This is a convenience function for removing what might be considered non-answers ("don't know", "refused", etc.) and rescaling the remaining values to add to 1.0.
sub_nonanswers(
data,
response = response,
value = value,
nons = c("Don't know", "Refused"),
factor_response = TRUE,
rescale = FALSE
)A data frame
Bare column name of where responses are found, including those considered to be non-answers. Default: response
Bare column name of values, Default: value
Character vector of responses to be removed. Default: c("Don't know", "Refused")
Logical: if TRUE (default), returns response variable
as a factor. This is likely a more useful way to handle response
categories once non-answers have been removed.
Logical: if TRUE, values will be scaled based on their total. If FALSE (the default), values are scaled based on an assumption that all responses add to 1. In some cases, crosstabs with heavy rounding might not add up to 1 when they should, so rescaling helps handle that.
A data frame with the same number of columns as the original, but fewer rows
if (interactive()) {
xt <- system.file("extdata/test_xtab2018.xlsx", package = "cwi")
df <- read_xtabs(xt, process = TRUE) |>
dplyr::filter(code == "Q1") |>
sub_nonanswers()
}