R/priormul.R

In InDisc: Obtaining and Estimating Unidimensional and Multidimensional IRT Dual Models

priormul <- function(alpha, PHI, nquad, df, ls){

  n<-size(alpha)[1]

  mvari <- 0
  resi <- matrix(0,n,1)

  for (i in 1:n){
    tmp <- alpha[i,]
    tmp2 <-1 - tmp%*%PHI%*%transpose(tmp)
    tmp3 <- tmp%*%transpose(tmp)
    resi[i] <- tmp2/tmp3
  }

  tmpmin <- min(resi)
  mvari <- tmpmin
  resi <- resi - (mvari*matrix(1,n,1))

  #nodes

  if (missing(df)==TRUE || missing(ls)==TRUE){
    df <- 5
    ls <- round(mvari*3)
    if (ls == 0){
      ls <- 1
    }
  }

  nodth <- creanodos(-4,4,nquad)
  nodichi<- creanchi(0.02,4,nquad,df,ls)
  if (df == 5){
    var_nodos <- 0.4159 # the variance of the nodes
  }
  if (df == 6){
    var_nodos <- 0.3832
  }
  if (df == 10){
    var_nodos <- 0.2631
  }


  OUT<-list("mvari" = mvari, "resi" = resi, "nodth" = nodth, "nodichi" = nodichi, "var_nodos" = var_nodos)
  return(OUT)
}