0
$\begingroup$

I wish to estimate a treatment effect using Kernel Matching, but I'm confused about the process. From a high level, Is A or B correct? Or are both considered Kernel matching?

A

(1) Estimate propensity scores

(2) Use kernel that uses the propensity scores to create weight matrix

(3) Calculate treatment effect using weights from matrix

B

(1) Compute distance between all observations (could be mahalanobis or euclidean), giving you a matrix of distances between observations

(2) Put distances through a kernel function, giving you a matrix of weights

(3) Calculate treatment effect using weights from matrix

I wish to code this up in R because there are no kernel matching functions that I'm aware of. Thanks for your help.

$\endgroup$
  • $\begingroup$ I think package KBAL supports kernel balancing. However, I am afraid that the package was removed from R repository. $\endgroup$ – James Park Mar 15 at 8:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.