Skip to contents

A plain CSV carries codes. dm_1 arrives in SPSS as a column of integers with empty variable and value labels, and the reader has to reconstruct all of it from the questionnaire. This attaches both from the instrument: the variable label is the item's label, and the value labels come from the item's choice set, which stores its values and labels side by side.

Usage

export_responses(data, instrument, path)

Arguments

data

A response data frame, as returned by read_responses().

instrument

The sframe the responses were collected with.

path

Output file. An .sav is written for SPSS, a .dta for Stata, chosen from the extension.

Value

The path, invisibly. When any name or label was changed, the path carries an attribute "renamed", a data frame with the columns kind ("name", "variable label" or "value label"), column, old and new. An unchanged file returns silently with an empty attribute.

Text codes in Stata

Stata attaches value labels to numbers alone. In a .dta, a choice set coded with text, such as "a" and "b", is written as 1, 2, 3 in the set's declared order, each carrying its label, so the file reads back in Stata with the labels intact. A .sav keeps the text codes, which SPSS labels directly. Codes such as "01" and "02" are text, as read_responses() reads them, since as numbers they would lose their leading zero and differ from the instrument.

How names and labels are adapted for each format

SPSS and Stata use column names that start with a letter and stay within their format limits of 64 bytes for SPSS and 32 characters for Stata. Matrix row columns such as session__Reading materials are renamed to valid names, with illegal characters turned into _, a leading v where the first character is other than a letter, and a number added where 2 names would otherwise match. Variable labels are cut at 256 bytes for SPSS and 80 characters for Stata, and value labels at 120 bytes for SPSS and 80 characters for Stata. Everything else keeps its full label. 1 message lists each change as old -> new.

Examples

# \donttest{
demo <- sframe_demo("two_group")
out <- file.path(tempdir(), "two_group.sav")
if (requireNamespace("haven", quietly = TRUE)) {
  export_responses(demo$responses, demo$instrument, out)
  sframe_show(out)
}
# }