R/ci2p.R

In depigner: A Utility Package to Help you Deal with "Pignas"

#' ci2p
#'
#' \code{ci2p} compute the p-value related with a provided confidence
#' interval. It considers a symmetric distribution (by default standard
#' normal).
#'
#' Confidence level can be customize (by default 95%) and it is also
#' possible to apply a log transformation in case of proportions.
#'
#' @param est estimated value
#' @param lower lower bound of the confidence level
#' @param upper upper bound of the confidence level
#' @param log_transform (default `FALSE`) flag indicating if a log
#'                     transformation as to apply to the data
#' @param conf (default `95\%`) confidence level
#' @param qdist (default `qnorm`) quantile function
#' @param pdist (default `pnorm`) distribution function
#'
#' @return a p-value
#' @export
#'
#' @examples
#'
#' ci2p(1.125, 0.634, 1.999, log_transform = TRUE)
#' ci2p(1.257, 1.126, 1.403, log_transform = TRUE)
ci2p <- function(
                 est, lower, upper,
                 log_transform = FALSE,
                 conf = 0.95,
                 qdist = stats::qnorm,
                 pdist = stats::pnorm) {
  if (upper < lower) {
    tmp <- lower
    lower <- upper
    upper <- tmp
    rm(tmp)
    ui_warn("upper < lower: they are considered reversed")
  }

  if (log_transform) {
    est <- log(est)
    lower <- log(lower)
    upper <- log(upper)
  }

  se <- (upper - lower) / (2L * qdist(conf))
  1L - pdist(est / se)
}