- The requirement that variables need at least 2 nonmissing
observations in each group for a significance test to be performed now
only applies to numeric variables
- Reintroduce stretch space to LaTeX tables.
- Added var_equal option to test_options which will lead to use of
Student’s t-test instead of Welch’s.
- Added functionality to automatically apply both “usual” summmary
statistics aswell as categorical descriptions to ordered variables which
can be converted to numeric without warning.
- Added max_first_col_width format option for LaTeX tables.
- Some random bugfixes i seem to have forgotten to push.
- Fix bug where confidence intervals for differences in rates where
displayed as differences in row-proportions instead of
- Make error message for improperly matched data more precise
- Fix a bug where ordered factors were not correctly converted to rank
data for non-parametric testing purposes
- Add check in .meanCIlower and .meanCIupper for non-constantness of
- Fixed a bug where indices could not be specified inside
- fixed bug with printing CI abbreviations in tex mode
- Add hline between footnotes and table in tex mode
- make pagebreak alogrithm less aggressive
- Fix bug with group_labels that lead to group N numbers not being
- Fix major bug where percentages were not calculated correctly if
factor levels are omitted.
- Fix bug with confidence interval functions
- Added lapply_descr function which allows applying descr to a list of
- Major code refactoring, removing alot of code duplication
- The confidence interval column now behaves more similarily to the
p-value colum and now supports footnotes.
- Fixed github actions and updated documentation
- Added Fisher’s exact test for KxL tables
- Added exact binomial test for 1xL tables
- Fixed some issue in the usage documentation where a code chuck was
- Some work on github actions
- knit_print now works properly. R chunks in .Rmd documents don’t need
the results = ‘asis’ option anymore.
- Tables can now have captions
- LaTeX tables will now properly wrap the names of long factor levels
and long variable names.
- .mean now returns NA_real_ in cases were NaN was previously
- Fixed a bug with paired data, where some of the pairs contain
missing. In some cases, tests were not calculated and a misleading error
message was produced.
- Added the ability to suppress the “Total” column via the
format_options(print_Total = FALSE) option.
- Exact McNemars test will now calculate confidence intervals for rate
differences by leveraging the distribution of the test statistic.
- Using (non-exact) McNemars test will now produce a warning, that the
confidence intervals for the differences in rates do not consider the
paired structure of the data.
- Datasets containing variables which inherit from the “Date” class
are now automatically converted to factors.
- Fixed a bug with print_CI = FALSE option in format_options.
- “&” signs in factor labels are now properly escaped in LaTeX
- Added the “combine_median_Q1_Q3” argument to format_options, which
reshapes these summary statistics to “median (Q1, Q3)”.
- Added the following members to format_options: percent_suffix,
row_percent, Nmiss_row_percent and absolute_relative_frequency_mode.
They can be used to control control how absolute and relative
frequencies are displayed.
- Added documentation for confidence intervals.
- All tests that previously used continuity correction by default do
not use continuity correction anymore.
- Added the
additional_test_args argument to
test_options. This lets the user pass arguments to the
underlying test functions,
additional_test_args = list(correct=TRUE) to request
continuity correction in