computeBounds {GroupSeq}R Documentation

calculate boundaries

Description

Function to calculate boundaries corresponding to particular type I error spending rates and times of interim analyses, see Lan & DeMets 83.

Usage

computeBounds(n, drift, alpha, phi, t, t2, OneOrTwoSidedBounds, whatSpendingFunctionIsUsed, boundsTruncation)

Arguments

n is the number of interim analyses
drift drift is the true mean of the process, the noncentrality - drift is set to zero, and so does not affect function
alpha is the total type I error.
phi optional Parameter referring to Power Family
t is the information times.
t2 is the 'second' time scale, used for the covariance of sequential statistics, usually number of events or subjects.
OneOrTwoSidedBounds indicates one or two sided boundaries.
whatSpendingFunctionIsUsed indicates type I error spending function.
boundsTruncation is the user selected truncation on integration limits.

Value

lowerBounds the vector of lower standardized boundaries
upperBounds the vector of upper standardized boundaries
probExit vector of exit probabilities
probDifference(i) probExit(i)-probExit(i-1)

Note

drift MAY NOT be changed without adding code!
This function is meant to be called in context of using groupseq

Author(s)

Roman Pahl

References

Lan & DeMets 1983 Biometrika http://www.biostat.wisc.edu/landemets/

See Also

groupseq, alphaByUseFunction stdDeviations, pnorm jointDensity, tailProbability, searchForBound

Examples

## fictious example
computeBounds(2,0,0.05,0,c(0.5,1.0),c(0.5,1.0),1,1,8)
##with following result:
#$lowerBounds
#[1] -8 -8
#
#$upperBounds
#[1] 2.537988 1.662070
#
#$exitProbabilities
#[1] 0.005574597 0.050000000
#
#$differencesExitProbabilities
#[1] 0.005574597 0.044425403

[Package GroupSeq version 1.0 Index]