Hello everyone,
I am currently doing a robustness check after my main analysis which used a PSM-OLS approach.
I am now using Entropy Balancing and had a question about the weights.
I used the 'ebalance' command in Stata and then ran my regression. Initially, I used [aw=_webal] for the OLS model and [pw_webal] for the probit model (as one of my outcome vars is binary).
However, I'm not sure which is more appropriate, using (1) the analytical weight or (2) the probability weight. The results don't change hugely either way, but I would like to be sure of the reasoning and justification.
I'd be grateful if you could share your thoughts
Best regards,
Gráinne
I am currently doing a robustness check after my main analysis which used a PSM-OLS approach.
I am now using Entropy Balancing and had a question about the weights.
I used the 'ebalance' command in Stata and then ran my regression. Initially, I used [aw=_webal] for the OLS model and [pw_webal] for the probit model (as one of my outcome vars is binary).
However, I'm not sure which is more appropriate, using (1) the analytical weight or (2) the probability weight. The results don't change hugely either way, but I would like to be sure of the reasoning and justification.
I'd be grateful if you could share your thoughts
Best regards,
Gráinne
Comment