@Beargrad04 wrote:
So on the one hand I have a method that is supposed to be beneficial; on the other hand that same method can be decrypted and is not as randomized as randomized digits I'm trying to find out how to do.
As it stands, it sounds as if the excel import process is the way to go so far.
So I import a set of random ID numbers paired with my IDs I want made anonymous, then I merge the file with said IDs and data with the anonymous ID imported by the said ID. That way my anonymous ID is paired with both files correctly and synched up with the rest of the fields I want to report.
Lets go through what @Reeza did with an example data set
This gets a unique list of names for associating. If your data already does not have any duplicates of the personal id values that would not be needed. BUT you really should verify that you have no duplicates before you start anything and with 30K+ plus records in an Excel sheet how did you verify no duplicates? You would use the name of your data set instead of SASHELP.Class.
proc sql;
create table unique_list as
select distinct name
from sashelp.class;
quit;
This added a random number with the RAND function. If your data has no duplicates you could place the name of your data set than needs random numbers on the SET statement. There are different RAND function options, if you really need an integer you could use the FLOOR function on the result of the RAND function. Or something like (rand('integer',1000000000,9999999999) to directly generate 10 significant digit values.
This is the FIRST step you need if your list is 100% unique.
data random_values;
set unique_list;
call streaminit(&random_seed.);
rand = rand('normal', 50, 10);
run;
This scrambles your list of ids into a random order based on the random number result. It is needed here because Reeza's example does not actually have an existing ID number, only the name and she want to create an id that is not based ont the alphabetical order of the names in the SASHELP.CLASS data set. The code below will create an "id" value based on the random order.
proc sort data=random_values;
by rand;
run;
Important note at this time: You likely should make sure that the data set Random_values is stored in a permanent library as you might need it later.
This step adds the Id, a simple numeric order assigned after the sort. You would not need that if you already have an ID value. The additional pieces are preparing to create a Format, which is one of the ways SAS can look up values quickly. A format will display one value, the "START" value with the actual value of the "LABEL" . In your case you would want the RAND value as the LABEL and your existing ID variable name as the START. If your ID is actually a numeric value (bad idea put folks do that) then change the Type='C' to Type='N' so SAS understand it will be manipulating a numeric value. This would be the SECOND step that you need.
data ID_key_pair;
set random_values;
label = put(_n_, z5.);
fmtname = 'anon_fmt';
type='C';
start=name;
run;
This code actually creates the format. CNTLIN tells the procedure that the data set contains the instructions to create a format.
If you have a permanent library associated with this project you could add LIBRARY= lib to create the format in that library. You would use the SAS option FMTSEARCH to add that library to the search path to make it useable. The default shown below would create the format in the WORK library and be found by default. BUT the proc format code shown would need to be rerun every time you want to use the format. The THIRD step.
proc format cntlin=id_key_pair;
run;
And the following code uses the PUT function to create the random Id based on the value of the START values used above. In the example that is the value of the Name variable.
data want;
set sashelp.class;
RandomID = put(name, $anon_fmt.);
*drop name;
run;
Your code would look very similar something like: and the LAST step needed to add the Newid to your data set. Note: you really do want to create a new data set. If you make a mistake you do not want to take the chance of destroying or corrupting your original data unless it is very easy to recreate exactly as it was at the start.
data want;
set yourdatsetname;
NewId = put(youridvariable, $anon_fmt.);
run;
I can't be any more specific with data set names or variables because you have not actually shared what you have.
Now, a critic piece of information that may have forgotten to mention: will you have to add new student ids to this list? Because you do not want to restart this process with a full list unless you actually intend to remove the previous "newid" with something else so any previous reports or such that seem to have needed the "newid" would not reference the current look up created between the added list and the old list of students.
