SPSS Statistics

You could write some SPSS looping code to do this, but here is the easiest way.

First, go to Extensions > Extension Hub and install the SPSSINC TRANS extension command.   If you already have it, check the Updates box to be sure you have the latest version.

Then run this code, where charges is the name of the charge variable, and charges1 to charges20 are the names of the output variables.  The command splits the string on the commas and writes the values in up to 20 string variables of length 7.  Change those settings as needed.

spssinc trans result=charges1 to charges20 type=7
    /formula "re.split(',', charges)".

either of these will work with parsing.  the second one you need to install jon's extension from the extension hub.  i'd probably pivot the tables after using varstocases.

1.

dataset close all.
DATA LIST LIST 
 / id (f3) variable (A35).
BEGIN DATA.
1, "122, 2, 3, 4"
2, "2, 34, 2, 4"
3, "5, 25, 1, 4"
4, "3, 2, 13, 2224"
5, "61, 254, 11, 9745"
END DATA.
string v1 to v4 (a5).
vector temp=v1 to v4.
loop i=1 to 4.
compute a=char.index(variable,', ').
compute temp(i)=char.substr(variable,1,a-1).
if a>0 variable=char.substr(variable,a+2,100).
if a=0 temp(i)=char.substr(variable,1,75).
end loop.
execute.

2. 
dataset close all.
DATA LIST LIST 
 / id (f3) variable (A35).
BEGIN DATA.
1, "122* 2, 3; 4"
2, "2, 34, 2, 4"
3, "5; 25, 1* 4"
4, "3* 2, 13; 2224"
5, "61, 254; 11, 9745"
END DATA.
spssinc trans result = v1 to v5 type=20
    /formula "re.split(r'[;,*\s]\s*', variable)".

Thank you so much! I just ran Jon's syntax, and it worked great. The only issue is that it stopped at the 825th case and there are 4552. Is there something I can change to get all 4552 cases like this? Thank you again!

The procedure should run through all the cases, but it respects any active SELECT or FILTER you have.

It will stop if there is a serious error, but in that case you would see an error message.  One common error source is too many values output.  If there are more charges found in a case than allowed for in the output specification, it will stop, but the message would be something like function returned too many values.

Hello @Andrew Denney.  Is the file you were given a .SAV file, or some other format that you had to import?  If the latter, what format?  Thanks for clarifying. 

Hi Jon,

When I ran this syntax...

spssinc trans result=S1 to S10 type = 40
/formula 're.split(";", Ethnicity)'.

...it correctly split out each instance, however how do we adjust it so that each unique value gets its own column/variable? See my attachment for the existing result.

Chris, do you mean that you want a 0/1 variable for each distinct value in the set of variables generated from SPSSINC TRANS?  TRANS can't easily do that, but you could do this.

Run the SPSSINC TRANS command
Define a multiple category set from the output variables
Use the STATS MCSET CONVERT (Tables > Convert Multiple Category Set) command to create a multiple dichotomy set.

Thank you, Jon. That solved it for me!!!