← SASnRAdding new variables

Introduction

Most analysis steps derive new variables from existing ones — BMI from height and weight, age bands from age, or a flag from a date comparison.
SAS does this inside a DATA step by assigning to a variable that does not yet exist. The new name appears as a column in the output.
In tidyverse we use mutate(), which adds one or more derived columns in a single call. In base R we assign to df$new_col directly.
This lesson derives BMI (weight in kg divided by height in metres squared) from the CLASS dataset. Height in the input is inches, weight is pounds, so we convert units in the formula too.
After running, the new column BMI holds a numeric BMI value for each subject.

Adding new variables

SAS (Base SAS)+

data DM;
 infile datalines dlm='|' dsd missover;
input subjid : $10. age : best32.;
label ;
 format ;
 datalines4;
1001|35
1002|64
1003|19
;;;;
 run;

data dm02;
     set dm;
     agemon=age*12;
    group=1;
run;

data dm02;
     set dm;
     length agegr1 $30;
    if age lt 20 then agegr1="<20 Years";
    else if 20 le age lt 60 then agegr1="20 - <60 Years";
    else if age ge 60 then  agegr1=">= 60 Years";
run;

To derive a new variable in SAS, we assign to it inside a DATA step: BMI = weight / (height**2) * 703;.
Any name that did not exist on the left side of an assignment before becomes a new column in the output dataset — no explicit declaration needed.
The formula uses the imperial BMI shortcut: weight (lb) divided by height (in) squared, times 703. We could equally convert to metric first and use the textbook formula.
SAS applies the formula to every row as it reads the input, so the computation is vectorised by default.
We could also derive many variables in one DATA step — just keep adding assignment lines before the run;.

R (tidyverse)+

dm=60 ~ ">= 60 Years",
   between(age,20,59) ~ "20 - < 60 Years",
   age < 20 ~ "< 20 Years" )
)

In tidyverse, we add derived columns with mutate().
mutate(class, BMI = Weight / (Height^2) * 703) adds a new column named BMI computed from the existing Weight and Height columns.
The right-hand side is a regular R expression. It is evaluated row-wise because each column is a vector and R applies arithmetic element-wise.
We can add multiple derived columns in one call: mutate(class, BMI = ..., BMI_cat = ifelse(BMI > 25, "High", "Normal")). The second derivation can even reference the first.
The result is a tibble with the original columns plus each new one appended on the right.

Happy learning!

Search lessons

Introduction

Adding new variables