multiple imputation (-mi-) with the synthetic control method (-synth-)

[Tom Scott]

Has anyone combined multiple imputation with the synthetic control method (https://fmwww.bc.edu/repec/bocode/s/synth.html)? Is this possible? I'm using Stata 16. Thank you!

[Amber Bosma]

Hello Tom Scott, this is exactly what I am trying to do at the moment, did you find an answer to whether it is possible? Thank you very much

[Tom Scott]

Amber Bosma no I haven't. I haven't been successful figuring out how to use mi with estimation methods that aren't already included in the mi package.

[Amber Bosma]

Tom Scott thank you for your response. I will look into it and let you know if I find out, kind regards

[Robert Kolesar]

Good day, I am checking in to find out if anybody has found a way to use synth after multiple imputation. Thank you.

[Jared Greathouse]

Before I give my thoughts, I'm quite curious about why you would want to do this. Can any of you explain why this might be desirable? I know what MI does, but I've never even heard of it being used with SCM. What might you need MI for so badly such that you need it for SCM to work?

What's the context here?

[Tom Scott]

Jared Greathouse I don't think the motivations for using MI would differ by analytic method, would they? For my purpose, I'm aggregating crime data from U.S. law enforcement agencies to the state level. The relationship between the victim and offender (e.g., stranger; ex-spouse; friend) in a crime is missing quite a bit and I was hoping to use MI to impute the missing relationships before summing crime counts to the state level to conduct a SCM.

[Jared Greathouse]

Tom Scott what's the intervention here? And no, they wouldn't differ by method necessarily, I've just never heard those two used together is all