topic Re: Creating new binomial variable from multiple binomial variables in SAS Programming

Creating new binary variable from multiple binary variables

tofov2 — Wed, 12 Jan 2022 18:42:02 GMT

Hello everyone,

I need to create a new binary variable (yes =1, no = 0) based on if ≥2 of a set of other binary variables are =1 per individual in the cohort.

In other words, I could name my new variable "multi_sport_athlete" and would set it as 1 only if this same individual played 2 or more of any sports, with variables such as "tennis" = 1, "football" = 1 , "rugby" = 1, "swimming" =1 .

Any help is greatly appreciated.

Re: Creating new binomial variable from multiple binomial variables

PaigeMiller — Wed, 12 Jan 2022 18:00:40 GMT

Sum the binary variables, and test to see if this sum is greater than or equal to 2.

if sum(football,tennis,rugby,swimming)>=2 then multi_sport_athlete=1;

Re: Creating new binomial variable from multiple binomial variables

ballardw — Wed, 12 Jan 2022 18:01:19 GMT

If you have numeric 0/1 coded variables then you can use the SUM function to get the number of values equal to 1.

But the layout of your data is needed to recommend exact code. Provide an example of the needed variables in the form of a data step and what the expected result for the example data might be.

If you have variables (which are very seldom in quotes in SAS code) named Tennis, Football,Rugby and swimming then a data step such as:

data want;
   set have;
   Multi_sport_athlete= ( sum(tennis,football, rugby, swimming) ge 2);
run;

should work. SAS will return 1 for true and 0 for false in a comparisons such as Somevar = ( <comparison code goes here>);

Note that if all the sport variable values are missing this will assign 0. So if you don't want that behavior you need to be a bit more descriptive about rules.