Transitional release. It adds the unified interface and deprecates the earlier one; nothing is removed in this version, and every function exported by 1.0.2 still works exactly as before.
rule() constructs a decision rule. The test statistic
(test), the monitoring scheme (stopping) and
the optimality criterion (criterion) are arguments, so one
function replaces the eleven separate decision-rule functions.op() evaluates a design by simulation, taking the
object returned by rule() and reporting both hypotheses in
a single call, with the futility and superiority stopping probabilities
separated and Monte Carlo standard errors on the rejection
probabilities.design_table() sweeps a set of tests and criteria and
returns a data frame.rule() and op() return classed objects
with named components and print() methods, so their output
need not be indexed by position.p1 and p2 are checked for numeric type, equal
length, at least two categories, absence of missing values,
non-negativity, and summation to one within 1e-6.Proportional_odds_assumption() explains itself and
warns instead of returning a bare NA.rule() detects a degenerate optimum and, by default,
substitutes the single-stage design; controlled by
on_degenerate.The functions below still work and still return exactly what they
always returned, but each now warns that it will be removed in version
1.1.0 and names its replacement. See
help("OptOTrials-deprecated").
ruleF() and ruleFS() become
rule(..., stopping = "F") and
rule(..., stopping = "FS").Decision_rule_S.F(), Decision_rule_M.F()
and Decision_rule_W.F() become
rule(..., test = "S"/"M"/"W", stopping = "F").Decision_rule_S.FS(), Decision_rule_M.FS()
and Decision_rule_W.FS() become
rule(..., test = "S"/"M"/"W", stopping = "FS").Decision_rule_S_1stage(),
Decision_rule_M_1stage() and
Decision_rule_W_1stage() become
rule(..., test = "S"/"M"/"W", stopping = "none").op.F(), op.FS() and
op.1stage() become op(design, nsim, seed),
where design is the object returned by
rule().Two habits change when converting a script. rule()
returns a named object, so res[3] becomes
res$n1, and the object is passed straight to
op() instead of its elements being transcribed.
op() evaluates both hypotheses in one call, so the old
pattern of calling the simulator twice, once with (p1, p2)
and once with (p1, p1), is no longer needed.
The printed operating-characteristic summary no longer reports a
Monte Carlo standard error keyed to a nominal level of 0.05, which was
wrong for any other choice of alpha. The table already
gives the standard error of each rejection probability, computed from
that probability.
The FS design’s continuation probability under the alternative,
beta3 = Pr(T1 <= t1s | Ha), was computed for the score
test without the factor n1 in the mean of the statistic.
The score statistic has mean theta * V * n, unlike the
rank-based statistics, whose means do not scale with the sample size, so
the omission made the value far too large. It entered the expected
sample size under the alternative and therefore the objective of
criteria 2, 3 and 5, so a score-test FS design could fail to minimise
what it was asked to minimise. The Mann-Whitney-Wilcoxon and win odds
expressions were already correct, and the F design does not use this
quantity.
Internal variable names now follow the article: t1f
and t1s for the interim futility and superiority
boundaries, and beta3 for
Pr(T1 <= t1s | Ha), in place of t1l,
t1u and beta2.
Typographical errors in the help-page titles of
QR_fun(), V_S.over.nk(),
Decision_rule_M.F(), ruleF() and
ruleFS().
The description of the value returned by op.F(),
op.FS() and op.1stage(): the first element is
the rejection probability, not the probability of an incorrect
decision.
Initial CRAN release.