Get Pocock or OBF-style boundaries at the design stage

get.boundaries.design(
  rates,
  obf,
  unequal_type = FALSE,
  rho = 1,
  change = 0,
  algorithm = Miwa(steps = 1000)
)

Arguments

rates: A vector of information rates (between 0 and 1)
obf: Whether to use OBF (TRUE) or Pocock (FALSE)
unequal_type: Correction for unequal sample sizes across stage
rho: Fraction of variance explained by fitting ANCOVA.
change: A vector indicating which stages use ANOVA v. ANCOVA.

Examples

# Information fractions
t <- 1:3/3

# OBF-type boundaries
get.boundaries.design(rates=t, obf=TRUE)
#> [1] 3.471090 2.454432 2.004035
get.boundaries.design(rates=c(0.3, 0.9, 1.0), obf=TRUE)
#> [1] 3.412065 2.412694 1.969957
get.boundaries.design(rates=c(0.3, 0.9, 1.0), obf=TRUE, unequal_type=TRUE)
#> [1] 3.699929 2.136155 2.026535

# ANCOVA at last stage, R^2 = 0.5
get.boundaries.design(rates=c(0.3, 0.9, 1.0), obf=TRUE, unequal_type=TRUE, rho=sqrt(0.5), change=c(0, 0, 1))
#> [1] 3.895111 2.248843 2.133440

# ANCOVA at last two stages, R^2 = 0.5
get.boundaries.design(rates=t, obf=TRUE, rho=sqrt(0.5), change=c(1, 0, 0))
#> [1] 3.473523 2.456152 2.005440