| NEWS | R Documentation |
Function frdAllPairsConoverTest internally
calculated wrong Friedmans test statistics that lead to
wrong pairwise tests statistics and p-values.
This was fixed.
Function skillingsMackTest only worked as expected,
if NA values were explicitly given in vectors to match
the size of the matrix. This has been extended to work with
some sort of 'sparse matrices'.
Data type caste from integer to double precision in Fortran source was changed in order to allow portability.
The iterative procedure that was included
in function power.williams.test occassionally
failled to return reasonable values for n, when power and
other parameters were given (Instead, it returned the given
upper limit). The procedure was replaced with the analytical expression
as given in the PASS manual p. 595-2.
Function chenJanTest was added. Step-down p-adjustment
according to Chen (1999) was added for functions chenTest and
chenJanTest.
Function frdHouseTest was fixed in such a way
that y, groups and blocks can be passed as vectors.
Functions chenTest, Dunnett,
frdHouseTest, power.dunnett.test,
power.tukey.test and power.williams.test
were added.
Code in init.c was re-written based on
tools::package_native_routine_registration_skeleton.
Function dscfAllPairsTest always returned NA.
This has been fixed.
Function manyOneUTest always returned NA.
This has been fixed.
An order restriction for factor levels
was implemented in the function jonckheereTest,
which allows an un-sorted input.
The Rd-files were updated for all trend-test procedures with a note on how to label the factor levels from zero-dose control to highest dose level.
The function manyOneUTest was corrected
in order to become workable.
The function mrrTest to conduct
Madhava Rao-Raghunath exact symmetry test was added.
The functions hayterStoneTest,
hsAllPairsTest and NPMTest were extended for
calculating asymptotic p-values with the code
of pHayStonLSA of the package NSM3.
The Shan-Young-Kang test for testing against
ordered alternatives
was implemented in function shanTest.
The function uryWigginsHochbergTest was
corrected in order to return the two-tailed p-value.
In the demo algae the step-down
trend examples were replaced with
the corresponding novel function
stepDownTrendTest.
The method for class "aov" was implemented
for the function williamsTest.
The function toTidy was extended for
functions of two-way rank ANOVA.
Several Rd files were updated and extended.
The functions hayterStoneTest
and hsAllPairsTest were implemented.
The function toTidy was implemented
based on a valuable recommendation of Indrajeet Patil.
The method for class "aov" was implemented
for the functions osrtTest and MTest.
The function stepDownTrendTest as well as
corresponding print and summary methods were
implemented to perform step-down tren test procedures.
the functions hayterStoneTest, hsAllPairsTest
and NPMTest were extended for method = "boot",
with which an asymetric permutation test is conducted
and p-values are returned.
The function chaAllPairsNashimotoTest was
enhanced for NPT'-test, where p-values are calculated from the
standard normal distribution and NPY'-test, where the statistic(s)
are compared with ciritical h-values.
The function chackoTest was extended for
alternative = c("greater","less").
The Rd-files of several functions were revised.
The flignerWolfeTest and
chackoTest were implemented
in the function powerOneWayTest.
The functions MTest and osrtTest were
corrected in such a way, that estimated statistics are
compared with critical h-values from an internal look-up
table.
The function chaAllPairsNashimotoTest was corrected
for the NPM-test. p-values are now calculated from the
standard normal distribution.
A new generic method print.osrt and
summary.osrt was coded for all functions that use
critical values for decisions, i.e. osrtTest,
MTest, steelTest, williamsTest and
shirleyWilliamsTest. Other print and
summary methods were removed.
The Fligner-Wolfe test was implemented in function flignerWolfeTest.
kwManyOneNdwTest was corrected for directed hypothesis testing.
S3 method for class‘aov’ was added
for tukeyTest, gamesHowellTest,
lsdTest, tamhaneT2Test,
uryWigginsHochbergTest, dunnettT3Test,
snkTest, scheffeTest, dunnettTest,
tamhaneDunnettTest and welchManyOneTTest.
Updated concept-tag in Rd files.
The examples in kruskalTest, adKSampleTest
and bwsKSampleTest were
changed for compatibility with
the upcoming ‘stringsAsFactors = FALSE’ default in R (4.0.0).
The names for the ‘method’ were corrected for adAllPairsTest and bwsAllPairsTest.
The example in adKSampleTest was corrected.
The S3 method for class ‘formula’ was added
for frdAllPairsNemenyiTest.
The novel internal function frdRanks was added
for consistent Friedman-type ranking.
Results of all frd[.]Test and quadeAllPairsTest depended on the sequential
order of groups and blocks. This has been fixed.
Thanks to Jakob Richter, email 2019-10-15.
The function frdManyOneExactTest was
corrected towards ‘alternative == "two.sided"’ only.
The function hartleyTest for performing
Hartley's Maximum F-Ratio Test of Homogeneity of Variances
was added.
The demo trout.R was updated.
The function siegelTukeyTest for
performing the Siegel-Tukey rank dispersion test
was added.
The function GSTTest for
performing the Generalized Siegel-Tukey Test
for rank dispersion was added.
The names of columns and rows of the lower
triangle matrix in any of the functions
frdAllPairs*, frdManyOne* and
quadeAllPairsTest were occasionally wrong.
The bug only appeared, if y, group, block
were provided as vectors and the group vector was
sorted. This bug has been fixed.
Example of Rd file of kruskalTest
was modified to show equality of Conover's F and
one-way ANOVA F-Test with ranks.
The function snkTest for performing
the Student-Newman-Keuls test was added.
The funtion duncanTest for performing
the Duncan's (new) multiple range test was added.
The funtion welchManyOneTTest for performing
many-to-one comparisons with Welch's t-test was added.
Likewise, welchManyOneTTest was added as test
in function powerMCtests
The funtion steelTest including
print and summary S3-methods for applying
Steel's Many-to-One Rank Test was added.
The function tamhaneT2Test was extended
for performing Tamhane's extended T2' test as well. Thanks to
Sirio Bolaños.
Several Rd files were corrected (e.g. links to other man files, corrected calls to functions in examples section)
The function tamhaneT2Test was corrected in order
to compute the two-tailed p-value with Sidaks p-adjustment using
m = k(k-1)/2.
The function shirleyWilliamsTest was extended
for alternatives less and greater using a look-up table
for critical Williams-t values including
print and summary methods.
Links in Rd file of bwsManyOneTest were corrected.
The function williamsTest
including print and summary methods were added.
Misspelled author name Terpstra was
corrected in function jonckheereTest
and corresponding Rd file.
Reference in the title of Rd files
for functions mandelhTest and mandelkTest
were corrected.