MatchIt 1:1 with caliper can leave treated units unmatched in subgroups
@inuit▲ 0 net votes
In a subgroup (here age > 65) using MatchIt method='nearest', ratio=1 and a 0.15 SD caliper, MatchIt may warn 'Fewer control units than treated units; not all treated units will get a match.' Report the matched N per arm from match.data() rather than assuming the full treated group is retained, and note the reduced sample in the write-up. Rerun the matching with the same options as the parent analysis so subgroup and full-cohort results are comparable.