Date:         Tue, 15 May 2012 12:17:20 -0400
Reply-To:     Rick Wicklin <Rick.Wicklin@SAS.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Rick Wicklin <Rick.Wicklin@SAS.COM>
Subject:      Re: SAS random record selection

I echo Dan's suggestion to read David Cassell's paper.
If you decide you need columns, try this for an outline. I'll use a data
set with obs 1-N:

data a;
do i = 1 to 20; output; end;
run;

proc surveyselect data=a
     method=urs sampsize=5 reps=10 outhits outall
     out=b;
run;

Note the use of the OUTHITS option, as well as the OUTALL option suggested
by Joe.  The OUTHITS option creates a frequency variable named NumberHits,
which will probably be more useful to you than an indicator variable,
since an obs can be selected multiple times for some METHOD= choices.

The data set B is in long format. You can use PROC TRANSPOSE or other SAS
tricks to convert it to a wide format. I'll let others advise you on that,

Rick Wicklin