R/T_hss_table_single.R

In RenRMT/hsstools: Simplify the HSS Data Process

#' Create contingency table for single-response question
#'
#' Create contingency tables for HSS questions with a single response option and
#' a cross-variable. Tables can be created with either counts or percentages and the number
#' of significant digits for percentages can be set. Tables created with this function
#' come with a couple of additions according to HSS preferences. Most notably: sums are
#' added over columns, and N-values are added to column headers.
#'
#' @param df A dataframe containing the variable of interest and grouping variable.
#' @param var A character string with the variable name of interest.
#' @param group A character string with the grouping (or disaggregation) variable.
#' @param percent Set to TRUE to show percentages. Set to FALSE to show counts.
#' @param digits The number of (significant) digits to display. Trailing zeroes are
#' always removed. Note that 'digits' does not mean 'decimals', so digits = 3 will display as 'mm.d' not 'mm.ddd'
#'
#' @return A contingency table with the variable of interest and grouping variable.
#' @seealso `T_hss_table_multi`
#' @export
#'
T_hss_table_single <- function(df, var, group, percent = TRUE, digits = 1) {

  # create table
  if (percent == TRUE) {
    # creates contingency table with 'total' column
    table <- table(
      haven::as_factor(df[[var]]),
      haven::as_factor(df[[group]])
    ) %>%
      addmargins(margin = 2) %>%
      # display as proportions and convert to percentage
      proportions(margin = 2) * 100
    # format percentage to show specified digits, add % sign. (this converts all
    # values to character)
    table <- table %>%
      formatC(digits = digits, format = "fg", mode = "real") %>%
      as.double() %>%
      sprintf("%s%%", .) %>%
      matrix(
        nrow(table),
        dimnames = dimnames(table)
      ) %>%
      as.data.frame()

    # removes 'refused to answer' category from grouping variable if present
    # table <- dplyr::select(table, !contains("refused"))
  } else if (percent == FALSE) {
    # create contingency table with 'total' column
    table <- table(
      haven::as_factor(df[[var]]),
      haven::as_factor(df[[group]])
    ) %>%
      addmargins(margin = 2)
  } else {
    stop("Invalid input for percent:", percent)
  }

  # add rownames (response options) as column and convert to dataframe and rename sum col
  table <- dplyr::bind_cols(
    !!var := rownames(table),
    as.data.frame.matrix(table, row.names = NULL),
  ) %>%
    ## NB: generally tables don't show the 'refused' answers as grouping category,
    ## but excluding it causes issues with calculation of N values. commented out for now
    # dplyr::select(!contains("Refused")) %>%
    dplyr::rename(Total = Sum)

  rownames(table) <- NULL

  #apply N value labels to column headers
  names(table) <- paste0(
    names(table),
    .get_nval_single(df, var, group)
  )

  return(table)
}