rdci() now includes the Brown-Li ‘Jeffreys’ method
for binomial RD (noted to be a superior non-iterative method in Laud
& Dane 2014).
orpairci() now includes a second version of the
transformed SCAS method, omitting the ‘N-1’ adjustment. (The adjusted
version previously labelled ‘Transformed SCASp’ in the output is now
labelled ‘Transformed SCAS(N-1)’).
orpairci() now includes the Transformed Blaker
method, produced when cc is TRUE.
scoreci() and scasci() have improved
computation of quadratic solution for the skewness-corrected score,
avoiding spurious results observed in 0.13% of a simulation exercise
(#38, thanks to Vincent Jaquet for reporting the issue and proposed
solution).
orpairci(), rrpairci() and
rdpairci() have fixed special cases to avoid warnings due
to negative variance estimate or intervals output as NaN.
orci() now fixed to give CIs for OR instead of RR,
and fixed handling of special cases.
rdci(), rrci() and
orci() give a selection of confidence intervals for each
contrast for 2 independent binomial or Poisson rates.rdpairci(), rrpairci() and
orpairci() give a selection of confidence intervals for
each contrast for 2 paired binomial proportions.scorepairci() and
moverpairci() for the score and MOVER methods,
respectively, for paired contrasts, mirroring scoreci() and
moverci() for independent contrasts.pairbinci(), which is deprecated.rateci():x / n, instead of an estimate that is consistent with a 0%
CI. std_est allows the user to choose. (#32, thanks to
Chelsea Dickens for raising the issue.)cc is TRUE, and
approximately match the corresponding mid-p intervals when
cc is FALSE.precis for setting the precision of the exact and mid-p
method.pairbinci() is soft-deprecated, and is replaced by
scorepairci() for score methods, moverpairci()
for MOVER methods, and rdpairci(), rrpairci
and orpairci() for other methods.rateci() and
moverci():rdpairci() with no discordant
pairs).exactci():moverci():clusterpci() for CI and test for a single
binomial proportion from clustered data.pairbinci():skew for skewness correction.bcf for variance bias correction.method_RD, method_RR and
method_OR are replaced with method.cc uses a new form of correction for RR giving
equivariant intervals. Also allows consistency with the
continuity-corrected McNemar test (or an intermediate correction of the
user’s choosing). cctype is deprecated.scaspci():bcf option now implemented for contrast = “p” (default
= FALSE).bign allows a different sample size to be used in the
bias correction (used within transformed SCASp method for paired OR in
pairbinci, for consistency with ‘N-1’ test).scoreci():bcf option now implemented for contrast = “p” (default
= FALSE).precis argument is improved for RR
and OR contrasts.exactci():pairbinci():cc continuity correction is now available for all
methods for all contrasts.cctype controls the type of correction to apply for
contrast = “RR”.method_RD = “Score_closed” for
non-iterative calculation of the Tango score interval for
contrast = “RD”. Thanks to Tony Yang for permission to use
the code in his 2013 paper.method_RR = “Score_closed” for
non-iterative calculation of the Tang score interval for
contrast = “RR”. Thanks to Guogen Shan for contributing
code via email.method_RD = “MOVER” and
method_RR = “MOVER”. Also “MOVER_newc” incorporates
Newcombe’s correlation correction.moverbase, for specifying different versions of
the MOVER methods (Wilson, Jeffreys, midp or SCAS).method_OR options for
transformed binomial methods for OR.scoreci():moverci():type = “wilson”.type = “SCAS” and “midp”
intervals.scoreci():moverci():moverci() with distrib = “poi” and
type = “wilson”]scoreci():tdasci()).Stheta = (p1hat - p2hat * theta) / p2d
(see Tang 2020)tdasci():scoreci() corrected for
distrib=“poi”.scoreci() for calculation of stratum CIs
with random=TRUE.scoreci() for distrib = “poi” and contrast
= “p” (#7).scaspci().rateci() for closed-form calculation of
continuity-corrected SCAS.scoreci() for stratified zero scores
calculated as NA, resulting in UL = 0. (Thanks to Lidia Mukina for
reporting the bug.)scoreci() for OR SCAS method
(derived from Gart 1985).pairbinci().scaspci() for non-iterative SCAS methods for
single binomial or Poisson rate.rateci() for selected methods for single binomial
or Poisson rate.pairbinci() for contrast=“OR”.moverci() for contrast=“p” and
type=“wilson”.scoreci()scoreci().pairbinci() for all comparisons of paired
binomial rates.scoreci().scoreci().scoreci() output
when stratified = TRUE.moverci().tdasci() wrapper function.moverci().moverci() to posterior
median for type = “jeff”, to ensure consistent calculations with
informative priors.