mariposa (version 0.7.3)

set_na: Declare Values as Missing

Description

Replaces specific numeric values with NA (or tagged NAs) across one or more variables. This is essential for data cleaning workflows where missing value codes (e.g., -9, -8, 99) are stored as regular values and need to be declared as missing after import.

Usage

set_na(data, ..., tag = TRUE, verbose = FALSE)

Value

The modified data (invisibly for data frames, visibly for vectors).

Arguments

data

A data frame, tibble, or a single vector.

...

Values to set as missing. Can be:

• Unnamed numeric values: Applied to all numeric columns (e.g., set_na(data, -9, -8))

• Named pairs: Applied to specific variables (e.g., set_na(data, income = c(-9, -8), age = -1))

tag

If TRUE (default), uses tagged NAs to preserve distinct missing types. The resulting tagged NAs integrate with na_frequencies(), frequency(), and codebook(). If FALSE, replaces with regular NA.

verbose

If TRUE, prints a summary of conversions.

Details

Tagged vs. Regular NA

When tag = TRUE (default), each missing value code gets a unique tag character, so you can distinguish between "No answer" (-9) and "Not applicable" (-8) in downstream analysis. This is the same system used by read_spss() with tag_na = TRUE.

When tag = FALSE, all specified values become regular NA and the distinction between different missing types is lost.

Interaction with Existing Labels

If a value being set to missing has an existing value label, that label is preserved as a tagged NA label (when tag = TRUE), making it visible in frequency() and codebook() output.

See Also

na_frequencies() for inspecting missing types, strip_tags() for converting tagged NAs to regular NA, untag_na() for recovering original codes

Other labels: copy_labels(), drop_labels(), find_var(), to_character(), to_label(), to_labelled(), to_numeric(), unlabel(), val_labels(), var_label()

Examples

# Use regular NA instead of tagged NA (no packages required)
data <- set_na(survey_data, -9, -8, tag = FALSE)

# Tagged NAs preserve the distinction between missing types
# (requires the haven package)
if (requireNamespace("haven", quietly = TRUE)) {
  # Set -9 and -8 as missing across all numeric variables
  data <- set_na(survey_data, -9, -8)

  # Set missing for specific variables only
  data <- set_na(survey_data,
    income = c(-9, -8, -42),
    life_satisfaction = c(-9, -11)
  )

  # Check the result
  na_frequencies(data$income)
}