Skip to contents

anovakun 1.1.0

  • Updated to ‘ANOVA-kun’ 4.9.0 (released 2026-11-02). Changes in the original: the prep option (p_rep) is removed; the output of iga/ciga for the simple main effects of between-subjects factors is fixed; anovatan() with tech = TRUE returns the results; the bootstrap (besci = TRUE) stops when a cell has fewer than three observations.
  • Where the original stops the bootstrap with message() and invokeRestart("abort"), the package signals an error with stop() and the same message, so that the condition can be caught.

anovakun 1.0.0

Results as objects

  • anovakun() returns an object of class "anovakun" instead of writing the tables to the console. The tables are written by its print() method in the same layout as the original, so calling anovakun() at the top level shows the same output as before, while res <- anovakun(...) shows nothing and keeps the results. The object holds the same list as tech = TRUE. tech = TRUE still returns the plain list.
  • anovatan() returns an object of class "anovatan" holding the "anovakun" results for each level (and the split data with data.frame = TRUE). With tech = TRUE it returns a list of the heading and the plain list of each level in turn, as the original does from 4.9.0 (up to 4.8.9 the original printed only the headings and returned NULL).
  • With data.frame = TRUE, the data frame is an element (DATA.FRAME) of the returned object; the original returned it alone.
  • With bgraph, the bar graph is drawn when anovakun() is called, i.e. before the tables are printed.
  • copy = TRUE writes the printed tables to the clipboard instead of redirecting the console with sink().
  • The progress of the bootstrap (besci = TRUE) is reported with message(); the progress bar is shown only in interactive sessions.

Changes from the original source (no effect on the results)

  • ciml = TRUE calls ‘lmerTest’ through its namespace instead of library(lmerTest); ‘lmerTest’ is a suggested package.
  • ci.bars() draws on the current graphics device instead of opening a new quartz() or x11() window.
  • options() and par() are restored with on.exit(), so they are restored even if the analysis stops with an error.
  • The operating system (for copy = TRUE, read.clip()) is detected with Sys.info(); the original test on .Platform did not recognize macOS when R was built from source.
  • Internal functions anova.modeler() and anova.output() were renamed to anova_modeler() and anova_output() so that they are not mistaken for S3 methods of anova(). The unused internal function hmean() was removed.