NEWS.md
Due to a bug in previous versions of WRS2::lincon, even if p-value correction was method = "none", it still applied Hochberg’s correction and the pairwiseComparisons package once again applied Holm’s correction, which means the p-values were over-corrected for multiple comparisons. This bug has been fixed in WRS2 1.1-3 and therefore the users should expect slightly different p-values for between-subjects post hoc robust tests.
pairwiseComparisons now relies on statsExpressions for statistical analysis.
All included datasets have now been removed since the same datasets are also present in statsExpressions package.
No longer depends on ipmisc package.
To avoid confusion among users, the trimming level for all functions is now changed from tr = 0.1 to tr = 0.2 (which is what WRS2 defaults to).
The ... are now passed to other methods. This can be used to specify additional arguments, like alternative (#28).
Gets rid of iris_long dataset, which was not used in the package.
Minor internal refactoring.
Removes the unnecessary (and confusing) significance column from all outputs.
To be consistent with the rest of the ggstatsverse, the Bayes Factor results are now always shown in favor of null over alternative (BF01).
pairwise_comparisons function gets subject.id argument relevant for repeated measures design.
The label column returned in pairwise_comparisons now displays the p-value adjustment method in the label itself.
pairwise_caption function has changed its output to reflect changes made to the p-value labels.
Major internal refactoring to get rid of the following dependencies: broomExtra, dunn.test, forcats, and tidyr. This comes at the cost of omission of few of the details that were previously included in the output (e.g., mean.difference column for Student’s t-test).
tidyBF 0.3.0.Fixes a bug which affected results for within-subjects design when the dataframe wasn’t sorted by x (#19).
This fix also now makes the results more consistent, such that irrespective of which type of statistics is chosen the group1 and group2 columns are in identical order.
R (3.6).For repeated measures datasets with NAs present, the Bayes Factor values were incorrect. This is fixed.
Internal refactoring to improve data wrangling using ipmisc.
Removes dependence on jmv and instead relies on dunn.test and PMCMRplus. This significantly reduces number of dependencies.
The non-parametric Dwass test has been changed to Dunn test.
Adapts to breaking changes in upcoming release of broom 0.7.0.
Thanks to Sarah, the package has a hexsticker. :)
Due to changes made to downstream dependencies, the minimum R version expected is bumped to 3.6.0.
Adds support for the Bayes Factor tests.
Exports the internal helper function pairwise_caption.
pairwise_comparisons_caption is removed since it was helpful only for ggstatsplot’s internal graphics display and wasn’t of much utility outside of that context.Instead of cluttering the terminal with messages, pairwise_comparisons function now instead adds two columns (test.details and p.value.adjustment) to all outputs specifying which test was carried out and which adjustment method is being used for p-value correction.
Gets rid of groupedstats and crayon from dependencies.
jmv 1.0.8, the results from the Dwass-Steel-Crichtlow-Fligner test will be slightly different.