Hello,
I am conducting propensity score matching using psmatch2 in Stata, and have a question about how Rubin's B and R are calculated. I have read the Stata help file for psmatch2, and also read the Rubin (2001) paper where this definition is given: “B is the number of standard deviations between the means of the x distributions in the two groups, and R is the ratio of the variances of x in the two groups.“
The reason why I wanted to hand calculate Rubin's B and R is that I am trying to exact match on a few important matching variables in addition to the normal list of matching variables by following this example:
logit union collgrad age tenure not_smsa c_city south nev_mar
predict pscore if e(sample), pr
gen pscore2=year*10+pscore
psmatch2 union, pscore(pscore2) outcome(ln_wage) caliper(0.5)
The pstest output for Rubin's B & R does not seem to make much sense to me, and I was not able to reproduce the numbers myself. I would appreciate it if somebody can help me:
1. find the exact formula for Rubin's B & R
2. let me know whether I can use the same formula to calculate a meaningful Rubin's B & R using pscore2 constructed above.
Thank you very much for you help!
I am conducting propensity score matching using psmatch2 in Stata, and have a question about how Rubin's B and R are calculated. I have read the Stata help file for psmatch2, and also read the Rubin (2001) paper where this definition is given: “B is the number of standard deviations between the means of the x distributions in the two groups, and R is the ratio of the variances of x in the two groups.“
The reason why I wanted to hand calculate Rubin's B and R is that I am trying to exact match on a few important matching variables in addition to the normal list of matching variables by following this example:
logit union collgrad age tenure not_smsa c_city south nev_mar
predict pscore if e(sample), pr
gen pscore2=year*10+pscore
psmatch2 union, pscore(pscore2) outcome(ln_wage) caliper(0.5)
The pstest output for Rubin's B & R does not seem to make much sense to me, and I was not able to reproduce the numbers myself. I would appreciate it if somebody can help me:
1. find the exact formula for Rubin's B & R
2. let me know whether I can use the same formula to calculate a meaningful Rubin's B & R using pscore2 constructed above.
Thank you very much for you help!
