From: elodie on
Hi everyone,

I have been working on selecting a small number of controls from a
bigger pool of control subjects, in a study where there are 2
treatment groups. My objective is to select about 20 controls so that
the 2 treatment groups and the control groups do not differ
statistically on age and another variable.

I have found a SAS codefile which will match controls to cases, but
only when there is one treatment group. So all I could think of is
matching 10 controls to the first treatment group and the remaining 10
controls to the second treatment groups. This is not great, but my
resulting three groups turn out to not be statistically significant on
age and other demographics, which is a good start.

I would appreciate it if somebody had a better method for matching
when there are several treatment groups. Thanks in advance.