haven zap_labels() in R: Remove Value Labels from Data
The haven zap_labels() function removes value labels from a labelled vector, the kind read_sav(), read_dta(), and read_sas() produce. It returns the underlying codes as a plain R vector, ready for analysis.
zap_labels(x) # drop value labels from a vector zap_labels(df) # drop value labels from every column zap_labels(read_sav("survey.sav")) # strip labels right after import str(zap_labels(x)) # confirm the labels are gone zap_label(x) # drop the VARIABLE label instead zap_formats(x) # drop SPSS or Stata display formats as_factor(x) # keep labels, convert to a factor
Need explanation? Read on for examples and pitfalls.
What zap_labels() does
zap_labels() converts a labelled vector into a plain one. When haven imports an SPSS, Stata, or SAS file, coded columns arrive as labelled vectors that pair each number with a text label. zap_labels() discards those value labels and hands back the bare codes, which is the form most modelling and summary functions expect.
zap_labels() syntax and arguments
zap_labels() takes a single argument. You pass either one labelled vector or a whole data frame, and it returns the same object with value labels removed.
zap_labels() is an S3 generic. It has methods for labelled vectors, for SPSS-style labelled vectors with user-defined missing values, and for data frames. Plain vectors with no labels pass through untouched, so the function is always safe to call. The full method list is in the haven reference.
Strip value labels with zap_labels()
Call zap_labels() on a labelled vector to get its codes back. Build a labelled vector with labelled(), then strip it down to numbers.
Pass a data frame and every labelled column is cleaned in one call. Columns that were never labelled are left as they are.
zap_labels() keeps the variable label, the short description SPSS and Stata attach to each column. Only the value labels go.
zap_labels() vs as_factor and other zap functions
zap_labels() is one of a family of haven attribute strippers. Each removes a different piece of the metadata haven attaches to imported columns. Pick the function by what you want gone.
| Function | Removes | Keeps | Use when |
|---|---|---|---|
zap_labels() |
value labels | codes, variable label | you need numeric codes for modelling |
zap_label() |
variable label | codes, value labels | the variable description clutters output |
zap_formats() |
display formats | codes, labels | SPSS or Stata formats break a package |
as_factor() |
the labelled class | label meaning, as levels | you want human-readable categories |
The decision rule is short. Use as_factor() when the label text matters for plots or grouped tables, and zap_labels() when you only need the raw numbers and the labels are in the way.
Common pitfalls
Three mistakes catch new haven users. Each is easy to avoid once you know zap_labels() drops information for good.
1. Expecting the labels to survive. zap_labels() deletes them, the codes lose their meaning, and there is no undo.
To keep the meaning, convert with as_factor() before, or instead of, stripping.
2. Forgetting the variable label stays. zap_labels() leaves the variable label in place. If a function still prints a description, reach for zap_label() as well.
3. Assuming it ignores user-missing values. For SPSS data with user-defined missing values, zap_labels() also converts those tagged values to ordinary NA. Check your missing-value counts after calling it.
Try it yourself
Try it: Build a labelled vector ex_rating with codes 1, 2, and 3 labelled Poor, Fair, and Good, then strip its value labels into ex_clean.
Click to reveal solution
Explanation: zap_labels() removes the value labels and returns the underlying numeric codes. The Poor, Fair, and Good text is discarded.
Related haven functions
These functions pair naturally with zap_labels(). Each handles a related part of importing and cleaning labelled data.
zap_label()removes the variable-level label that SPSS and Stata store.zap_formats()removes the display format string attached to a column.zap_missing()converts user-defined missing values to plain NA.as_factor()turns a labelled vector into a factor, keeping the label text.read_sav()imports the SPSS files whose columns you will be stripping.
FAQ
What does zap_labels() do in R?
zap_labels() removes the value labels from a labelled vector created by haven. A labelled vector pairs each code, such as 1 or 2, with a text label like Male or Female. zap_labels() discards those labels and returns the plain codes. The result is an ordinary numeric or character vector that base R and modelling functions handle without any special treatment.
What is the difference between zap_labels() and zap_label()?
They remove different things. zap_labels(), plural, removes the value labels that map codes to text. zap_label(), singular, removes the variable label, the short description of the whole column. A column read from SPSS often carries both. Call zap_labels() to drop the code-to-text mapping and zap_label() to drop the column description. Use both for a completely bare vector.
Does zap_labels() remove the variable label?
No. zap_labels() removes only the value labels and leaves the variable label untouched. If you want the variable label gone too, call zap_label() on the result. Many users expect zap_labels() to clear everything, so a description still showing in output is a common surprise.
How do I remove all labels from a haven data frame?
Pass the data frame straight to zap_labels(). It applies the labelled-vector method to every column, so one call cleans the whole table. Columns that were never labelled pass through unchanged. To also drop variable labels and display formats, chain zap_label() and zap_formats() over the same data frame.
Should I use zap_labels() or as_factor()?
Use as_factor() when the label text matters, such as for axis labels or grouped summaries, because it keeps the meaning as factor levels. Use zap_labels() when you only need the numeric codes, for example before fitting a model that expects numbers. zap_labels() throws the labels away, while as_factor() preserves them in a different form.