Creating Matched Pairs

[lalaktgrau]

Hi everyone,

I'm trying to create a dataset where I match subjects by age (+/- 1 yr), gender, and county. I would want the final dataset to have individual observations for each subject, with an additional matched ID variable that identifies each group.

How can I go about it??

When I do it using SQL, I end up with one observation with all the ids of everyone that has been matched.

proc sql;
create table matches_age as
select
a.id as v_id,
b.id as nv_id,
a.birth_yr as v_birth_yr,
b.birth_yr as nv_birth_yr,
a.county_2013 as v_county_2013,
b.county_2013 as nv_county_2013,
a.gender_cd as v_gender_cd,
b.gender_cd as nv_gender_cd,
b.rand_num as rand_num
from v a, nv2 b
where ((a.birth_yr between b.birth_yr_high and b.birth_yr_low)
and a.county_2013=b.county_2013 and a.gender_cd=b.gender_cd);
quit;

[audrey]

Hi,

Did you try the DQMATCH proc?

http://go.documentation.sas.com/?docsetId=dqclref&docsetVersion=9.4&docsetTarget=n1xjvmh658gp9mn1rl9...

Thanks,
Audrey

[lalaktgrau]

From what I'm reading, it seems like it's used to identify multiple observations of the same subject. I would want to match cases and controls with similar characteristics and assign each group of cases/controls a unique value.

[Reeza]

What version of SAS do you have?

PROC PSMATCH is for probabilistic linkages and/or the Mayo Clinic matching macros?

PROC PSMATCH

https://support.sas.com/rnd/app/stat/procedures/psmatch.html

Mayo Clinic macros

http://www.mayo.edu/research/departments-divisions/department-health-sciences-research/division-biom...

[lalaktgrau]

I have SAS 9.4 TS Level 1M2 X64_7PRO.

I don't want to do propensity score as in PSMATCH. 

I just want to select observations that have the same birth_yr, gender, and county and assign a unique number to it.

[Reeza]

Then you want this most likely.

https://stats.idre.ucla.edu/sas/faq/how-can-i-create-an-enumeration-variable-by-groups/