topic Re: Array/loop for new categorized variables in SAS Programming

Array/loop for new categorized variables

Adele3 — Mon, 17 Jun 2024 19:29:09 GMT

Hello,

I have a series of 10 Likert scale (1-3) variables that need to be recategorized using the same criteria. The variables are as follows:

a1_c_q1 - a1_c_q5, where 'a1' is administrator 1, 'c' is cognitive test, and 'q' represents the questionnaire number (1 though 5).

a2_c_q1 - a2_c_q5, where 'a2' is administrator 2, 'c ' is cognitive test, and 'q' represents the questionnaire number (1 through 5).

For each of the 10 variables, I am looking to create a new recategorized variable where 1=3 and 3=1. Is there a way to use an array or loop in the code so the same lines do not need to be repeated 10 times?

DATA new1;
     SET new;

	 a1_c_q1_new=a1_c_q1;
	 IF a1_c_q1=3 THEN a1_c_q1_new=1; 
	 ELSE IF a1_c_q1=3 THEN a1_c_q1_new=1; 

RUN;

Re: Array/loop for new categorized variables

ballardw — Mon, 17 Jun 2024 20:47:20 GMT

I think you need to look very closely that the IF that you show and see if that is correct:

	 IF a1_c_q1=3 THEN a1_c_q1_new=1; 
    ELSE IF a1_c_q1=3 THEN a1_c_q1_new=1;

I suspect that you meant

	 IF a1_c_q1=3 THEN a1_c_q1_new=1; 
    ELSE IF a1_c_q1=1 THEN a1_c_q1_new=3;

I moved the code a bit to show that the IF and the ELSE are basically identical.

Here is a litteral translation of that into arrays:

DATA new1;
    SET new;
    array a (*)  a1_c_q1 - a1_c_q5  a2_c_q1 - a2_c_q5
    array new (*) newa1_c_q1 - newa1_c_q5 newa2_c_q1 - newa2_c_q5,;
    do i=1 to dim(a);
       new[i] = a[i];
       if a[i]=3 then new[i]=1;
       else if a[i]=3 then new[i]=1;      
    end;
RUN;

Strongly recommend putting a suffix (start) on your new variables as you can then use list syntax as shown in the array definition. Otherwise you have to write out every single "_new" on the array definition and cannot use lists in any of the functions that might be useful to work with the values later.

Best might be to actually show a few values and the real result.

Another option could be a custom INFORMAT when reading text for those variables.

Re: Array/loop for new categorized variables

Reeza — Mon, 17 Jun 2024 20:47:22 GMT

You can definitely do this, but you'll need to create two sets of arrays, one for the old variables and one for the new variables.

Note that I moved the _new to the middle of the questions to make it easier to shortcut the name references. Untested but this is the idea.

DATA new1;
     SET new;
    
      array _old (*) a1_c_q1 - a1_c_q5 a2_c_q1 - a2_c_q5;
      array _new(*)  a1_c_new_q1 - a1_c_new_q5 a2_c_new_q1 - a2_c_new_q5 ;

     do i=1 to dim(_old);
	_new(i) = _old(i)
	 IF _old(i)=3 THEN _new(i)=1; 
	 ELSE IF _old(i)=1 THEN _old(i)=3; 
      end;

RUN;

Here's a tutorial on using Arrays in SAS
https://stats.idre.ucla.edu/sas/seminars/sas-arrays/

@Adele3 wrote:

Hello,

I have a series of 10 Likert scale (1-3) variables that need to be recategorized using the same criteria. The variables are as follows:

a1_c_q1 - a1_c_q5, where 'a1' is administrator 1, 'c' is cognitive test, and 'q' represents the questionnaire number (1 though 5).

a2_c_q1 - a2_c_q5, where 'a2' is administrator 2, 'c ' is cognitive test, and 'q' represents the questionnaire number (1 through 5).

For each of the 10 variables, I am looking to create a new recategorized variable where 1=3 and 3=1. Is there a way to use an array or loop in the code so the same lines do not need to be repeated 10 times?
DATA new1;
     SET new;

	 a1_c_q1_new=a1_c_q1;
	 IF a1_c_q1=3 THEN a1_c_q1_new=1; 
	 ELSE IF a1_c_q1=3 THEN a1_c_q1_new=1; 

RUN;