← SASnRCount Missing Values in Selected Variables

Introduction

A common audit question: how many of a given set of columns are missing on each row? — useful for completeness checks, data-quality flags, and reviewer queries.
We want a count per row, looking only at the columns we care about, not every column in the dataset.
SAS offers CMISS(), which works on both character and numeric inputs — just list the columns to check.
Tidyverse uses across() with is.na() and rowSums() to count NAs across selected columns.
Base R uses rowSums(is.na(df[, cols])) — the bracket subset picks the columns, is.na() builds a logical matrix, and rowSums() sums each row.
This lesson counts missingness across name and score on a small four-row dataset so each value is easy to inspect.

Count Missing Values in Selected Variables

SAS (Base SAS)+

data example;
    input id name $ age score;
    datalines;
1 Alice 25 85
2 . 30 90
3 Charlie . 88
4 . . .
;
run;

data example2;
    set example;
    cmiss_subset = cmiss(name, score);
run;

CMISS() is the SAS function that counts the missing values among its arguments — it accepts both numeric and character variables in the same call.
cmiss_subset = cmiss(name, score); counts how many of name and score are missing on each row.
On a row where both are missing the result is 2; where only one is missing it is 1; where both are present it is 0.
CMISS() ignores the type difference between name (character) and score (numeric) — no need to split the call.
For the count of non-missing values, N(of var1 var2 ...) is the complementary function — useful for "complete n" summaries.
We run and check example2: the cmiss_subset value on each row should equal the number of missing cells among name and score.

R (tidyverse)+

df <- tibble(
id = 1:4,
name = c("Alice", NA, "Charlie", NA),
age = c(25, 30, NA, NA),
score = c(85, 90, 88, NA) 
)
 

df2 <- df %>% mutate(cmiss_subset = rowSums(is.na(across(c(name, score)))))

In tidyverse, we combine across() with is.na() and rowSums() to reproduce the SAS CMISS() row-wise count.
across(c(name, score)) selects the target columns; using across() keeps the call compact and supports any tidyselect helper.
is.na(across(...)) returns a logical data frame where TRUE marks missing cells.
rowSums(...) sums those TRUE/FALSE values row by row, giving the missing-count per row.
We can swap c(name, score) for where(is.character) or starts_with("lb") to count missingness in any column family without listing names.
We run and confirm df2$cmiss_subset matches the SAS example2.cmiss_subset row for row.

Happy learning!

Search lessons

Introduction

Count Missing Values in Selected Variables