← SASnRconditional logic - case_when - .default and missing values

Introduction

Many derivations are multi-branch: if X then A, else if Y then B, else C. A typical example is age banding — under 18, 18 to 59, 60 and over.
SAS handles this with if-else if-else inside a DATA step. Tidyverse offers case_when() from dplyr, which reads top-to-bottom like a truth table. Base R chains ifelse() calls.
We also have to think about missing values — they can be handled explicitly (a dedicated branch) or default to the final fallback.
This lesson builds an age band using both approaches: one where missing ages fall through to the default branch, and one where they get their own "Age Missing" category.
After running, dm1 has four rows with agegrp values derived from the age, and dm2 treats the missing age as a separate category.

conditional logic - case_when - .default and missing values

R (tidyverse)+

library(dplyr)

# Sample data
dm <- tibble(
  USUBJID = c("101-001", "101-002", "101-003", "101-004"),
  AGE = c(35, 64, NA, 17)
)

# .default used when no condition matches
dm1 <- dm %>%
  mutate(agegrp = case_when(
    AGE < 18 ~ "< 18 Years",
    AGE >= 60 ~ ">= 60 Years",
    .default = "18-59 Years"
  ))

# Missing values handled separately
dm2 <- dm %>%
  mutate(agegrp = case_when(
    is.na(AGE) ~ "Age Missing",
    AGE < 18 ~ "< 18 Years",
    AGE >= 60 ~ ">= 60 Years",
    .default = "18-59 Years"
  ))

For multi-branch logic in tidyverse, we reach for case_when() from dplyr.
Each line inside case_when() is a condition ~ value pair; dplyr evaluates them top-to-bottom and returns the value for the first matching condition.
In the first example, a row with AGE = NA does not match AGE < 18 or AGE >= 60 (NA comparisons return NA, not TRUE). So it falls through to .default and gets "18-59 Years" — probably not what we want.
In the second example, we add an explicit is.na(AGE) ~ "Age Missing" branch as the first line. Because case_when stops at the first match, missing ages get their own label and never reach the numeric comparisons.
.default = replaces the older TRUE ~ value trick. Both still work; .default is clearer.
After running, we check that the missing row has the right label in each output.

Happy learning!

Search lessons

Introduction

conditional logic - case_when - .default and missing values