Hi,
I'm fairly new to Stata and am using version 12 for Windows. Recently, I am just starting to implement DID with propensity score matching into practice, I have so many confusions and questions before applying matching.
1. I found most people are using command psmatch2, but some people using pscore. What is the difference among psmatch2, nnmatch and pscore?
2. I conclude the procedure as follows:
A.Estimate the probability of participation using a probit or logit model and create the predicted ‘propensity score’ to participate for each observation.
B.Limit the original sample to a new dataset which only include matched pairs.
C.DID
The procedure (matching -> newdata -> DID) is vague. If the dataset is an unbalanced panel, how to do the first step? matching year by year? I am confused about a method to do pairwise matching with unbalanced panel data in Stata. Could you offer me more details which should be taken care during this process.
As for my project, 2 dimensions for our DID are: inside zone vs. outside zone and before the year of implementation(2007) vs after 2007.
As some people told me to calculate ATT difference year by year, so it will have a DID estimate every year. How to explain them? Or only calculating ATT difference between "pre" and "post" year?
I have read some related materials including some video tutorials on YouTube, one of which told me that a "Propensity Score Matching" in combination with a Difference in Difference approach is to calculate the ATT on the differences in outcomes after and before. The method is to put the differences in outcomes after and before instead of outcome into the program command of matching procedure. Simply put, the process is: differencing -> matching -> calculating ATT. But, I'm still confused about it.
Hopefully I have described it clearly. The task is so pressing, so I was wondering if you could give me some advice. Looking forward to your reply~
Best Regards,
Min DAI
I'm fairly new to Stata and am using version 12 for Windows. Recently, I am just starting to implement DID with propensity score matching into practice, I have so many confusions and questions before applying matching.
1. I found most people are using command psmatch2, but some people using pscore. What is the difference among psmatch2, nnmatch and pscore?
2. I conclude the procedure as follows:
A.Estimate the probability of participation using a probit or logit model and create the predicted ‘propensity score’ to participate for each observation.
B.Limit the original sample to a new dataset which only include matched pairs.
C.DID
The procedure (matching -> newdata -> DID) is vague. If the dataset is an unbalanced panel, how to do the first step? matching year by year? I am confused about a method to do pairwise matching with unbalanced panel data in Stata. Could you offer me more details which should be taken care during this process.
As for my project, 2 dimensions for our DID are: inside zone vs. outside zone and before the year of implementation(2007) vs after 2007.
As some people told me to calculate ATT difference year by year, so it will have a DID estimate every year. How to explain them? Or only calculating ATT difference between "pre" and "post" year?
I have read some related materials including some video tutorials on YouTube, one of which told me that a "Propensity Score Matching" in combination with a Difference in Difference approach is to calculate the ATT on the differences in outcomes after and before. The method is to put the differences in outcomes after and before instead of outcome into the program command of matching procedure. Simply put, the process is: differencing -> matching -> calculating ATT. But, I'm still confused about it.
Hopefully I have described it clearly. The task is so pressing, so I was wondering if you could give me some advice. Looking forward to your reply~
Best Regards,
Min DAI
