Modify several different aspects of a binary sequential boundary. Most modifications do not change the stopping boundaries. The exceptions are 'missN', which allows modifications for missing assessments, and 'closeout', which allows for early stopping of the trial for administrative reasons (i.e., reasons that do not depend on the responses in the trial). Other modifications possible: level of the confidence intervals (using tsalpha, conf.level, or alternative), which boundaries are binding (i.e., can change from a boundary with binding futility boundaries to one with non-binding futility boundaries), null hypothesis value (theta0), and whether the confidence intervals should match the non-binding futility p-values on the superiority boundaries (cipMatch).
modify(b, missN = NULL, theta0 = NULL, tsalpha = NULL,
conf.level = NULL, alternative = NULL, cipMatch = TRUE,
binding = NULL, closeout=NULL, ...)
an object of class boundEst
an object of class boundEst
a vector of missed assessments
null hypothesis probability of success
vector of length 2 with nominal significance level, if not NULL overrides conf.level
and alternative
confidence level, ignored if tsalpha is not NULL
character, alternative hypothesis, either 'less', 'greater' or 'two.sided'
logical, for non-binding futility boundaries, should CI match the p-values on the binding boundary
character, which sides are binding: 'both', 'upper', or 'lower'
total number of trials at early closeout
other parameters passed
b<-designOBF(50)
bmod<-modify(b,missN=30:36)
par(mfrow=c(2,1))
plot(b)
plot(bmod)
Run the code above in your browser using DataLab