Skip to contents

Convert LAPOP missing-value codes to haven tagged NAs

lpr_na_attributes.Rd

Converts numeric sentinel codes used in LAPOP/Stata workflows to [haven::tagged_na()] values. By default, `888888`, `988888`, and `999999` are converted to `NA(a)`, `NA(b)`, and `NA(c)`, respectively.

Usage

lpr_na_attributes(
  data,
  na_values = c(888888, 988888, 999999),
  na_tags = c("a", "b", "c"),
  vars = NULL,
  preserve_labels = TRUE,
  print = FALSE
)

Arguments

data

A data frame.

na_values

Numeric vector of sentinel values to convert.

na_tags

Character vector of single-letter missing-value tags. Must be the same length as `na_values`.

vars

Optional character vector of variable names to process. Defaults to all variables.

preserve_labels

Logical. If `TRUE`, labels attached to converted sentinel values are reassigned to the corresponding tagged NAs.

print

Logical. If `TRUE`, prints a compact missing-value summary for each processed variable. Defaults to `FALSE`.

Value

`data` with selected numeric variables converted from sentinel missing-value codes to haven tagged NAs.

Details

Unlike ordinary `NA` values or variable-level attributes, tagged NAs retain the missing-value reason for each observation and can be written to Stata as extended missing values by haven.

Examples

x <- haven::labelled(
  c(1, 888888, 988888, 999999),
  labels = c(Yes = 1, DK = 888888, NR = 988888, NotApplicable = 999999)
)
dat <- data.frame(x = x)
out <- lpr_na_attributes(dat)
haven::is_tagged_na(out$x)
#> [1] FALSE  TRUE  TRUE  TRUE