBR and Coarsened Exact Matching (CEM) is an example of MIB. I have a rough idea of how those 2 classes might be defined based on their names and the example algorithms.
However, I would like a clear and compact definition of each, with a comparison of their advantages and disadvantages. An example of when you might prefer one class over the other would be helpful.
When I try to find this from online resources I mostly find only journal articles that are more narrowly focused on other topics rather than any educational comparison. One exception is Definition 3 of this Harvard thesis for EPBR.
