Covariate balance, current implementation, future developments

[Naeemkh]

Achieving an acceptable covariate balance can be challenging for small datasets or data with extreme exposure values. Please add your thoughts about your experience with the current implementation and suggestions for improvement. Feel free to add any references that you think can be helpful.

[Naeemkh]

Possibility of matching on a subset of covariates

According to Stuart (2010), researchers should be liberal in choosing covariates. However, this might not be possible in small data samples because it can significantly increase variance.
To have robust analyses and to make sure that the ignobility assumption is satisfied, one can conduct the matching process on a subset of covariates (those we know have an effect on outcome based on domain knowledge) but conduct the covariate balance on all covariates.

This can be a valuable feature to add to the package. Please let me know your thoughts.

Stuart, E. A. (2010). Matching methods for causal inference: A review and a look forward. Statistical science: a review journal of the Institute of Mathematical Statistics, 25(1), 1.