Missing Values: NA and NULL

Real data has gaps: a survey question left blank, a sensor that failed, a value not yet measured. R represents a missing value with NA ("not available").

NA spreads through calculations

heights <- c(162, NA, 181, 158)
heights + 1
mean(heights)
mean(heights, na.rm = TRUE)

[1] 163  NA 182 159
[1] NA
[1] 167

Any calculation involving an unknown value is unknown — so the mean is NA. na.rm = TRUE ("remove NAs") tells the function to ignore missing values. Most summary functions accept it: sum, min, max, sd...

Finding missing values

heights <- c(162, NA, 181, NA)
is.na(heights)
sum(is.na(heights))
heights[!is.na(heights)]
heights == NA

[1] FALSE  TRUE FALSE  TRUE
[1] 2
[1] 162 181
[1] NA NA NA NA

heights == NA doesn't work — comparing anything with an unknown value gives NA. Always use is.na() to test for missing values.

NULL means "nothing"

NULL is different from NA: it means the value doesn't exist at all, rather than being missing.

length(NA)
length(NULL)
c(1, NULL, 2)

[1] 1
[1] 0
[1] 1 2

NA is a placeholder that takes up a slot; NULL disappears. You'll mostly meet NULL as the result of looking up something that doesn't exist, such as a missing list element.

Common mistakes

• Forgetting na.rm = TRUE and getting NA for a whole summary.
• Silently dropping missing values without asking why they're missing. Always count them first — how much data is missing, and where, can change your conclusions.

Practice

Given temps <- c(21, NA, 25, 19, NA, 23), print how many values are missing and the average of the rest.

temps <- c(21, NA, 25, 19, NA, 23)
sum(is.na(temps))
mean(temps, na.rm = TRUE)

[1] 2
[1] 22