John Baj <jbaj@NIU.EDU> wrote:

>> What I need to do is to build a comparision group >> for the welfare population from the non welfare >> population (matched pairs). For this comparison >> group, I am trying to match each job spell of welfare >> recipients to a job spell of a non recipient who >> began working for same employer at the same >> time and for approximately the same wage (using >> the earnings variable as a proxy) as the recipient.

The BioStats folks at Mayo Clinic have kindly made a macro available to the public that I believe will do what you describe (a post-observational assignment of control cases) ....check out "Match" at http://www.mayo.edu/hsr/sasmac.html

Match

Computerized matching of cases to potential controls. The purpose of this macro is to match 1 or more controls (from a total of M) for each of N cases. The controls may be matched to the cases by one or more factors(X's). Jon Kosanke & Erik Bergstralh [28/OCT/1999]

HTH

Douglas Dame Shands HealthCare Gainesville FL USA