Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!umich!samsung!brutus.cs.uiuc.edu!psuvax1!psuvm!FORDMURH!AOWALSH
From: AOWALSH@FORDMURH.BITNET
Newsgroups: bit.listserv.sas-l
Subject: duplicates in sampling with replacement
Message-ID: <SAS-L%90021208010372@MARIST.BITNET>
Date: 12 Feb 90 12:57:00 GMT
Sender: "SAS(r) Discussion" <SAS-L@OHSTVMA>
Reply-To: AOWALSH@FORDMURH
Lines: 10
Approved: NETNEWS@PSUVM Gateway
Original_To:  BITNET%"sas-l@marist"

When bootstrapping the correlation coefficient with small n's it is possible
to get samples containing only 1 or 2 different observations. IN the code below
taken from the SAS Applications Guide,how can the above cases be detected ?
Any suggestions would be appreciated. I am using SAS 5.18 under vms 5.2 without
IML.            do i=1 to n;
                iobs=int(ranuni(0)*N+1);
                set sample point=iobs nobs=N;
                output;
                end;
                stop;