Movatterモバイル変換

News for Packagecoin

Changes in version 1.4-3 (2023-09-26)

Bug fixes

• Supplying case weights to tests for marginal homogeneity or symmetryhypotheses appeared to work, but returned incorrect test statisticsandp-values; it now results in an error.

User-visible changes

• The shift algorithm is now computationally more efficient and allowsfor user interrupt.

• Documentation updates.

Other changes

• The Fligner-Killeen test, the Conover-Iman test and the Quade testperformed transformations of the response variable prior to applyingthe influence function; these transformations are now done within theinfluence function.

Changes in version 1.4-2 (2021-10-07)

Bug fixes

• Partial matching of argumentteststat failed forformula methods.

User-visible changes

• Documentation updates.

Other changes

• Packagelibcoin\ge1.0-9 is now required.

Changes in version 1.4-1 (2021-02-08)

Bug fixes

• Fix LaTeX problem.

Changes in version 1.4-0 (2021-01-15)

New features

• The Peto-Peto test for right-censored data is now available (see?logrank_test and?logrank_weight).

• The Moore-Penrose inverse of the conditional covariance matrix cannow be extracted (see?covariance).

• For tests of conditional independence within blocks, i.e.,stratified tests, the partial linear statistic for each block andits conditional expectation and variance as well as its conditionalcovariance and corresponding Moore-Penrose inverse can now beextracted (see?statistic,?expectation,?variance and?covariance).

Bug fixes

• Exact multivariate two-sample tests with quadratic test statisticsappeared to work but returned incorrectp-values; it nowresults in an error.

• Extraction of the standardized linear statistic from objects ofclass"IndependenceLinearStatistic" resulted in an error.

User-visible changes

• Extraction of the conditional expectation or variance now returns amatrix instead of a vector (for consistency with linear statisticsandp-values).

• The test size is now computed more efficiently.

• Documentation updates.

Other changes

• The internal storage of the linear statistic and its conditionalexpectation and covariance, i.e., slotslinearstatistic,expectation andcovariance in classes inheriting from"IndependenceLinearStatistic", as well as the Moore-Penroseinverse of the conditional covariance, i.e., slotcovarianceplus in classes inheriting from"QuadTypeIndependenceTestStatistic", has been changed. Inparticular, the conditional covariance and its Moore-Penrose inversenow use a memory-efficient packed format.

• R\ge3.6.0 and packagelibcoin\ge1.0-5are now required.

Deprecated and defunct

• Classes"Variance","CovarianceMatrix" and"VarCovar" are now deprecated.

Changes in version 1.3-1 (2019-08-22)

Bug fixes

• Precision issues in point and interval estimates associated withlocation and scale problems.

User-visible changes

• The density, distribution and quantile functions now propagate anyattributes of thex,q andp arguments,respectively, to the results.

• Documentation updates.

Changes in version 1.3-0 (2019-03-04)

New features

• Ordered factors can now be transformed to a finite collection oforder-restricted scores (see?zheng_trafo); this can be usedto formulate score-independent tests for ordered categorical data(see?malformations,?jobsatisfaction and?vision).

• The test size, i.e., the actual significance level, of approximative(Monte Carlo) and exact tests can now be computed (see?size).

• The centered linear statistic can now be extracted (see?statistic).

• P-values can now be extracted from objects of class"PValue" usingpvalue().

Bug fixes

• Adjustedp-values based on the approximative (Monte Carlo)marginal null distributions were incorrect under some circumstances.

• The point and/or interval estimate of the ratio of scales parameter(for tests against scale alternatives) were incorrect when using theasymptotic null distribution.

• Exact distributions of symmetry tests for paired data, e.g., theexact sign and Wilcoxon signed-rank tests, used theStreitberg-Röhmel algorithm for independent samplesinstead of the more efficient version for dependent samples.

• The class definition of"PValue" erroneously included themidpvalue andpvalueinterval slots; these slots arenow part of the"NullDistribution" subclass (as they shouldhave been originally).

User-visible changes

• Asymptotic and exactp-values smaller than machine precision,i.e.,.Machine$double.eps, or approximative (Monte Carlo)p-values smaller than resampling precision, i.e.,1 / nresample, are now reported as such in the printedoutput.

• Computations on the approximative (Monte Carlo) null distribution,e.g., adjustedp-values for maximum-type tests, are now moreefficient.

• The Pearson chi-squared test is now computed more efficiently.

• Extracting the test statistic or the (global)p-value returneda named result under some circumstances; the results are nowunnamed.

• The quantile function now returns a named vector if itspargument is named.

• Objects of class"ApproxNullDistribution" now has a slotnresample containing the number of Monte Carlo replicates.

• Classes"ExpectCovar" and"ExpectCovarInfluence" havebeen moved to theparty package.

• The ‘coin_implementation’ vignette has been renamed to‘Implementation’.

• Documentation updates.

Other changes

• The linear statistic, its conditional expectation and covariance,and permuted linear statistics are now computed by packagelibcoin.

• confint() is now an S4 method.

• R\ge3.4.0 and packagesstats4 andmatrixStats are now required.

Deprecated and defunct

• TheB argument ofapproximate() andApproxNullDistribution() is now deprecated and has beenreplaced bynresample (for consistency with thelibcoin,party andpartykit packages).

• Class"PValue" is now deprecated.

Changes in version 1.2-2 (2017-11-28)

Other changes

• Thelibcoin regression tests are now inlibcoin 1.0-0.

Changes in version 1.2-1 (2017-07-16)

Bug fixes

• Precision issue in the quantile function of exact null distributionscomputed by the shift algorithm.

Changes in version 1.2-0 (2017-06-20)

New features

• The Gaugler-Kim-Liao class of tests for right-censored data is nowavailable (see?logrank_test and?logrank_weight).

• Thepvalue(),midpvalue() andpvalue_interval()methods for objects of class"NullDistribution" are nowvectorized.

Bug fixes

• Precision issue in the density function of approximative (MonteCarlo) null distributions. This also impacts mid-p-values andp-value intervals.

• Precision issue in the support of approximative (Monte Carlo) nulldistributions.

• Precision issue in the symmetric median scores transformation.

• The weighted logrank transformation returned nonsense scores whensupplied non-right-censored input; it now results in an error.

• The maximally selected statistics transformation for ordered factorsreturned the wrong labels under some circumstances.

User-visible changes

• Theties.method argument oflogrank_trafo() no longerallows using"logrank" and"HL" to specify mid-ranksand the Hothorn-Lausen method, respectively.

• Documentation updates.

Deprecated and defunct

• Functionsurv_test() is now defunct.

• Optionmethod = "discrete" inpvalue() is now defunct.

• Thedperm(),pperm() andqperm() methods forobjects of class"AsymptNullDistribution" are no longerneeded and have been removed.

Changes in version 1.1-3 (2016-11-24)

Bug fixes

• Objects of class"AsIs" were handled incorrectly in somecases, leading to, e.g., misleading error messages or teststatistics having the wrong sign.

• Supplying decreasing scores for ordered factors with two levelsusing the test functions'scores argument resulted inincreasing normalized scores, leading to test statistics having thewrong sign.

• The transformation function for ordered factors did not returnnormalized scores in the two-level case when argumentscoreswas used.

• Precision issues in confidence intervals for location and scaleproblems.

• When the general symmetry and marginal homogeneity tests wereapplied to an object of class"table", the levels of themeasurement conditions were not preserved (but reorderedalphabetically).

User-visible changes

• Documentation updates.

Other changes

• Packagemvtnorm\ge1.0-5 is now required.

Changes in version 1.1-2 (2015-11-13)

Bug fixes

• of_trafo() did not handle ordered factors of length one.

• Parallel operation (viaapproximate()) changed the randomnumber generator state upon first invocation unless packageparallel's namespace had already been loaded.

User-visible changes

• Documentation updates.

Changes in version 1.1-1 (2015-10-25)

Bug fixes

• Methodsinitialize() andshow() were not imported frompackagemethods.

User-visible changes

• Documentation and vignette updates.

Changes in version 1.1-0 (2015-08-14)

New features

• The Fisher-Yates correlation test (based on van der Waerden scores)is now available (see?fisyat_test).

• The quadrant test is now available (see?quadrant_test).

• The Koziol-Nemec test and its corresponding transformation functionare now available (see?koziol_test and?koziol_trafo).

• The Savage test and its corresponding transformation function arenow available (see?savage_test and?savage_trafo).

• The Taha test is now available (see?taha_test).

• The Klotz test and its corresponding transformation function are nowavailable (see?klotz_test and?klotz_trafo).

• The Mood test and its corresponding transformation function are nowavailable (see?mood_test and?mood_trafo).

• The Conover-Iman test is now available (see?conover_test).

• The weighted logrank test and its corresponding transformationfunction are now available, including weights for generalizedWilcoxon tests (Gehan-Breslow, Prentice, Prentice-Marek andAndersen-Borgan-Gill-Keiding), the Tarone-Ware class of tests, theFleming-Harrington class of tests and the Self class of tests (see?logrank_test,?logrank_trafo and?logrank_weight).

• The sign test for paired data is now available (see?sign_test).

• The Quade test is now available (see?quade_test).

• The Pearson chi-squared test, the generalizedCochran-Mantel-Haenszel test and the linear-by-linear associationtest now allow the alternative hypothesis to be specified in thedoubly ordered case, i.e., when both the response variable and theexplanatory variable are ordered factors (see?chisq_test,?cmh_test and?lbl_test).

• The Kruskal-Wallis test, the van der Waerden test, the Brown-Moodmedian test, the logrank test and the Friedman test now allow thealternative hypothesis to be specified when the explanatory variableis an ordered factor (see?kruskal_test,?normal_test,?median_test,?logrank_test and?friedman_test).

• The marginal homogeneity test now handles ordered responses and/orordered measurement conditions and allows the alternative hypothesisto be specified in the doubly ordered case (see?mh_test).

• The Brown-Mood median test and its corresponding transformationfunction now offer the choice of three different versions of themid-score (see?median_test and?median_trafo).

• The van der Waerden test, the Brown-Mood median test and theAnsari-Bradley test now handleK-sample problems (see?normal_test,?median_test and?ansari_test).

• The Fligner-Killeen test now handles two-sample problems (see?fligner_test).

• Maximally selected statistics can now be computed for objects ofclass"table" (see?maxstat_test).

• symmetry_test() has gained a new argumentpaired thatis used to indicate if paired data have been transformed in such away that the resulting unstandardized linear statistic is the sum ofthe absolute values of the positive differences between the pairedobservations (see?symmetry_test).

• All transformation functions are now passing through missing values.

• New function for rank transformations (see?rank_trafo).

• New function for maximally selected statistics transformation ofordered factors (see?ofmaxstat_trafo).

• The Conover-Salsburg transformation function now allows theaconstant to be set (see?consal_trafo and?neuropathy).

• The transformation function for ordered factors has gained a newargumentscores that provides an alternative way of attachingscores to ordinal variables (see?of_trafo).

• The approximative (Monte Carlo) null distribution can now beobtained by parallel operation using packageparallel (see?approximate).

• The exact null distribution of test statistics based on quadraticforms can now be obtained in the univariate two-sample case.

• The exact null distribution can now be obtained for tests specifiedwith case weights, implying that exactp-values can becomputed for2 \times 2 tables orK \times 2 tables withK ordered categories.

• Evaluation of the null distribution can now be suppressed for alltest procedures by settingdistribution = "none".

• The mid-p-value can now be computed and is accompanied by a 99% mid-p confidence interval when resampling has been used toobtain the null distribution (see?midpvalue).

• Thep-value interval can now be computed (see?pvalue_interval).

• Step-down adjustedp-values based on the joint distribution ofthe test statistics can now be obtained when the asymptotic nulldistribution is used.

• Adjustedp-values based on the marginal distributions of thetest statistics can now be obtained when the asymptotic nulldistribution is used.

• Adjustedp-values based on the marginal distributions of thetest statistics now use a max-T procedure, instead of amin-P (which is now deprecated), incorporating discretedistributional characteristics when the approximative (Monte Carlo)null distribution is used (see?pvalue).

• Single-step or step-down adjustedp-values based on themarginal distributions of the test statistics can now be obtainedusing the Bonferroni or the Šidák method (see?pvalue).

• Unadjustedp-values can now be obtained (see?pvalue).

• Random numbers can now be generated from the permutationdistribution (see?rperm).

• New data sets on gastric cancer, congenital sex organmalformation and unaided distance vision have been added (see?GTSG,?malformations and?vision).

Bug fixes

• The standardized linear statistic of the Pearson chi-squared testwas off by a factor\sqrt{n / (n - 1)}.

• The Pearson chi-squared test ignored user-specified transformations.

• Partial matching of thescores argument did not work forPearson's chi-squared test and the marginal homogeneity test.

• The Fisher-Pitman permutation test, the Wilcoxon-Mann-Whitney test,the Kruskal-Wallis test, the van der Waerden test, the Brown-Moodmedian test, the Ansari-Bradley test, the Fligner-Killeen test, theWilcoxon signed-rank test, the Friedman test and the Page test didnot return an error message when supplied with a censored responsevariable.

• The Brown-Mood median testp-value computation used the wrongtail of the null distribution in the one-sided case.

• The test statistic of the Wilcoxon signed-rank test had the wrongsign under some circumstances, due to being based on the sum of theabsolute values of the negative differences, instead of thepositive, between the paired measurements.

• The Friedman test did not warn when case weights were supplied.

• Maximally selected statistics did not handle user-specified scoresfor ordinal variables correctly.

• Two-sample problems were not identified as such if the groupingvariable had unused factor levels, causing problems if, e.g., tryingto use thesubset argument offormula methods.

• Stratified tests were not allowed ifblock had unused factorlevels, causing problems, e.g., if trying to use thesubsetargument of theformula methods.

• Transformations involving ordered factors or survival data failedunder some circumstances when performed within blocks.

• The maximally selected statistics transformation for factors droppedits column names under some circumstances.

• The maximally selected statistics transformation for factorsretained unused levels.

• Approximative (Monte Carlo)p-values for stratified tests werewrong when the data were not ordered with respect to the blocks.

• Exactp-values for stratified tests were wrong when each blockcontained precisely two observations.

• Exactp-values for tests with an ordinal grouping variablewere wrong.

• The smallest single-step and step-down adjustedp-valuesdiffered from the globalp-value under some circumstances.

• Single-step adjustedp-values were too small when the supportof the null distribution contained values equal to the observed teststatistics.

• Monotonicity of step-down adjustedp-values was not enforcedwhen using the approximative (Monte Carlo) null distribution.

• The density function of asymptotic permutation distributions formaximum-type tests returned nonsense; it now reportsNA.

• The density function of approximative (Monte Carlo) permutationdistributions returned nonzero probabilities for values not includedin the support.

• The density function of exact permutation distributions returned azero-length vector for values not included in the support; theprobability of such values are now reported as ‘⁠0⁠’.

• The distribution and quantile functions of asymptotic permutationdistributions for maximum-type tests were not handling vectorarguments correctly.

• The quantile function of exact permutation distributions reportedNA when the 100 % quantile was requested under somecircumstances.

• The support of exact permutation distributions for stratified testswas not guaranteed to be distinct and ordered, which in turn causedfailure when trying to obtain values from the density and quantilefunctions.

• The printed output of the Pearson chi-squared test and thegeneralized Cochran-Mantel-Haenszel test claimed that a‘⁠Linear-by-Linear Association Test⁠’ had been carried out evenin the singly ordered case, i.e., when either the response variableor the explanatory variable is an ordered factor; these are nowreported as ‘⁠Generalized Pearson Chi-squared Test⁠’ and‘⁠Generalized Cochran-Mantel-Haenszel Test⁠’, respectively.

• The printed output of the Fisher-Pitman permutation test reportedthe parameter tested in the two-sample case when the explanatoryvariable was an ordered factor.

• The printed output of the one-sided Ansari-Bradley test reported theopposite direction of the specified alternative (but performed thetest in the specified direction and thus reported the correctp-value).

• The printed output of the logrank test did not report whether a two-orK-sample test had been carried out.

• The printed output of the logrank test did not report the testedparameter in the two-sample case.

• The printed output of maximum-type tests did not include thealternative hypothesis.

• The printed output of the general symmetry test claimed that a‘⁠General Independence Test⁠’ had been carried out; it is nowreported as ‘⁠General Symmetry Test⁠’.

• Objects of class"SymmetryProblem" were not checked forvalidity.

User-visible changes

• The Pearson chi-squared test, the generalizedCochran-Mantel-Haenszel test, the linear-by-linear association testand the marginal homogeneity test now use a scalar test statistic inthe doubly ordered case.

• The Fisher-Pitman permutation test, the Kruskal-Wallis test, thelogrank test and the Friedman test now use a scalar test statisticin case the explanatory variable is an ordered factor.

• The Fisher-Pitman permutation test no longer allows the teststatistic to be specified; a quadratic form is now used in theK-sample case.

• The Pearson chi-squared test, the generalizedCochran-Mantel-Haenszel test, the Kruskal-Wallis test, the Friedmantest, the Fligner-Killeen test and the marginal homogeneity test nolonger features thedistribution argument (but still allowthe type of null distribution to be specified).

• The Ansari-Bradley test no longer features thealternativeargument (but still allows the alternative hypothesis to bespecified).

• When the general symmetry test or the marginal homogeneity test isapplied to a table, the variable indicating the measurementconditions has been renamed to ‘⁠conditions⁠’ (from‘⁠groups⁠’).

• Theteststat argument no longer allows the use ofmaximum-type statistics and quadratic forms to be specified using"maxtype" and"quadtype", respectively.

• The median and logrank transformations are now increasing functions.

• The default number of Monte Carlo replicates used to approximate theexact conditional distribution has been increased to 10000.

• The exact distribution of stratified test statistics is now obtainedmore efficiently, and especially so for large numbers of strata.

• exact() hasalgorithm = "auto" as its new default,preserving the behavior ofalgorithm = "shift" in previousversions of the package.

• The shift algorithm (exact(algorithm = "shift")) no longersilently switches to the split-up algorithm when non-integer scoresare detected; it now fails with an error message.

• Single-step adjustedp-values based on the joint distributionof the test statistics are now computed slightly faster due to usingonly the unique realizations of the test statistics.

• The support of an asymptotic null distribution is now reported asNA.

• Small changes in the printed output of the Pearson chi-squared test,the Fisher-Pitman permutation test, the Wilcoxon-Mann-Whitney test,the van der Waerden test, the Brown-Mood median test, the marginalhomogeneity test, the Wilcoxon signed-rank test, and generalizedmaximally selected statistics.

• The printed output of the Fisher-Pitman permutation test now reportsthat a ‘⁠Linear-by-Linear Association Test⁠’ is carried out whenthe explanatory variable is an ordered factor.

• In the printed output of the Ansari-Bradley test and the Spearmantest the tested parameter has been renamed to ‘⁠ratio of scales⁠’and ‘⁠rho⁠’, respectively.

• wilcoxsign_test.IndependenceProblem() has been replaced bywilcoxsign_test.SymmetryProblem().

• The"IndependenceTestStatistic" class is now virtual.

• Objects of class"IndependenceProblem" drop unused factorlevels.

• Objects of classes"ScalarIndependenceTestStatistic" or"QuadTypeIndependenceTestStatistic" now have apairedslot.

• Objects of class"ScalarIndependenceTest" now has aparameter slot.

• Objects of classes"ApproxNullDistribution" or"AsymptNullDistribution" now have a slotseedcontaining the state of the random number generator.

• The package manual has been completely revised.

• The memory consumption during computations has been reduced.

Other changes

• Henric Winell, a long-time collaborator who donated almost allchanges incoin 1.1-0, is now listed as author. Welcome onboard, Henric!

Deprecated and defunct

• surv_test() is now deprecated and has been replaced bylogrank_test().

• method = "discrete" inpvalue() is now deprecated.

• The Wilcoxon signed-rank test no longer features the defunctties.method argument.

Changes in version 1.0-24 (2014-09-17)

• as.integer(round(x)) is safer thanas.integer(x) due totruncation in the latter.

Changes in version 1.0-23 (2013-09-03)

• Fix ‘DESCRIPTION’ and ‘NAMESPACE’ issues.

Changes in version 1.0-22 (2013-04-16)

• Documentation updates.

• Test update (for Sparc Solaris).

Changes in version 1.0-21 (2012-03-08)

• Move ‘inst/doc/vignettes/*’ to ‘vignettes/*’.

• NewCWD data.

Changes in version 1.0-20 (2011-10-04)

• Report a warning for rank tests with case weights.

Changes in version 1.0-19 (2011-08-17)

• Norequire() in.onLoad().

Changes in version 1.0-18 (2010-12-21)

• exact() can not work forsymmetry_test(); spotted by LEPAPE Gilleslepape.gilles@neuf.fr.

Changes in version 1.0-17 (2010-09-13

• Add dependencies required for ‘tests/*’.

Changes in version 1.0-16 (2010-08-23)

• Changeties.method = c("HollanderWolfe", "Pratt") tozero.method = c("Pratt", "Wilcoxon") inwilcoxsign_test() following a suggestion by Fritz Scholz. Awarning is fired to avoid misleading and nonreproducible results.

• Usequantile(..., type = 1) for computing approximate quantiles(suggested by Fritz Scholz).

Changes in version 1.0-14 (2010-08-10

• DocumentationBiobase::expressionSet.

• Further checks and better error messages for nonsense data in (paired)two-sample tests.

Changes in version 1.0-12 (2010-07-03)

• Better error message forwilcoxsign_test(c(1, 1, 1) ~ c(1, 1, 1)).

Changes in version 1.0-11 (2010-04-13)

• Bugfix in discrete MTP adjustment.

Changes in version 1.0-10 (2010-01-27)

• Improve uponqperm() for van de Wiel algorithm again.

Changes in version 1.0-9 (2009-12-30)

• Again precision issues in van de Wiel algorithm:qperm()suffered from too large tolerances of the inverted probabilityfunction.

Changes in version 1.0-8 (2009-10-30)

• Precision issues in van de Wiel algorithm fixed. In rare cases,P(T < t) was returned asp-value instead ofP(T \le t). Two statistics are now considered equal (in all computations)when the difference is smaller thansqrt(.Machine$double.eps).

Changes in version 1.0-7 (2009-09-29)

• dperm(),pperm() andqperm() are vectorized alsofor exact and approximative null distributions.

Changes in version 1.0-6 (2009-09-08)

• Exact distribution for independent two-sample problems with only twoobservations was wrong, spotted by Fritz Scholzfscholz@u.washington.edu.

Changes in version 1.0-5 (2009-06-04)

• Documentation updates.

Changes in version 1.0-4 (2009-05-11)

• Evaluate allformulae inxxx_test()'sparent.frame.

Changes in version 1.0-3 (2009-04-01)

• Add alternative ties handling towilcoxsign_test() featurerequest by Fritz Scholzfscholz@u.washington.edu.

• Vignette update.

Changes in version 1.0-2 (2009-03-16)

• Fix location confidence interval problem spotted by Fritz Scholzfscholz@u.washington.edu.

Changes in version 1.0-1 (2009-01-09)

• Add average scores for logrank test.

• Fix Rd problems.

Changes in version 1.0-0 (2008-11-06)

• JSS paperdoi:10.18637/jss.v028.i08 documents version 1.0-0.

• Set default for<IndependenceTest>@method to‘⁠General Independence Test⁠’.

• covariance() always returns a covariance matrix.

• Addshow() method for"IndependenceTest" objects.

• Functions supplied viadistribution does not need to have aclass.

• Export all classes.

• Compute linear statistics, expectations and (co)variances whenconstructing"IndependenceLinearStatistic" objects instead of"IndependenceTestStatistic" and define methods for this class.

• maxstat_test() with integer case weights gaveNA orwrong cutpoints.

Changes in version 0.6-9 (2008-05-05)

• any() was misused in one place, spotted by Kasper Daniel Hansenkhansen@stat.Berkeley.EDU.

• Code cosmetics by Johannes Huesingjohannes@huesing.name.

Changes in version 0.6-7 (2007-10-09)

• Remove non_function entries from Rd files.

• DisableBiobase example in vignette.

Changes in version 0.6-6 (2007-08-13)

• Fix precision issues with exactp-values (spotted by Michael Faymfay@niaid.nih.gov).

Changes in version 0.6-5 (2007-07-23)

• Add new vignette on technical details.

• Update to newmvtnorm 0.8-0.

• Check for overflow errors in ‘StreitbergRoehmel.c’ (thanks toMichael Faymfay@niaid.nih.gov for spotting this).

Changes in version 0.6-4 (2007-06-28)

• Add new argumentordered_trafo totrafo() and deal withordered factors in a more transparent way (via a new functionof_trafo()).

Changes in version 0.6-3 (2007-06-21)

• New vignette ‘MAXtest’.

Changes in version 0.6-2 (2007-05-13)

• Print name ofx variable and levels whenx is a factor.

• New class"IndependenceLinearStatistic".

Changes in version 0.6-1 (2007-04-25)

• LazyLoad: yes.

Changes in version 0.6-0 (2007-03-01)

• Rename some C source files & update doxygen documentation.

• Simplify some methods and the class structure.

• Clean up code and simplify wrapper functions.

• Documentation updates.

• Deal with factors containing only one level.

• Do not compute design matrices inModelEnvFormula.

• Expand case weights ifdistribution = approximate().

• Assign names to user-specified transformation in case they aremissing.

• Thedistribution argument may now be a function with oneargument allowing for user-specified distributions.

• surv_test() ignored thealternative argument.

Changes in version 0.5-2 (2006-11-03)

• maxstat_trafo did not always choose the correct maximalcutpoint.

• Formula evaluation withoutdata argument was partially broken(thanks to Achim Zeileis for spotting this).

Changes in version 0.5-1 (2006-10-27)

• Improve formula interface forwilcoxsign_test().

Changes in version 0.5-0 (2006-10-17)

• Implement exact distribution for symmetry problems (especiallywilcoxsign_test()).

• Add more checks on StatXact examples.

Changes in version 0.4-14 (2006-09-18)

• Fix problems reported by valgrind.

Changes in version 0.4-13 (2006-09-13)

• ‘coin-Ex.R’ generated byR 2.4.0.

• Enhances: Biobase.

Changes in version 0.4-12 (2006-09-07)

• Add$(FLIBS) to ‘Makevars’.

• Include doxygen documentation for C functions.

• Documentation updates.

Changes in version 0.4-10 (2006-08-25)

• maxstat_test(y ~ x) is now able to deal with unorderedxvariables.

• maxstat_test() returns estimates of both the selected variableand the cutpoint in multivariate situations (as a list).

• More checks formaxstat_test().

• Addcodetools checks to tests.

Changes in version 0.4-9 (2006-07-24)

• Fix typo: Homegeneity.

• Add ASA copyright to ‘LegoCondInf’ vignette.

Changes in version 0.4-8 (2006-06-23)

• Clarify that the Stuart-Maxwell test is computed bymh_test()and add a further example (thanks to Henric Nilssonhenric.nilsson@phadia.com for pointing this out!).

Changes in version 0.4-7 (2006-05-16)

• Improvemaxstat_trafo(): some potential cutpoints could havebeen overlooked in case of ties equal to the maxprob and minprobsample quantiles.

• Reproduce two examples from Hothorn and Lausen (2003) in?maxstat_test (and addedhohnloser data set).

• Add more regression tests formaxstat_test().

Changes in version 0.4-6 (2006-04-24)

• Updated ‘LegoCondInf’ vignette.

• Printed names of test statistics are now ‘⁠chi-squared⁠’,‘⁠maxT⁠’, or ‘⁠Z⁠’ instead of ‘⁠T⁠’.

Changes in version 0.4-5 (2006-04-12)

• User-supplied transformations can return a vector which is coerced tomatrix(..., ncol = 1) internally.

• ytrafo andxtrafo may take functions likerank()directly. However, the use oftrafo() is recommended.

• Argumentteststat may take values"max","quad"or"scalar".

• Internal reimplementation of score handling for ordered factors.

• Handling of multiple ordered factors / ordered factors in multivariatesituations implemented.

• f_trafo() returns a design matrix withNA rows in casemissing values were present.

Changes in version 0.4-4 (2006-02-03)

• MakeR CMD check happy and move ‘src/README’ to‘inst/README’.

Changes in version 0.4-3 (2006-01-12)

• isequal() must not check equality of attributes.

Changes in version 0.4-2 (2005-12-21)

• New data setalpha.

• Data setalzheimer is now a"data.frame", not a"table".

• New vignette with more applications.

Changes in version 0.4-1 (2005-12-02)

• Documentation updates.

• Remove unusedsetAs definitions.

• expectation(),variance() andcovariance() returnnamed vectors or matrices. Names for objects returned bystatistic() have been partially improved.

• All observations withNAs are removed now.

• Some internal improvements (removed unused code, avoid duplicated codechunks etc.).

• More checks onNAs, blocks and multiple ordinal variables (oneunivariate problems are currently allowed to have ordinal variables).

Changes in version 0.3-8 (2005-11-29)

• Exportindependence_test.table().

• Addconf.level attribute for"MCp" objects whenquasi-randomized Monte Carlo procedures (frommvtnorm) have beenused.

• Add newvar_trafo argument totrafo().

• Documentation updates.

Changes in version 0.3-7 (2005-11-23)

• statistic() now returns correct linear and standardizedstatistics when scores are in play.

• alzheimer data added.

• photocar data added.

• Documentation updates.

• support() anddperm() methods have been improved forasymptotic and approximative null distributions.

Changes in version 0.3-6 (2005-11-17)

• Internal C functionR_MonteCarloIndependenceTestreturns apq \times B matrix instead of a list withB elements (the linear statistic for each random permutation ofthe data).

• Various simplifications for the computations of adjustedp-values.

• I() informulae could still cause trouble with class"AsIs", now fixed.

• Logical variables are now allowed informulae (and are treatedlikefactors).

• Fixprint method for marginal homogeneity tests(‘⁠...stratified by block...⁠’).

• R_kronecker in C now available.

• C versions ofnrow() andncol() return 1 orlength for vectors.

• Some internal optimisation.

• C functionsnrow(x) andncol(x) returnLENGTH(x) or1 whenx has nodimattribute.

• NewR interface functionR_kronecker toC_kronecker(which returns avector!).

• Speedup of variance computations in internal functions.

Changes in version 0.3-3 (2005-09-07)

• Fix more problems reported by newcodetools. Try to work aroundtheterms(y ~ ., data = data.frame(1:10)) problem inR 2.2.0.

Changes in version 0.3-2 (2005-08-31)

• Fix some problems reported bycodetools.

Changes in version 0.3-1 (2005-08-20)

• alternative = "less" andalternative = "greater" are nowdefined for"maxtype" statistics as well.

• One- and two-sided single-step and step-down max-Tp-valueadjustments are now available from the appropriatepvalue()method. (NOTE: those procedures have not been tested carefully, yet.)

  In addition, the Bonferroni-adjustment by Westfall & Wolfinger (1997)is available now. Note that the interface topvalue() changedslightly,adjustment = TRUE was replaced bymethod = "single-step".

• More examples added to the vignette.

Changes in version 0.2-14 (2005-07-14)

• distribution = "approximate" for"maxtype" statisticswas wrong in case bothxtrafoandytrafo weremultivariate (the conditional expectation was computed incorrectly).

Changes in version 0.2-13 (2005-07-11)

• I() informulae returns objects of class"AsIs"which caused troubles intrafo().

Changes in version 0.2-12 (2005-06-15)

• Functionality forformula parsing and evaluation is nowimported from packagemodeltools.

• show() returns objects (of class"htest", for example)invisibly, really.

• maxstat_trafo() is much faster now and returns a matrix withboth row- and column names set appropriately.

Changes in version 0.2-11 (2005-06-02)

• Thedistribution argument now takes the return values offunctionsexact(),approximate() orasymptotic()as well. Those functions can be used to specify parameters, such asthe number of Monte Carlo replications via..., distribution = approximate(B = 9999), ....

• show() returns objects (of class"htest", for example)invisibly.

• expectation() returns a vector, not a matrix.

• New genericvariance() for extracting the variance(s) of linearstatistics.

• Only variances (instead of the whole covariance matrix) is computedwhen the distribution of maximum-type test statistics is to beapproximated.

• data may be an object of class"exprSet" (->Biobase), the vignette has an example.

• logrank_trafo() (andsurv_test()) now have aties.method argument, see?surv_test for moreinformation.

Changes in version 0.2-10 (2005-05-04)

• ‘⁠asymptotical⁠’\rightarrow ‘⁠asymptotic⁠’ inprint methods.

• mercuryfish example added.

• <x,y>trafo now can return matrices with number of columnsdifferent from the lhs and rhs offormula.

Changes in version 0.2-9 (2005-03-15)

• mergesort is already defined in‘/usr/include/stdlib.h’:270 on some platforms.

• delay is deprecated inR 2.1.0.