NEWS
afex 1.4
Significant User Visible Changes and New Features
- Added plot_first argument to afex_plot(), which allows passing ggplot2 geom's which will be plotted first, before any other graphical elements. This allows plotting reference lines in the background of the plot, e.g. using geom_hline().
Bugfixes
- Address #87 hopefully increasing speed of emmeans() for mixed objects significantly.
afex 1.3-1 (2024-02-25)
Bugfixes
- prepared afex for ggplot2 version 3.5.0
afex 1.3
Significant User Visible Changes and New Features
- Nicer warning if there is missing data for ANOVA models.
- Removed all_fit() function. mixed(..., all_fit = TRUE) now uses lme4::allFit(). In case you have relied on all_fit() use lme4::allFit().
- Removed support for deprecated ANOVA functions aov.car, aov.ez, and aov4. Instead use aov_car(), aov_ez(), and aov_4().
Bugfixes
- Updates to documentation in various places.
- Using vdiffr for unit tests of afex_plot
- Fixed broken link to www.talkstats.com (replaced with link in web.archive.org)
afex 1.2
Significant User Visible Changes and New Features
- Multiple geoms for data in the background can now be passed to afex_plot() as a list. Likewise, for each geom, a separate list of additional arguments can be passed as a list of lists to data_arg.
- afex_plot() gets a new argument, data_color with default "darkgrey". Like data_alpha for alpha, data_color sets the colour for the geom (or now geoms) used to plot the data in the background.
- Both data_color and data_alpha can now be set to NULL, in which case they will not be set but should be set by the user through data_arg.
Bugfixes
- Data in mixed() output is shown nicer (#113)
- afex_plot() should now work with factor_levels arguments even if corresponding variable is not a factor in the data.
- Resaved plots for new version of ggbeeswarm (#117).
afex 1.1
Significant User Visible Changes and New Features
- Fitting messages for mixed() are now only shown per default in interactive mode, thanks to Marius Barth (#111).
- Nicer error message about missing variables for ANOVAs.
- Removed check_contrasts() argument for ANOVAs, which did not work as described for designs with repeated-measures factors: https://github.com/singmann/afex/issues/109
Bugfixes
- Fixed overzealous data completeness check introduced in version 1.0 in aov_car() which prevented ANOVAs with additive factors and no cross over data from working.
- mixed(..., method = "LRT") and similar nested methods should work with empty factor levels: https://github.com/singmann/afex/issues/96
afex 1.0
Significant User Visible Changes and New Features
- ANOVA objects do not by default contain the aov slot any more
- emmeans() for ANOVA now uses multivariate model as default
- Default p-value method for mixed() is now Satterthwaite: method = "S"
- New data set, Urry et al. (2021, Psych Science): laptop_urry
The following changes are thanks to Mattan S. Ben-Shachar
- Added new vignette: "Testing the Assumptions of ANOVAs"
- Added predict() method for afex_aov objects.
- Deprecated functions test_levene() and test_sphercitiy(), which are now part of the performance package as performance::check_homogeneity() and performance::check_sphericity(), respectively.
Bugfixes
- All vignettes should run without suggested packages present.
- Added ez to suggests, to avoid NOTE on CRAN solaris.