Reference Files/dfba_interval_BF.R

In danbarch/dfba: What the Package Does (One Line, Title Case)

#' Beta Interval Bayes Factor
#'
#'
#' @param n1 Number of values for variable 1
#' @param n2 Number of values for variable 2
#' @param lowerH0value (Optional) Lower limit of interval estimate for the null hypothesis (default is 0)
#' @param upperH0value (Optional) Upper limit of interval estimate for the null hypothesis (default is 0.5)
#' @param a0 (Optional) Shape parameter alpha for the prior beta distribution (default is 1)
#' @param b0 (Optional) Shape parameter beta for the prior beta distribution (default is 1)
#' @return \item{H0_range}{Interval estimate of the null hypothesis -- either the default of [0, 0.5] or the value specified by user}
#' @return \item{H0_prior}{Prior probability of the null interval}
#' @return \item{H1_prior}{Prior probability of the alternative interval}
#' @return \item{H0_posterior}{Posterior probability of the null interval}
#' @return \item{H1_posterior}{Posterior probability of the alternative interval}
#' @return \item{Direction}{Statement describing the numerator and denominator of Bayes Factor estimate}
#' @return \item{BF}{Bayes Factor estimate}
#'

dfba_interval_BF<-function(n1,
                           n2,
                           lowerH0value = 0,
                           upperH0value = .5,
                           a0=1,
                           b0=1){
  if ((lowerH0value < 0)|(lowerH0value > 1)|(lowerH0value >= upperH0value)|(upperH0value > 1)) {
    lowerH0value = 0
    upperH0value = .5}
#  else {}
  rangeH0<-c(lowerH0value,
             upperH0value)
  if ((a0<=0)|(b0<=0)) {
    a0=1
    b0=1
  }
#  else {}
  if ((n1<0)|(n2<0)){
    stop("Both n1 and n2 must be nonnegative")} else {
      a = n1+a0
      b = n2+b0
      pH0up = pbeta(upperH0value,
                    a0,
                    b0)
      pH0lo = pbeta(lowerH0value,
                    a0,
                    b0)
      pH0 = pH0up-pH0lo
      pH1 = 1-pH0
      postH0up = pbeta(upperH0value,
                       a,
                       b)
      postH0lo = pbeta(lowerH0value,
                       a,
                       b)
      postH0 = postH0up-postH0lo
      postH1 = 1-postH0
      if (pH0 >= postH0){
        Direction<-"Bayes factor is for alternative over null"
        if(postH0==0) {
          BF10<-Inf
#          message("BF10 approaches infinity")
          } else {
            BF10=((1/postH0)-1)/((1/pH0)-1)
            BF<-BF10
        }
        } else {
          Direction<-"Bayes factor is for null over alternative"
          if (postH0==1){
            BF01<-Inf
#            message("BF01 approaches infinity")
            } else {
              BF01=((1/pH0)-1)/((1/postH0)-1)

          BF<-BF01
            }
          }
    }
  dfba_interval_BF_out<-list(H0_Range = rangeH0,
                             H0_prior = pH0,
                             H1_prior = pH1,
                             H0_posterior = postH0,
                             H1_posterior = postH1,
                             Direction = Direction,
                             BF = BF
       )
}