Date:         Thu, 3 Feb 2000 17:19:01 +0100
Reply-To:     NASSAR NAJI <nassar@CYBERCABLE.FR>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         NASSAR NAJI <nassar@CYBERCABLE.FR>
In-Reply-To:  <20000203150431.YKUQ1148.mta05.onebox.com@onebox.com>
Content-Type: text/plain; charset="iso-8859-1"

Kevin,

I hope this will help:

- If only complaint 1 interests you, this syntaxe should give you what
needed
sort cases by casenum(A) complaint (A).
compute cplt1=(complaint=1).
if (casenum=lag(casenum)) cplt1=lag(cplt1).
execute.
You can encounter some problems in case of missing values or 0 for complaint
variable (in such case record those values to higher value, for ex 999 and
1000).

- You want to do this for many complaint values (say A, B, C,..), I think
you'll have to dichotomize complaint variable.

Define the macro developped by Fabrizio Arosio (September 15, 99)
        DEFINE Macro(OutVar=!TOKENS(1) /InVar=!ENCLOSE('(',')')
        /Values=!ENCLOSE('(',')')).
        Count
        !LET !SEP=!NULL
        !DO !I !IN (!Values)
        !CONCAT(!SEP,!OutVar,!I,"=",!InVar,"(",!I,")")
        !LET !SEP="/"
        !DOEND
        !ENDDEFINE

Macro OutVar=cplt InVar=(complaint) Values=(A B C ...).
execute.
AGGREGATE
  /OUTFILE='c:\countcplt.sav'
  /BREAK=casenum
  /cplt1 to cpltx=SUM(cplt1 to cpltx). @x= max(A,B,C...)
@this file -countcplt.sav- contains one observation for each case and the
frequency for each complaint
sort cases by casenum (A).
MATCH FILES /FILE=*
  /TABLE='c:\countcplt.sav'
  /BY casenum.
- now you can use the count as variable or recode them as filter
RECODE  cplt1 to cpltx  (0=SYSMIS)  (ELSE=1)  .
EXECUTE .
You can now use cplt1 to cpltx as Multi response variable and use it in
general tables..

Alert : I didn't test it...
Regards
Naji

> -----Message d'origine-----
> De: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU]De la part de
> Kevin Borders
> Date: jeudi 3 février 2000 16:05
> À: SPSSX-L@LISTSERV.UGA.EDU
> Objet:
>
>
> Hello.  I'm a doctoral student in social work with a record selection
> question. I have a database made up of cases and complaint information.
>  Each case may have more than one complaint associated with it. The data
> file looks something like this:
>
> case number        complaint
> 001                     1
> 001                     45
> 002                     17
> 002                     87
> 003                     1
> 003                     22
> 003                     129
> 004                     1
> 004                     36
> 005                     77
> 005                     110
>
> While I know how to select the case numbers where the complaint is 1,
> what I now need is to be able to also select all of the other complaints
> that are associated with that case number.  What I get now is all of
> the case numbers where complaint = 1, and the rest of the complaints
> are left out.  Does anybody have any suggestions?  Thanks very much.
> --
> Kevin Borders, MSSW University of Louisville
> greenhiker@zdnetonebox.com
>
>
> ___________________________________________________________________
> To get your own FREE ZDNet onebox - FREE voicemail, email, and fax,
> all in one place - sign up today at http://www.zdnetonebox.com
>