R/report.multipleChoice.R

In handyReport: Handy package for reports

#' Report the distributions of several dichotomous variables
#'
#' This function reports the frequencies of a specified event in multiple variables of a dataset.
#' It converts the event of interest to numeric, calculates the frequency of the event,
#' the number of valid and invalid entries for each variable, and optionally attaches variable labels.
#'
#' @param data A data frame or an object of class \code{survey.design} from \code{survey} package.
#' @param varNames A vector of strings representing the names of the variables in the data frame.
#' @param dec.freq Number of decimal places for frequencies. Defaults is \code{0}. Only used when an object of class \code{survey.design} is specified with the parameter \code{data}.
#' @param language The language used for displaying the statistics in the frequency table. This parameter accepts two values: \code{english} or \code{slovene}. Depending on the chosen language, all statistical terms and output will be adjusted accordingly. Default is \code{english}.
#' @param event A value (numeric or character) representing the event of interest to be reported.
#'              This event is transformed to numeric 1 in each specified variable.
#'              Default is \code{1}.
#' @param attachLabels Logical, if \code{TRUE}, attempts to attach variable labels from the attributes of the data frame.
#'                     Default is \code{FALSE}.
#'
#' @return A data frame with columns representing:
#'         1) the frequency of the event,
#'         2) the percentage of the event,
#'         3) the number of valid (non-NA) entries,
#'         4) the number of invalid (NA) entries for each specified variable.
#'
#' @examples
#' data <- data.frame(A = c(1, 1, 2, NA), B = c(1, 2, 2, 1), C = c(NA, 1, 1, 1))
#' varNames <- c("A", "B", "C")
#' report.multipleChoice(data, varNames)
#' @author Marjan Cugmas
#' @importFrom stats t.test aov chisq.test oneway.test sd t.test as.formula formula
#' @importFrom survey svyby svymean
#' @export

report.multipleChoice <- function(data, varNames, event = 1, attachLabels = FALSE, dec.freq = 0, language = "english") {
  utezi <- any(class(data) %in% c("survey.design2", "survey.design"))
  if (!utezi){
    for (i in varNames) data[,i] <- as.numeric(data[,i]== event)
    frekvence <- colSums(data[,varNames], na.rm = TRUE)
    veljavne <- apply(data[,varNames], 2, FUN = function(x) sum(!is.na(x)))
    neveljavne <- apply(data[,varNames], 2, FUN = function(x) sum(is.na(x)))
    ret <- data.frame(varNames,
                      frekvence,
                      round(frekvence/veljavne*100),
                      veljavne, check.names = FALSE,
                      neveljavne)
    colnames(ret) <- c("Category", paste0("Frequency (", event, ")"),
                       paste0("Share (", event, ")"),
                       "n", "NA")
    if (attachLabels) try({ret$Category <- attributes(data)$variable.labels[varNames]}, silent = TRUE)
    if (language %in% c("Slovene", "slovene", "slo", "s")) {
      colnames(ret) <- c("Kategorija", paste0("Frekvenca (", event, ")"),
                         paste0("Dele\u017e (", event, ")"),
                         "n", "Brez odgovora")
    }
  }
  if (utezi){
    frekvence <- NULL
    veljavne <- NULL
    for (i in varNames) {
      tmp <- survey::svytable(stats::as.formula(paste("~", i)), data)
      frekvence <- c(frekvence, round(tmp[event], dec.freq))
      veljavne <- c(veljavne, round(sum(tmp), dec.freq))
    }
    ret <- data.frame(varNames,
                      frekvence,
                      round(frekvence/veljavne*100),
                      veljavne, check.names = FALSE)
    colnames(ret) <- c("Category", paste0("Frequency (", event, ")"),
                       paste0("Share (", event, ")"),
                       "n")

    if (attachLabels) try({ret$Category <- attributes(data$variables)$variable.labels[varNames]}, silent = TRUE)
    if (language %in% c("Slovene", "slovene", "slo", "s")) {
      colnames(ret) <- c("Kategorija", paste0("Frekvenca (", event, ")"),
                         paste0("Dele\u017e (", event, ")"),
                         "n")
    }
  }

  return(ret)
}