Extract original response codes and labels — lpr_extract

Reads dataset dictionaries and haven labels before considering factor levels. Factor positions are never substituted for the original survey codes.

Usage

lpr_extract_ros(
  data,
  lang_id = "en",
  include_special = FALSE,
  restrict_to_present = TRUE,
  one_row_per_var = FALSE,
  pair_sep = " | ",
  attr_name = "label.table",
  special_values = NULL
)

Arguments

data: A data frame imported with readstata13 or haven.
lang_id: Requested dictionary language, e.g. "en", "es", or "pt". `NULL` or `""` uses the variable's `val.labels` link, then the dataset's active language. A sole unambiguous dictionary is a final fallback. For haven vectors without language metadata, their attached labels are used.
include_special: Include tagged missing codes and explicitly labelled nonresponses. Valid codes are not excluded merely because they exceed 1000.
restrict_to_present: Keep only observed options. For questionnaire inventories use `FALSE`, so unobserved response options are retained.
one_row_per_var: Collapse into one row per variable with code-label pairs.
pair_sep: Separator between collapsed pairs.
attr_name: Preferred dictionary attribute. Even when `"levels"` is requested, an available original code dictionary takes precedence.
special_values: Additional response codes to exclude when `include_special = FALSE`.

Value

A tibble with `variable_name`, `value`, and `answer_text`, or just `variable_name` and `answer_text` when collapsed. Numeric codes remain numeric (including haven tagged NAs). If a dictionary has character codes, `value` is character. Empty variables have genuine NA cells. Plain factors without a code dictionary return their labels with unknown (`NA`) codes and a warning.

Examples

toy <- data.frame(x = c(0, 1))
attr(toy, "label.table") <- list(x_en = c(No = 0, Yes = 1, Other = 152501))
lpr_extract_ros(toy, restrict_to_present = FALSE)
#> # A tibble: 3 × 3
#>   variable_name  value answer_text
#>   <chr>          <dbl> <chr>      
#> 1 x                  0 No         
#> 2 x                  1 Yes        
#> 3 x             152501 Other      
lpr_extract_ros(toy, restrict_to_present = FALSE, one_row_per_var = TRUE)
#> # A tibble: 1 × 2
#>   variable_name answer_text                      
#>   <chr>         <chr>                            
#> 1 x             (0) No | (1) Yes | (152501) Other