select.rule:
Selection Rules for Interim Analysis in ASD
Description
Function select.rule provides a number of options for selecting treatments at an interim analysis in ASD.
Usage
select.rule(x, type = 0, epsilon = 1, thresh = 1)
Arguments
x
Vector of test statistics.
type
Decision rule type; 0, 1, 2, 3, 4, 5 or 6 (see below for details); default is 0.
epsilon
For type = 4, set epsilon criterion
thresh
For type = 6, set threshold criterion
Value
select
Indicator vector that shows treatments selected (1) or not selected (0)
z
Vector of same length as select set to -Inf if not selected and 0 otherwise. For use with function dunnett.test
Details
There are seven types of selction rule available:
(0) Select all treatments
(1) Select one treatment; largest value of x
(2) Select two treatments; two largest values of x
(3) Select three treatments; three largest values of x
(4) Epsilon rule; select all x within epsilon of maximum
(5) Randomly select one treatment
(6) Threshold rule; select all x larger than thresh