Creating a numeric ID from name and surnames of cases

[pmulonge]

I would appreciate help with the following problem in SPSS syntax.
I have a dataset where the cases only have name and surname and no numeric
ID
e.g.
DATA LIST / name(a5) surname(a5).
BEGIN DATA
Johny Smith
Rachel Cohen
Ingrid Brown
END DATA.

I also have another dataset where presumably the same names and surname
appear  with the same spelling.
DATA LIST / name(a5) surname(a5).
BEGIN  DATA
JOHN SMITH
PAUL BLUE
RACHEL COHEN
TONY HILLS
INGRID BROWN
END DATA.

My question is then: is there a way to merge these datasets based on a
unique identifier? We have chosen name and surname.
My solution was to concatenate the name and surname into one string var
COMPUTE NOME = CONCAT(NAME,SURNAME).

I then proceed to try and use
RECODE NOME (CONVERT) NOME1.
ALTER TYPE NOME (F10).
COMPUTE NOME1 = NUMBER(NOME,F10)

But each time it just converts it to missing values, with no numbers.
The only one that worked was
AUTORECODE VAR = NOME /INTO NOME1.

The problem with AUTORECODE is that they are numbered alphabetically
according in the dataset, so INGRID BROWN will be a different number in the
first dataset from the number assigned in the second dataset. Therefore,
there is no unique ID given based on spelling of the string.
Is there a way around this?
Thanks

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[David Marso]

Several problematic issues come to mind.
*1. the two files appear to have different case!
See UPCASE and LOWER functions to address this.
*2. reading the two strings as A5 will truncate them.  Maybe that's what you want, but I doubt it as far as data integrity is concerned.
OTOH: for matching purposes you will mismatch Johny and JOHN even after fixing case.
*3.  It is unlikely *ALL* names will be unique so I wouldn't count on just first and last name as sufficient keys *DO YOU HAVE OTHER* variables which appear in both files.
*4.  Why concat?  MATCH allows many key variables to be used concurrently.
...
*5. CONVERT on RECODE only applies to "*NUMERIC*" data values
RTFM: "CONVERT recodes the string representation of numbers to their numeric representation."
..... blah blah blah...
"If RECODE encounters a value that cannot be converted, it scans the remaining value
specifications. If there is no specific recode specification for that value, the target variable
will be system-missing for that case."
6. AUTORECODE has a SAVE TEMPLATE and APPLY TEMPLATE subcommand.  Maybe that will help (don't know, my old reliable version doesn't support it)
So?
------
DATA LIST LIST/ name(a10) surname(a10) othrdat1 (F8).
BEGIN DATA
Johny Smith 12345678
Rachel Cohen 23456789
Ingrid Brown 23451672
END DATA.
LIST.
STRING Merge1 Merge2 (A10).
COMPUTE Merge1=UPCASE(Name).
COMPUTE Merge2=UPCASE(Surname).
SORT CASES BY Merge1 Merge2.
SAVE OUTFILE "temp1.sav".
DATA LIST LIST/ name(a10) surname(a10) othrdat2 (F8).
BEGIN  DATA
JOHN SMITH 344353
PAUL BLUE 345256
RACHEL COHEN 234353
TONY HILLS 262737
INGRID BROWN 123344
END DATA.
STRING Merge1 Merge2 (A10).
COMPUTE Merge1=UPCASE(Name).
COMPUTE Merge2=UPCASE(Surname).
SORT CASES BY Merge1 Merge2.

MATCH FILES / FILE "temp1.sav" /IN=Flag1/ FILE * /IN=Flag2/ BY Merge1 Merge2 /MAP.
LIST.



NAME       SURNAME    OTHRDAT1 MERGE1     MERGE2     OTHRDAT2 FLAG1 FLAG2

Ingrid     Brown      23451672 INGRID     BROWN        123344   1     1
JOHN       SMITH             . JOHN       SMITH        344353   0     1
Johny      Smith      12345678 JOHNY      SMITH             .   1     0
PAUL       BLUE              . PAUL       BLUE         345256   0     1
Rachel     Cohen      23456789 RACHEL     COHEN        234353   1     1
TONY       HILLS             . TONY       HILLS        262737   0     1


Number of cases read:  6    Number of cases listed:  6
 

pmulonge wrote

I would appreciate help with the following problem in SPSS syntax.
I have a dataset where the cases only have name and surname and no numeric
ID
e.g.
DATA LIST / name(a5) surname(a5).
BEGIN DATA
Johny Smith
Rachel Cohen
Ingrid Brown
END DATA.

I also have another dataset where presumably the same names and surname
appear  with the same spelling.
DATA LIST / name(a5) surname(a5).
BEGIN  DATA
JOHN SMITH
PAUL BLUE
RACHEL COHEN
TONY HILLS
INGRID BROWN
END DATA.

My question is then: is there a way to merge these datasets based on a
unique identifier? We have chosen name and surname.
My solution was to concatenate the name and surname into one string var
COMPUTE NOME = CONCAT(NAME,SURNAME).

I then proceed to try and use
RECODE NOME (CONVERT) NOME1.
ALTER TYPE NOME (F10).
COMPUTE NOME1 = NUMBER(NOME,F10)

But each time it just converts it to missing values, with no numbers.
The only one that worked was
AUTORECODE VAR = NOME /INTO NOME1.

The problem with AUTORECODE is that they are numbered alphabetically
according in the dataset, so INGRID BROWN will be a different number in the
first dataset from the number assigned in the second dataset. Therefore,
there is no unique ID given based on spelling of the string.
Is there a way around this?
Thanks

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD