hilgard-functions.R

In Joe-Hilgard/hilgard: Miscellaneous Hilgard Convenience Functions

#F2R, t2R, r.CI
# made by joe hilgard in early 2014 *flexes biceps*








Cramer = function(chisq, n, k=2) {
  v = sqrt(chisq/(n*(k-1)))
  return(v)
}

# Standard error according to
# http://stats.stackexchange.com/questions/8487/how-do-you-calculate-confidence-intervals-for-cohens-d
stderr.d = function(d, n1, n2) {
  term1 = (n1+n2)/(n1*n2) + d^2 / (2*(n1+n2-2))
  term2 = (n1+n2)/(n1+n2-2)
  return(sqrt(term1*term2))
}

# Odds Ratio into Cohen's d
# Hasselblad & Hedges (1995) technique
OR.to.d = function(OR=NULL, b=NULL) {
  if (!is.null(OR)) b1 = log(OR)
  if (!is.null(b) & !is.null(OR)) if (b1 != b) print("Nonmatching OR and b! One or the other, please.")
  if (is.null(b)) b = b1
  return(b * sqrt(3)/pi)
}


d2r = function(d, n1, n2, width=.95) {
  r = d / sqrt(d ^ 2 + 4)
  term1 = (n1+n2)/(n1*n2) + d^2/(2*(n1+n2-2))
  term2 = (n1+n2)/(n1+n2-2)
  StdErr.d = sqrt(term1*term2)
  a = ((n1+n2)^2)/(n1*n2)
  StdErr.r = sqrt(a^2 * StdErr.d ^ 2 / ((d ^ 2 + a) ^ 3))
  return(list("r"=r, "StdErr.r"=StdErr.r))
}

d2r2z = function(d, n1, n2) {
  r = d2r(d, n1, n2)[[1]]
  StdErr.r = d2r(d, n1, n2)[[2]]
  FisherZ = 0.5 * log((1 + r) / (1 - r))
  StdErr.z = StdErr.r / (1 - r ^ 2)
  return(list("Z" = FisherZ, "StdErr.z" = StdErr.z))
}

d2t = function(d, n1, n2) {
  t = d * sqrt(n1 + n2 - 2) / 2
  return(t)
}