observations before and after propensity score matching

[minji kwak]

Hi, all.

I am doing a propensity score matching using observation n=7,089.
I have used psmatch2 and posttest using mahalaoibis method. But, I realized that even for the unmatched group, total observation used for analysis was only 3,949.

How can I explain it?

Thank you.

[Amin Sofla]

I think that some of the observations are removed because they were not properly matched. You can e.g., modify the caliper, and (or) change the 1:1 matching ratio. Alternatively, you can check other balancing methods such as Entropy Balancing.

ssc install ebalance  
help ebalance 

[minji kwak]

I will try. Thanks!