gridFinalPvalue: P-value over the Ordering Space

In paulowhite/DelayedGSD: Planning and analysis of group sequential trial with pipeline data

P-value over the Ordering Space

Description

Evaluate the p-value over stages and test statistics, following armitage ordering (Armitage 1957).

Usage

gridFinalPvalue(
  object,
  continuity.correction,
  seq.futility = c(-3, 0),
  seq.kMax = c(-3, 1, 3, 6),
  seq.efficacy = c(2.5, 3.25, 4, 5),
  plot = TRUE,
  col = NULL,
  title = NULL,
  xlim = NULL,
  pch = 20,
  cex = 2,
  lwd = 2,
  digits = 2
)

Arguments

object
continuity.correction	[logical] whether to add the probability of stopping between ck and ck.uncorrected to ensure continuity of the p-value across stages. When used the p-value will always be greater than this probability of stopping bettwen ck and ck.uncorrected.

Value

a data.frame containing

• k the stage

• z the test statistic value

• p.value the p-value

References

P. Armitage, Restricted sequential procedures. Biometrika 1957 (9-56)

Examples

## 2 stages, no fixC
myBound2 <- CalcBoundaries(kMax = 2,  
                         alpha = 0.025, 
                         beta = 0.2,  
                         InfoR.i = c(0.55, 1.00),  
                         InfoR.d = c(0.62, 1),  
                         rho_alpha = 2,  
                         rho_beta = 2,  
                         method = 1,  
                         cNotBelowFixedc = FALSE,
                         bindingFutility= TRUE,
                         delta = 0.6)
grid.p2 <- gridFinalPvalue(myBound2)
grid.p2

## 2 stages, fixC
myBound2.fixC <- CalcBoundaries(kMax = 2,  
                         alpha = 0.025, 
                         beta = 0.2,  
                         InfoR.i = c(0.55, 1.00),  
                         InfoR.d = c(0.62, 1),  
                         rho_alpha = 2,  
                         rho_beta = 2,  
                         method = 1,  
                         cNotBelowFixedc = TRUE,
                         bindingFutility= TRUE,
                         delta = 0.6)
gridFinalPvalue(myBound2.fixC,
                continuity.correction = 0)
gridFinalPvalue(myBound2.fixC,
                continuity.correction = 1)
gridFinalPvalue(myBound2.fixC,
                continuity.correction = 2)

## 3 stages, no fixC
myBound3 <- CalcBoundaries(kMax = 3,  
                         alpha = 0.025, 
                         beta = 0.2,  
                         InfoR.i = c(0.45, 0.65, 1.00),  
                         InfoR.d = c(0.52, 0.72, 1),  
                         rho_alpha = 2,  
                         rho_beta = 2,  
                         method = 1,  
                         cNotBelowFixedc = FALSE,
                         bindingFutility= TRUE,
                         delta = 0.6)
grid.p3 <- gridFinalPvalue(myBound3)
grid.p3

## 3 stages, fixC
myBound3.fixC <- CalcBoundaries(kMax = 3,  
                         alpha = 0.025, 
                         beta = 0.2,  
                         InfoR.i = c(0.45, 0.65, 1.00),  
                         InfoR.d = c(0.52, 0.72, 1),  
                         rho_alpha = 2,  
                         rho_beta = 2,  
                         method = 1,  
                         cNotBelowFixedc = TRUE,
                         bindingFutility= TRUE,
                         delta = 0.6)
gridFinalPvalue(myBound3.fixC, digits = 3,
                continuity.correction = 0)
gridFinalPvalue(myBound3.fixC, digits = 3,
                continuity.correction = 1)
gridFinalPvalue(myBound3.fixC, digits = 3,
                continuity.correction = 2)

paulowhite/DelayedGSD documentation built on Nov. 1, 2023, 5:36 p.m.