haven zap_formats() in R: Clear SPSS and Stata Formats

The haven zap_formats() function removes the SPSS, Stata, and SAS display-format attributes that haven attaches to imported columns, leaving plain R vectors that print and behave normally.

zap_formats(x)                          # strip formats from one column
zap_formats(df)                         # strip formats from every column
df$pay <- zap_formats(df$pay)           # strip one column in place
attributes(zap_formats(x))              # confirm format attrs are gone
df |> mutate(across(everything(), zap_formats))  # tidyverse style
zap_labels(zap_formats(x))              # chain with other zappers

What zap_formats() does

zap_formats() deletes display-format metadata, not data. When haven reads an SPSS, Stata, or SAS file, it preserves each variable's original display format so the data can round-trip back to the source software unchanged. That format is stored as an attribute (format.spss, format.stata, or format.sas) on every affected column.

Those attributes are harmless most of the time, but they can confuse functions that inspect attributes, clutter str() output, or trip up packages that do not expect them. zap_formats() strips all three attributes and returns the column or data frame with identical values.

zap_formats() syntax and arguments

The signature is minimal: zap_formats(x). The single argument x is either a vector or a data frame. For a vector, zap_formats() removes the format attributes directly. For a data frame, it applies itself to every column and returns the cleaned frame. There are no other parameters to tune.

The function targets exactly three attributes, one per supported statistical package:

The values are untouched. Only the metadata disappears.

Remove format attributes with zap_formats()

Pass a whole data frame to clean every column at once. This is the common case after reading a survey export, where many columns each carry a stray format string. A single call clears them all.

You can also chain it with other zappers, for example zap_labels(zap_formats(x)), when a column carries both formats and value labels.

zap_formats() vs zap_labels and other zap functions

Each zap function removes one specific kind of haven metadata. They are easy to confuse because they share a prefix, but they act on different attributes. Picking the wrong one leaves the attribute you actually wanted to remove still attached.

The rule is simple: use zap_formats() for display formats, zap_labels() for coded value labels, and zap_label() for the human-readable variable description. Only zap_empty() alters the data itself.

Common pitfalls

The most frequent mistake is forgetting to reassign the result. zap_formats() returns a cleaned copy; it does not modify its input in place. Calling it without capturing the output does nothing visible.

The third pitfall is harmless but worth knowing: running zap_formats() on a column that has no format attribute is a safe no-op. It returns the column unchanged rather than raising an error, so you can apply it defensively across mixed data.

Try it yourself

Related haven functions

These functions handle the other metadata that haven attaches to imported SPSS, Stata, and SAS columns:

• zap_labels() removes coded value labels and returns the underlying values.
• zap_label() removes the variable's human-readable description.
• zap_widths() removes SPSS display-width attributes.
• as_factor() converts labelled vectors into ordinary R factors.
• read_sav(), read_dta(), and read_sas() import the files that carry these formats. See the official haven zap_formats reference for package details.

FAQ

What does zap_formats() do in haven?

zap_formats() removes the display-format attributes (format.spss, format.stata, and format.sas) that haven attaches to columns when it imports SPSS, Stata, or SAS files. These attributes record how each variable was formatted in the source software. The function returns the same vector or data frame with identical values, just without the format metadata. It is a cleanup step, not a transformation, so the numbers and strings you care about never change.

Does zap_formats() remove value labels?

No. zap_formats() only touches display-format attributes. Value labels are stored in a separate labels attribute and are removed by zap_labels(). If an imported column still prints with labelled categories after you run zap_formats(), that label attribute is still attached. Use zap_labels() to drop the coded labels, or as_factor() to convert them into a proper factor instead.

Do I need zap_formats() after read_sav()?

Not always. The format attributes are harmless for most analysis and are only needed if you plan to export the data back to SPSS or Stata. Use zap_formats() when the attributes interfere with code that inspects attributes, when str() output is cluttered, or when a downstream package fails on the unexpected metadata. If you never export to a statistical package, zapping formats keeps your data frame clean.

What is the difference between zap_formats() and zap_widths()?

zap_formats() removes the display-format strings such as F8.2 or %9.0g. zap_widths() removes the separate display_width attribute that SPSS uses to set column width in its data editor. They target different attributes, so removing one does not affect the other. A column imported from SPSS can carry both, in which case you call each function once or chain them together.