Larger calipers permit more possible matches between treated and control
groups, which can be better for creating matches with larger effective sample
sizes. The downside is that wide calipers may make the matching problem too big
for processor or memory constraints. maxCaliper
attempts to find a
caliper value, for a given vector of scores and a treatment indicator, that
will be possible given the maximum problem size constraints imposed by
fullmatch
and pairmatch
.
1  maxCaliper(scores, z, widths, structure = NULL, exact = TRUE)

scores 
A numeric vector of scores providing 1D position of units 
z 
Treatment indicator vector 
widths 
A vector of caliper widths to try, will be sorted largest to smallest. 
structure 
Optional factor variable that groups the scores, as would
be used by 
exact 
A logical indicating if the exact problem size should be
computed ( 
numeric The value of the largest caliper that creates a feasible
problem. If no such caliper exists in widths
, an error will be
generated.
Questions? Problems? Suggestions? Tweet to @rdrrHQ or email at ian@mutexlabs.com.
All documentation is copyright its authors; we didn't write any of that.