Option for turning on/off labelled columns (particularly Stata files) about haven

[hadley]

Are you worried about variable labels or value labels?

[nickbloom]

Either/both. I'd be happy to have both labels if standard vector classes (e.g. numeric, character) can be preserved. But I am willing to forgo labeling of any kind that necessitates adding the class "labelled" to a variable, as with rl$type below:

> class(rl$int_date)
[1] "numeric"
> class(rl$type)
[1] "labelled"
> str(rl$type)
Class 'labelled'  atomic [1:35556] 1 1 1 1 1 1 1 1 1 1 ...
  ..- attr(*, "label")= chr "Type of sample used"
  ..- attr(*, "labels")= Named int [1:2] 1 2
  .. ..- attr(*, "names")= chr [1:2] "RDD" "Callback"

rl$type is actually a numeric vector, but is classed "labelled." I'm envisioning it as an "unlabeled" boolean function parameter that (default can be TRUE), that, when FALSE, returns a standard vector class of that column, which would be either "numeric" or "integer" for rl$type. I always have a codebook to reference alongside analysis, so labels aren't that important, at least to me.

[hadley]

In that case, you should do df[] <- lapply(df, zap_labels)

[nickbloom]

Excellent. Thanks!