Date:         Thu, 15 Sep 2005 23:18:34 +0200
Reply-To:     Roland van der Laak <rwvanderlaak@hotmail.com>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Roland van der Laak <rwvanderlaak@hotmail.com>
Subject:      creating new respondentcode based on country-variable
Content-Type: text/plain; charset="iso-8859-1"

Hi list,

I've got the following challenge:

- repeating projects in several countries
- original respondentcodes (unique key) have a range from 1 - 1070000000 (no system per country whatsoever, only thing is that every respondent has a unique number in this range)
- client wants to have other respondentcodes, 6 positions max. The range(s) should be per country, e.g.: country 1 range 1-99999 , country 2 range 100000-199999, etc.
- preferably there should be a system where the transformation of the 'old/original' respondentcode should always lead to the same 'new' respondentcode, independent of the number of respondents in datafile, way the datafile is sorted etc.
- solution should be as general as possible, as we will have lots of projects where we will have this issue.


example data (before transformation/without new desired unique respondentcode):

DATA LIST LIST / coderesp (f10.0) country (f1.0).
BEGIN DATA.

1 1

2 2

3 3

100 1

200 2

900 3

1013 1

1015 2

200000 3

34234234 1

94584545 2

1067804002 1

1067806267 2

1067806886 3

1067806928 1

1067807364 2

1067808740 3

1067813733 1

1067816486 2

1067816614 3 

1067817859 1

END DATA.



Any ideas?



Thanks in advance,

kind regards,

Roland van der Laak