Hello Everyone,
I am trying to do nearest neighbour matching using the teffects nnmatch command in stata and estimate the average treatment effect on the treated. I first tried to do the matching on the full sample and obtained the following results.
Then, I reduced my sample to only those observations from the control group that are matched with some observation from the treatment group. I implemented the teffects nnmatch on this reduced sample, matching on the same set of variables and using the same conditions. I also confirmed that the matches produced with the reduced sample are the same as the matches with the full sample. However, the coefficient estimates produced from these two samples are different as can be seen below:
Could someone explain why I am getting these different results even when the matched observations are the same?
I am trying to do nearest neighbour matching using the teffects nnmatch command in stata and estimate the average treatment effect on the treated. I first tried to do the matching on the full sample and obtained the following results.
Then, I reduced my sample to only those observations from the control group that are matched with some observation from the treatment group. I implemented the teffects nnmatch on this reduced sample, matching on the same set of variables and using the same conditions. I also confirmed that the matches produced with the reduced sample are the same as the matches with the full sample. However, the coefficient estimates produced from these two samples are different as can be seen below:
Could someone explain why I am getting these different results even when the matched observations are the same?
