topic Matched Subjects Design With SAS in Statistical Procedures

Matched Subjects Design With SAS

riccardo88 — Tue, 16 Oct 2018 13:14:54 GMT

Hello everybody!

I have to randomized the subjects for an Experiment and assigned them to a "Treatment" Group or a "Placebo" Group.

I would like to write a programm with SAS that find the best matches in a list of subjects and matches them based on different Parameters (such as Age and Sex).

For example having one dataset that Looks as follow:

ID Sex Age

01 M 23

02 M 37

03 F 29

04 M 25

05 F 30

....

The Programm should give me back as result that the best matches are:

Subj 01 -> Subj 04 (Sex M=M; Age 23=25)

Subj 03 -> Subj 05 (Sex F=F; Age 29=30)

and so on....

I looked in the Internet but I can´t find any tutorial about it. Can someone help me posting the link of a source where I can learn to do this or give me some tipps?

Thank you!

Re: Matched Subjects Design With SAS

Ksharp — Tue, 16 Oct 2018 13:55:14 GMT

data have;
input ID   Sex $  Age ;
cards;
01   M     23
02   M     37
03   F      29
04   M     25
05   F      30
;
run;
proc sql;
select a.*,b.id as _id
 from have as a,have as b
  where a.id ne b.id and a.sex=b.sex
   group by a.id
    having abs(a.age-b.age) =  min(abs(a.age-b.age));
  quit;

Re: Matched Subjects Design With SAS

cousineddie — Tue, 16 Oct 2018 16:32:41 GMT

Try the PSMATCH procedure documentation: https://support.sas.com/documentation/onlinedoc/stat/142/psmatch.pdf

One spot (of many) to look at specifically is Example 95.6: Mahalanobis Distance Matching. In the output statement of PSMATCH, include the MATCHID=_MatchID option to create a variable named _MatchID that identifies the matched sets of observations.

You'll need to include the column that designates each subject as either "test" or "control".

Re: Matched Subjects Design With SAS

riccardo88 — Wed, 17 Oct 2018 08:10:57 GMT

Works perfect! Thank you very much!!