SPSSX Discussion

simple question

Classic

List

Threaded

4 messages Options

mmcguire

simple question

Hello
I tend to use SPSS for simple data-analysis and so am not as acquainted
with its wider functions and syntax as I should be. Hence I have a
simple (and probably straightforward) problem that is frustrating me.

I am analysing a crime data-base and need to find if there are multiple
offenders incolved in an offence. My guide is the crime number for the
offence - however many times the crime number is repeated indicates the
number of offenders.
So - I need to recode (or compute?) a new variable from the variable
'crimeno' which translates the number of times the crime number recurs
into a value for the new variable.
For example, if the crime number 30445567 occurs three times, this
should be recoded into the value '3' in the new variable. Likewise if
the number 34512743 occurs three times, it should also be recoded into
the value '3' in the new variable.
Similarly, if the number 20566774 occurs 8 times, then it is recoded as
'8' in the new variable. And so on.

Simple.... (or it should be!)

I've tried using a frequency function to assign the new values but
without success so far. If anyone has an easy-fix procedure I could use
or other suggestions I'd be extremely grateful!!

thanks

--
===============================
Dr. Michael McGuire
London Metropolitan University
===============================

Maguin, Eugene

Re: simple question

Michael,

The solution is to use Aggregate along with the option to add new variables
to the existing data file.

In syntax this would be

Aggregate outfile=* mode=addvariables/break=crimeno/offendercout=nu.

You can also do all of this through the menu system: Data dropdown box in
the Data window.

I'm unsure whether you want this or to create a new file with the number of
times each offender is listed.

Gene Maguin

Hector Maletta

Re: simple question

In reply to this post by mmcguire

Apparently your database has one record per crime occurrence, and
"crimeno" is actually a "crime code", the coded designation of a single type
of offense. If this is the case, the solution is AGGREGATE. This function
creates a new file with fewer records, each representing a group of former
records. In your case, each new record would represent a crime type (a crime
number in your jargon), and you can create variables summarizing the group
of occurrences of that crime, including the number of times it showed up in
your database.

In a syntax window, you may write:
SORT CASES BY CRIMENO.
AGGREGATE OUTFILE * /PRESORTED/BREAK CRIMENO
/OCCURRENCES 'Number of occurrences'= nu.

This first sorts your working file by crime type. Then it creates a
new working file with one record per crime type, containing only two
variables: the crime type (CRIMENO) used as an aggregating key, and
OCCURRENCES representing the number of times that crime appears in your
original database.
The function NU defining OCCURRENCES is the Unweighted Number of
cases in each group (a "group" is the bunch of original records with the
same crime code).
AGGREGATE can be also be obtained from the Data Menu. It is
straightforward: Data -Aggregate, then use CRIMENO as break variable, and
then include the number of cases as variable (one of the choices in the
dialog box). You can also create other aggregate variables (such as the
average age of offenders, or the proportion of males, if you happen to have
their age or gender in the original records).

This new working file could then be saved to the disk for future
use if you want.

Hector

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
mmcguire
Sent: 16 July 2007 12:39
To: [hidden email]
Subject: simple question

Hello
I tend to use SPSS for simple data-analysis and so am not as
acquainted
with its wider functions and syntax as I should be. Hence I have a
simple (and probably straightforward) problem that is frustrating
me.

I am analysing a crime data-base and need to find if there are
multiple
offenders incolved in an offence. My guide is the crime number for
the
offence - however many times the crime number is repeated indicates
the
number of offenders.
So - I need to recode (or compute?) a new variable from the
variable
'crimeno' which translates the number of times the crime number
recurs
into a value for the new variable.
For example, if the crime number 30445567 occurs three times, this
should be recoded into the value '3' in the new variable. Likewise
if
the number 34512743 occurs three times, it should also be recoded
into
the value '3' in the new variable.
Similarly, if the number 20566774 occurs 8 times, then it is
recoded as
'8' in the new variable. And so on.

Simple.... (or it should be!)

I've tried using a frequency function to assign the new values but
without success so far. If anyone has an easy-fix procedure I could
use
or other suggestions I'd be extremely grateful!!

thanks

--
===============================
Dr. Michael McGuire
London Metropolitan University
===============================

Hector Maletta

Re: simple question - Addition

In reply to this post by Maguin, Eugene

I forgot to add that you can also ADD the aggregate variables to
your original file, as Gene Maguin suggests, but this is an additional
quirk, involving going up to the aggregate level and then back to the
elementary level. Since your acquaintance with SPSS seems to be still
elementary, I recommended doing just the aggregating, leaving the ADD
VARIABLES option for later.

Hector

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Gene Maguin
Sent: 16 July 2007 13:07
To: [hidden email]
Subject: Re: simple question

Michael,

The solution is to use Aggregate along with the option to add new
variables
to the existing data file.

In syntax this would be

Aggregate outfile=*
mode=addvariables/break=crimeno/offendercout=nu.

You can also do all of this through the menu system: Data dropdown
box in
the Data window.

I'm unsure whether you want this or to create a new file with the
number of
times each offender is listed.

Gene Maguin