set macro as denominator for a dataset

[monday89]

i have the following macro 

%let group_1=(n=4);
%let group_2=(n=10);

I have the following dataset. What I could like to do is that if group_strata = 1 then the denominator = 4 from macro above and some with group_strata = 2

medication	Group_strata	numerator
AAA	1	3
BBB	1	2
CCC	1	1
AAA	2	2
BBB	2	4
CCC	2	1

I tried the following and it doesn't work

data want; set have;

if group_strata = 1 then total_denominator = (&group_1);
else if group_strata = 2 then total_denominator = (&group_2);
run;

[Tom · Accepted solution]

If your dataset does not have a variable named N then running this code:

data want; set have;
  if group_strata = 1 then total_denominator = ((n=4));
  else if group_strata = 2 then total_denominator = ((n=10));
run;

Will result in TOTAL_DENOMINATOR always being set to 0.  SAS will create a new variable named N and set it to missing and missing is not equal to either 4 or 10.

Can you change how those macro variables are created?

If not then remove the N= part.  For example by using %SCAN() function.

data want; set have;
  if group_strata = 1 then total_denominator = %scan(&group_1,2,(=)) ;
  else if group_strata = 2 then total_denominator =  %scan(&group_2,2,(=)) ;
run;

[monday89]

Thanks! The second box worked! 

[ballardw]

Where do those values come from and why do you think that they need to be in macro variables?

[PaigeMiller]

---

@ballardw wrote:

Where do those values come from and why do you think that they need to be in macro variables?

---

Agreed. Unless you have no control over the contents of the macro variable ... then %let group_1=(n=4); is a poor way to set things up; and its hard to see why macro variables are needed anyway.

[Kurt_Bremser]

You want to have an informat:

proc format;
invalue denom
  1 = 4
  2 = 10
  other = .
;
run;

data want;
set have;
total_denominator = input(group_strata,denom.);
run;

Untested, for lack of usable example data.

[kelxxx]

%let group_1=(n=4);
%let group_2=(n=10);

data have;
input medication $3 group_strata numerator;
cards;
AAA 1 3
BBB 1 2
CCC 1 1
AAA 2 2
BBB 2 4
CCC 2 1
;
run;

data want;
set have;
total_denominator=input(scan(symget(cats("group_",group_strata)),2,"(=)"),8.);
run;