R/g.R

In handyReport: Handy package for reports

#' Calculate the Hedge's g statistic
#'
#' @description The function calculate the Hedge's g statistics, which is the effect size measure.
#' @param ordinal The (binary) norminal or ordinal variable.
#' @param interval The numberical (interval) variable.
#' @param correct Wheter to apply correction for chance.
#' @examples
#' g(ordinal = mtcars$vs, interval = mtcars$gear, correct = TRUE)
#' @author Marjan Cugmas
#' @importFrom stats complete.cases
#' @export

g <- function (ordinal, interval, correct = TRUE) {
  tmp <- stats::complete.cases(cbind(ordinal, interval))
  ordinal <- ordinal[tmp]
  interval <- interval[tmp]
  gs <- diff(by(data = interval, ordinal, mean))/(sum((by(data = interval,
                                                          ordinal, length) - 1) * by(data = interval, ordinal,
                                                                                     sd))/(length(interval) - 2))
  if (correct == TRUE) {
    a <- length(interval) - 2
    gs <- ((gamma(a/2))/(sqrt(a/2) * gamma((a - 1)/2))) *
      gs
  }
  return(gs)
}