Date:         Mon, 15 Nov 1999 21:15:36 GMT
Reply-To:     "Daniel J. Nordlund" <dnordlund@AOL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Daniel J. Nordlund" <dnordlund@AOL.COM>
Organization: AOL http://www.aol.com
Subject:      Re: Converting to events/trials data format?

Jin Kim <kimjin43@chollian.net> wrote:
>Dear SAS-L
>
>As a SAS novice, I want to know the method how I convert the following
>ordinary data to 'events/trials' data format, which can be used in PROC
>LOGISTIC or GENMOD.
>
>(Of course, I know that above PROCs are possible with ordinary data.
>This question is of educational purpose.)
>
>In the following example data, 'var1-3' are categorical explanatory
>variables, and 'result' is a binary response variable.
>
>        var1    var2    var3    result
>obs1    1        1        2        0
>obs2    2        2        2        1
>obs3    3        1        1        0
>obs4    2        2        1        1
>obs5    1        1        2        1
>  .     .        .        .        .
>  .     .        .        .        .
>  .     .        .        .        .
>
>How can I convert the ordinary data like above to 'events/trials'
>format?
>
>Thank you.
>
>Jin Kim
>

You can use PROC SUMMARY to accomplish what you want.  In the example,
'yourdata' is your original file.  The output file, which I called 'event',
will contain your classification variables and 2 new variables called events
and trials.

proc summary data=yourdata nway;
  class var1 var2 var3;
  var result;
  output out=event(rename=(_freq_=trials) drop=_type_)
            sum=events;
run;

Hope this helps,

Dan Nordlund