"kmatch": How to find matched individuals in the control group in "kmatch"

[Hossein Kavand]

Hello,
I am using "kmatch" with the different matching algorithm. However, I need to find those individuals in the control group who are matched with each person in the treatment group. I could not find any option in "kmatch" to show me which individuals have been matched to every person in the treatment group.



I have the two following questions about "kmatch":

1) In "kmatch", is there any way to find all individuals that have been matched with a person in the treatment group.


2) If "kmatch" is using kernel method, is there any way to find the weights assigned to each selected person in the control group as well as its reference in the treatment group? For example, if kernel method is using weights for 4 individuals who have been matched with a person in the treatment group, is there any way that I know who are these 4 individuals and their reference in the treatment group, and what is the assigned weights to them?

I really appreciate your help.
Thanks
Hossein

[Phil Bromiley]

You'll increase your chances of a useful answer by following the FAQ on asking questions – provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

You are using a user written routine. Stata now has a lot of treatment model estimators. You might find them easier to work with.

Without installing and running your routine, I can't tell for sure, but almost any program that matches treated individuals to non-treated must have some identifier for groups. If you have groups identified and you know which one is the treated, then you can identify the non-treated.

[eric_a_booth]

You are using a user written routine. Stata now has a lot of treatment model estimators. You might find them easier to work with.

Without installing and running your routine, I can't tell for sure, but almost any program that matches treated individuals to non-treated must have some identifier for groups. If you have groups identified and you know which one is the treated, then you can identify the non-treated.

In line with this advice, the example that Dimitriy V. Masterov provides on this stack exchange post is valuable for doing what you describe using -teffects- and a similar approach might be useful for -kmatch- https://stats.stackexchange.com/ques...see-my-matches

[Hossein Kavand]

Thanks.