Fill in missing and non-missing values across interrelated survey questions

Usage

qualtrics_define_missing(
  df,
  question_code_include,
  question_code_omit = NULL,
  default_values = list("No", 0, as.Date(0)),
  predicate_question = NULL,
  predicate_question_negative_value = NULL
)

Arguments

df: A dataframe of survey responses
question_code_include: A regex that matches the columns to include in the missing non-missing value imputation
question_code_omit: A regex that matches the columns to omit from the missing non-missing value imputation
default_values: A list of length three, specifying the default, non-missing values to be used for character, numeric, and Date columns, respectively
predicate_question: Optional. The name of a single column that controls whether columns selected with question_code_include
predicate_question_negative_value: If predicate_question is specified, provide the value that indicates a negative response to the predicate question. For responses where the predicate question has this value, this value will be imputed to the specified columns

Value

A tibble containing only the columns selected by question_code_include (excluding those matching question_code_omit), with missing values handled according to the following logic:

Without predicate_question: If all selected columns are NA for a row, values remain NA. If any selected column has a non-NA value, NA values in other selected columns are replaced with the appropriate default value from default_values based on column type.
With predicate_question: If the predicate question is NA, all selected columns are set to NA. If the predicate question equals predicate_question_negative_value , all selected columns are set to the appropriate default value. Otherwise, original values are preserved.

Column types and their default value mappings: character uses default_values[[1]], numeric uses default_values[[2]], and Date/POSIXct uses default_values[[3]].