R/constEffect.R

In randomizationInference: Flexible Randomization-Based Inference

# constant, additive treatment effect for factors at equally-spaced levels
# treatment indicators in 0s and 1s
# tau can be a vector

constEffect <- function(y, w, w_new, poOptions) {
  # formatting
  if (sum(factor(w) == w) > 0) w <- data.matrix(w)
  if (ncol(w) == 1) w <- as.vector(w)
  # calculations
  if (is.vector(w)) {
    y + t(t((w_new - w)) * poOptions$tau)
  } else {
    y + rowSums(t(t(w_new - w) * poOptions$tau))
  }
}