Replicable data with calipmatch

[leireberrocal6]

Hi! I just wanted to know if there is a way to execute calipmatch and get everytime the same matching for the same data, that is, having guarantee to be able to replicate the data. I have just tried to get the same pairs, but as the helps says and the code shows, there is a randomization in the data order at some point.

[taliabenheim]

Hi! I was running into this same problem, but figured out that simply setting the exact order of the dataset and the seed before running calipmatch seems to make it reproducible:

sort id //make sure this is a unique sort order
set seed 1234
calipmatch...

I believe the non-replication issue happens because calipmatch generates a random number variable and then sorts the data on that variable (lines 61-64 in calipmatch.ado). This step reshuffles the dataset and will only be reproducible if both: (a) the data start out sorted in the same exact order, and (b) the random number generator starts at the same seed. Importantly, the first line requires a totally unique sort order. If there are any duplicates in the sort variables, the order won't be fixed across different runs.