R/genvecs.R

In extraSuperpower: Power Calculation for Two-Way Factorial Designs

#' Internal function to generate a vector of coefficients to be applied to the reference mean to generate an experimental model.
#'
#' The vector's length is length equal to the number of levels of the experimental factor, defined in reps.
#' he coefficients are estimated from a user defined change. Default is to generate linear vector, this can be modified with the
#'
#' @importFrom methods is
#'
#' @noRd

## quiets concerns of R CMD check re: the .'s that appear in pipelines
if(getRversion() >= "2.15.1")  utils::globalVariables(c("refmean", "n", "power", "effect", "y", "mean.group.size", "lower.bound.ci", "upper.bound.ci"))


genvecs <- function(change, reps, bystart=TRUE, scaler=refmean)
{
  if(reps%%1>0 | reps<=0)
  {stop("Number of factor levels (nlfA and nlfB) must be positive integers")}
  if(!is(change, "numeric"))
  {stop("Effect sizes (fAeffect and fBeffect) must be numeric")}
  if(change<0)
  {stop("Effect sizes (fAeffect and fBeffect) must be positive")}
  if(!bystart)
  {
    reps <- reps-1
    increments <- scaler*change - scaler
    increments <- increments*(1:reps)
    increments <- c(1, 1+ (increments/scaler))
  } else if (bystart)
  {
    increments <- seq(1, change, length.out=reps)
  }
  increments
}