GroupSeq (version 1.0)

computeAlphaLevel: Computes alpha level and other probabilities

Description

Computes alpha level and other probabilities for the number of tests of hypotheses. Method is a modification of Armitage, McPherson and Rowe. Consistent with Lan & DeMets 83, does not assume equally spaced information times.

Usage

computeAlphaLevel(n, t, t2, lowerBounds, upperBounds, drift, nMax)

Arguments

n

the number of analyses

t

the vector of analysis time point on (0,1).

t2

is the second or information time scale, for covariances

lowerBounds

the vector of upper bounds (standardized)

upperBounds

the vector of lower bounds (standardized)

drift

the noncentrality; at t=1, process has mean drift.

nMax

Number of interim analyses is limited to nMax=25.

Value

• probStopping(i)
  is the prob of reaching ith analysis and stopping.
• probExceedingUpper(i)
  is the prob of reaching ith and exceeding upper.
• probExceedingLower(i)
  is the prob of reaching ith and exceeding lower.
• expectedStoppingTime
  is expected stopping time (not implemented).
• probTotal
  is the total type I error used.

References

Armitage, McPherson & Rowe 1969 JRSSA Lan & DeMets 1983 Biometrika http://www.biostat.wisc.edu/landemets/

See Also

groupseq, stdDeviations pnorm, jointDensity

Examples

## fictious example 
computeAlphaLevel(3, c(0.33,0.66,1),c(0.33,0.66,1), c(-1,0,1), c(2,3,4),0.2, 25)
## with following result:
#  $probAndStop
# [1] 0.1621561 1.0000000 1.0000000
#
# $probAndExceedingUpper
# [1] 2.970758e-02 7.447682e-04 9.675729e-06
#
# $probAndExceedingLower
# [1] 0.1324485 0.3115237 0.3327052
#
# $expectedStoppingTime
# [1] 0
#
# $totalTypeOneError
# [1] 0.8071395
#