在数据步骤中分配宏变量

Question

Confused about setting and recalling macro variables within a data step. I have an array of "Tumor" variables, and only one of them contains the information I need. There are a series of flags (BIN) to help me know which part of the Tumor array to reference. How can I do something like the following:

data tumors;
   input ID $ BIN1 BIN2 Tumor1 Tumor2;
   datalines;
1001 0 0 12 00
1002 1 0 01 01
1003 0 1 00 12
;

data newdata;
  set tumors;
  if BIN1 = 1 then do; %let value = 1; end;
  if BIN2 = 1 then do; %let value = 2; end;
  if Tumor&value in ('00','01','02') then Stage=0;
run;

The code outputs all blanks for "Stage" because I am doing this improperly, but I am not sure where the mistake is (there should be many listed as Stage 0). Any advice? I would want it to output the following:

data tumors_new;
   input ID $ BIN1 BIN2 Tumor1 Tumor2 Stage;
   datalines;
1001 0 0 12 00 ""
1002 1 0 01 02 01
1003 0 1 00 12 12
;

Answer 1

In your case, you do not need to use macro variables. You can do all of this with data step logic:

data newdata;
  set olddata;
  array Tumor[2];

  if(BIN1 = 1) then value = 1;
  if(BIN2 = 2) then value = 2;
  if(value IN(1, 2) ) then do; *Prevent errors from occuring if value is missing;
       if(Tumor[value]) in ('00','01','02') then Stage=0;
  end;
run;

Assuming your variable names are Tumor1, Tumor2 , we initialize an array named Tumor containing 2 values, which will automatically be named Tumor1 and Tumor2.

Explanation

The macro facility is a separate programming language from the SAS Data Step. There are only a few Data Step functions that connect the two languages. The reason why this is not working is because SAS always compiles macro language elements first before compiling any other code. When programming, always assume that your macro code will be interpreted first. I simply remember it in this compiling order:

1. Macro Code
2. SAS Code

In the program above, SAS does the following order of operations:

1. Assign 1 to the macro variable value
2. Assign 2 to the macro variable value
3. Resolve the macro variable value to 2
4. Compile the data step, then execute it

To bridge the gap between the Data Step and Macro Language, you need to use one of two functions:

call symput('macro variable name', variable or constant)

call symputx('macro variable name', scope <'G' or 'L'> )

symput (standing for S ymbol P ut) will read in the value of a data step variable into a macro variable, but only for that record. This is the tricky part. Because the Data Step naturally loops, it will constantly overwrite the value of your macro variable until the end of file marker. For this reason, it is very common to find call symput routines embedded in conditionals.

With symput , you are unable to use that macro variable directly within the Data Step. It only is usable after the data step completes. For example, you cannot use this logic:

data foo;
    set bar;
    call symput('macvar', var);
    if(&macvar = 1) then put 'Woo!';
run;

This will produce an error

ERROR 22-322: Syntax error, expecting one of the following: a name, a quoted string, a numeric constant, a datetime constant, a missing value, INPUT, PUT.

WARNING: Apparent symbolic reference MACVAR not resolved.

This is because macvar is not given a value until the end of the data step, causing an error in the if statement (and, because of the error, symput never runs, so macvar is never created). To SAS, the if statement above looks like:

if( = 1);

You can confirm this by typing in that code and finding that Error 22-322 occurs again.