R/08_estimate_bias.R

In nutriverse/anthrocheckr: Anthropometric Measurement Standardisation Tests

#'
#' Calculate bias of a measurement
#'
#' Calculate the bias of a measurement against a gold standard. Two gold
#' standards are used in this function: 1) measurements made by a supervisor or
#' a known expert; and, 2) median of all measurements made by the observers.
#'
#' @param msur A numeric value or vector of mean measurements from observers
#' @param msup A numeric value or vector of mean measurements from supervisor
#' @param mall A numeric value or vector of the median of measurements from
#'     observers
#'
#' @return A numeric value or vector signifying bias
#'
#' @examples
#' x <- subset(smartStdLong, measure_type == "height")
#'
#' y <- calculate_mean(
#'   measures = x$measure_value, index = x$observer
#' )
#'
#' z <- calculate_mean(x$measure_value)
#'
#' estimate_bias(msur = y$Freq[x$Var1 != 0],
#'               msup = y$Freq[x$Var1 == 0],
#'               mall = z)
#'
#' @export
#'

estimate_bias <- function(msur, msup, mall) {
  ## Bias from supervisor
  bias_super <- abs(msur - msup) / msup

  ## Bias from median
  bias_med <- abs(msur - mall) / mall

  ## Concatenate into a data frame
  bias <- data.frame(bias_super, bias_med)

  ## Return output
  bias
}