Hi stata-users,

I haven't quite been able to figure out some of the variables psmatch2 produces. I have used
Code:
set seed 1234
generate id=uniform()
sort id
psmatch2 admit, pscore(pscore) caliper(0.0024762) noreplacement descending common logit
to produce a 1:1 match (as per the manual) but it doesn't quite give me a 1:1 match. After dropping
Code:
drop if _weight==.
I get 436,924 in treated and 418,311 in control when I have LOTS of patients in control in the original dataset. I am not sure why it doesn't select them. Does it have to do with the common support but then in a huge dataset like this one (n=35 million), it shouldn't be an issue.

Any advice?

PS. I am sort of stuck with psmatch2 because no other package can run through such a big dataset in minutes.
For those wondering what the weights look like:

psmatch2: |
weight of |
matched |
controls | Freq. Percent Cum.
------------+-----------------------------------
1 | 845,674 98.88 98.88
2 | 1,197 0.14 99.02
3 | 8,000 0.94 99.96
4 | 40 0.00 99.96
5 | 324 0.04 100.00
------------+-----------------------------------
Total | 855,235 100.00