Classes for binary sequential boundaries

Description

There are several classes that represent binary sequential boundaries. The most simple is the abparms class, then comes the bounds class (which contains abparms), then comes boundNBF class (which contains bound class), or boundEst class (which contains bounds class), then comes boundNBFEst (which contains boundNBF). See details for which slots go with which classes.

Details

The simplest representation of a binary sequential boundary is the abparms class, represented by a vector of the total number of trials (Nk) where to stop, and denoting stopping when number of sucesses, S, is S>=b or S<= a. One sided boundaries can be represented by all NA values for either a or b. Often times a two-sided boundary treats one side as a superiority boundary which must be stopped if crossed (a binding boundary), while the other side of the boundary is a futility boundary which may be ignored (a non-binding boundary). For example when binding='upper', then p-values for the upper boundary are calculated as if the lower boundaries are ignored if crossed and stopping happens on the lower side at max(Nk) instead, while the p-values for the lower and end boundary points are calculated using all (lower,upper and end) boundaries.

Next is the bound class which adds the slots N (number of trials at each boundary point), S (number of sucesses at each point), K (number of ways to get to each point), order (ordering of points for p-value calculations), UL('upper','lower' or 'end').

Slots

Nk:

vector of number of samples at boundary stopping points

a:

vector for lower bound, stop if S out of Nk is less than or equal a. NA denotes do not stop.

b:

vector for upper bound, stop if S out of Nk is greater than or equal b. NA denotes do not stop.

binding:

character specifying which boundary section is binding, either 'both', 'upper', or 'lower'

alternative:

character specifying alternative, 'two.sided', 'less', or 'greater'

N:

vector of number of samples at boundary stopping points

S:

vector of number of sucesses at boundary stopping points

K:

vector of number of ways to get to each boundary point

order:

vector of ordering of points

UL:

character vector denoting part of boundary, either 'lower' or 'upper' or 'end'

estimate:

vector of estimates of theta, probability of success

lower:

vector of lower confidence intervals for theta

upper:

vector of upper confidence intervals for theta

conf.level:

confidence level associated with confidence intervals

alpha:

error on either side

theta0:

null value for theta

plower:

vector of lower p-values

pupper:

vector of upper p-values

pval:

vector of p-values as directed by alternative slot

Methods

There is a plot and a points method for boundEst objects.

Author(s)

Jenn Kirk, Michael P. Fay

Examples

1
new("abparms",Nk=200)