Date:         Tue, 27 Apr 2004 08:25:15 -0500
Reply-To:     "Peck, Jon" <peck@spss.com>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         "Peck, Jon" <peck@spss.com>
Subject:      Re: 0/missing = 0
Comments: To: Sally Zitzer <sallyz@u.washington.edu>
Content-Type: text/plain; charset="us-ascii"

You can write this type of expression a little more directly and transparently with a style like this.  The first term gives the normal result, and the second converts it to sysmis when the defined conditions occur.


compute newvar = vara/(varb+varc) + $sysmis*missing(varb+varc).

Sample results:

vara    varb    varc    newvar
1.00    2.00    3.00    .20
1.00    .       .       .
.00     .       .       .
.00     1.00    2.00    .00
.       .       .       .
1.00    .00     .00     .
.       .00     .00     .

Regards,
Jon Peck


-----Original Message-----
From: Sally Zitzer [mailto:sallyz@u.washington.edu]
Sent: Monday, April 26, 2004 5:11 PM
To: SPSSX-L@LISTSERV.UGA.EDU
Subject: [SPSSX-L] 0/missing = 0


Hi,

  It turns out that 0/missing = 0 by definition. (also 0*missing = 0, this
is shown in the syntax manual)

  I have a user who wants 0/missing = missing.  I'm wondering if anyone
knows of better way than the following --

His sample equation is --

compute newvar = (vara/(varb+varc)).
execute.

   If varb or varc are missing, and vara = 0, he wants newvar = missing
   But SPSS sets newvar = 0

What works for this specific example is --

compute newvar = (vara/(varb+varc)).
if (missing(varb+varc) and vara=0) newvar = $sysmis.
execute.

   newvar is first computed to be 0, but then recomputed to be missing

For more complicated equations it could get really messy.

   Thanks,

     Sally in Seattle

Sally Zitzer, C&C, Senior Computing Specialist, Statistical Software
sallyz@cac.washington.edu or sallyz@u.washington.edu  (206)-543-8515