Date:         Wed, 12 Jul 2000 19:50:18 -0500
Reply-To:     "John J. Shon" <pjshon@GSBPHD.UCHICAGO.EDU>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "John J. Shon" <pjshon@GSBPHD.UCHICAGO.EDU>
Subject:      proc univariate output
Content-Type: TEXT/PLAIN; charset=US-ASCII

my gut tells me this is a trivial question, but for the life of me, i
can't do it. the code:

proc univariate noprint data=original;
  var sales
output out=outt pctlpts=1 99
    pctlpre=sales_;
run;


the dataset OUTT now has the 1st and 99th percentile for the population's
sales. what i need to do is bring these 2 values (ie, the 1st and 99th
percentiles) into the original dataset, such that for each observation, i
can say something like:

if sales<sales_1 then delete;
if sales>sales_99 then delete;

can someone help? thanks.

~John Shon