I didn't capture one aspect clearly. In the groups of records identified by casenum, only one record in some of the groups has a 1 in it. I need to capture all the records associated with that record according to casenum. Thanks.

-----Original Message----- From: Richard Ristow [mailto:wrristow@mindspring.com] Sent: Wednesday, October 22, 2003 7:13 PM To: Raffe, Sydelle, SSA; SPSSX-L@LISTSERV.UGA.EDU Subject: Re: data structure.xls

At 01:20 PM 10/22/2003 -0700, Raffe, Sydelle, SSA wrote:

>I have a file wherein records are keyed together by >"casenum". Casenum groups can be from 2 to 16(highly unlikely) >records. I need to identify groups of records with the same casenum >where there is a code 1 in the last column. > >I'd like to end up with an excel data base with only those groups of >records keyed by casenum where there is a one in that column (variable).

Well, in SPSS this is easy with AGGREGATE/MATCH FILES. I'll let you move it to Excel yourself.

If I have this right, you want all records where *any* record with the same "casenum" has a 1 in a variable I'll call LASTCOL.

This is untested code. I assume you can write scratch files freely to c:\TEMP.

*----------------------------------------------. SORT CASES BY CASENUM. /* Put groups together */

TEMPORARY. COMPUTE LAST_IS1 = 0. IF (LASTCOL = 1) LAST_IS1 = 1. AGGREGATE/OUTFILE='c:\TEMP\KEEP_GRP.SAV' /BREAK= CASENUM /KEEP_GRP = MAX(LAST_IS1).

MATCH FILES/FILE=*/FILE='c:\TEMP\KEEP_GRP.SAV' /BY CASENUM. SELECT IF (KEEP_GRP = 1). *----------------------------------------------. There are variations, of course, but this should work quickly and easily.