Use AGGREGATE. Go to the Data screen, and choose Data-Aggregate. This
would create a new file of aggregated cases. If you want the sum of the
variable Y for particular subgroups, say for each value of the variable
X, then you should indicate that the aggretation is to be broken down
by X. If you want just the overall sum, you should first create a
variable that is constant for everybody, then use that variable to
"break" the aggregation.
This new aggregated file could then be merged with the original file,
assigning to each original case the aggregated variable from the
aggregated file. Of course, the aggregated variable need not be a sum:
it could also be an average, or a fraction of cases, or the value of the
first case in each group, or the value of the last case, or the maximum
or minimum value encountered within each group, and so on (look for help
Example with syntax:
GET FILE 'original.sav'.
save outfile 'original.sav'.
MATCH FILES /FILE = 'ORIGINAL.SAV'/TABLE=*/by X.
save outfile 'augmented original.sav'.
In the menu:
1. Go to Data - Transform - Compute to create the X variable if needed.
Save the file.
2. Go to Data - Aggregate. Follow instructions. Create only one
aggregated variable (the sum you need) or more variables as needed. The
aggregated file will become the active file.
3. Go to Data - Merge files - Add new variables. External file:
'original.sav'. Choose that the working file is a keyed table. Choose X
as the key variable for the matching. Proceed. Save the resulting file
with the name 'augmented original.sav' or whatever you like.
Notice that the file resulting from AGGREGATE is a file with one record
per group (only one record if you want the sum total of all cases). For
the matching, this smaller file is designated as a "TABLE", i.e. a
lookup table whose variables would be assigned to any case in the other
file that matches the value of the key variable (X). In the case of an
overall sum, the overall sum is added to all cases in the original file;
if the sum was obtained separately for different groups (say, for
various neighbourhoods, or for each sex, or whatever), all cases in each
neighbourhood or sex would be assigned the sum of Y obtained in that
neighbourhood or sex.
Universidad del Salvador
Buenos Aires, Argentina
Will Franks wrote:
> Hi My name is Will and I new to the group.
> I was wondering if there is anyway to add across cases and make that sum a
> new variable. For example:
> number of people Total number of people
> 1 6
> 2 6
> 3 6
> Thanks in advance
> MSN Photos is the easiest way to share and print your photos: