ggcorrmatno longer returns matrices of correlation coefficients or other details. It now returns either a plot or a dataframe and this can dataframe can then be used to create matrices.
output = "proptest" for
ggbarstats functions will now return a dataframe containing results from proportion test.
ggwithinstats will display pairwise comparisons even if
results.subtitle is set to
ggcorrmat supports computing Bayes Factors for Pearson’s r correlation.
ggcorrmat legend, in case of missing values, shows mode - instead of median - for the distribution of sample pairs.
caption.default argument in
ggcorrmat is retired.
ggcorrmatnow internally relies on
correlationfor correlation analyses.
This is a hotfix release to correct some of the failing tests and other minor breakages resulting from the new release of
sample.size.labelargument since this information is included in the goodness of fit test results itself. So setting
FALSEwill suppress this information.
To give users more flexibility in terms of modifying the aesthetic defaults for all
geoms included in the
ggstatsplot plots (each plot typically has multiple geoms), the package now uses a new form of syntax. Previously, each
geom had a separate argument to specify each aesthetic (e.g.,
geom_point would get arguments like
point.color, etc.), which resulted in functions with a massive number of arguments and was unsustainable in the long run. Instead,
ggstatsplot functions now expect a list of such arguments for the respective geom (e.g.,
geom_point will have
point.args argument where a list of arguments
list(size = 5, color = "darkgreen", alpha = 0.8) can be supplied).
grouped_ functions have been refactored to reduce the number of arguments. These functions now internally use the new
combine_plots2 instead of
combine_plots. The additional arguments to primary functions can be provided through
.... These changes will not necessarily break the existing code but will lead to some minor graphical changes (e.g., if you were providing
labels argument explicitly, it will be ignored).
All functions lose the
return argument, which was supposed to be alternative to enter
output. But this was just leading to more confusion on the user’s part. The biggest user-visible impact this is going to have is that
ggcorrmat will no longer be backward-compatible. The older scripts will still work but if the
return argument was anything except
"plot", it will just be ignored.
ggcorrmat no longer has
corr.method argument. To be consistent with rest of the functions in this package, the type of statistics should be specified using
type argument. Additional, it gains a new argument
ggcorrplot.args, which can be used to pass additional arguments to the underlying plotting function (
ggdotplotstats now use the following arguments to modify
geoms corresponding to the lines and labels:
centrality.label.args. This helps avoid specifying millions of arguments.
Removes the vestigial
ggbarstats remove the following vestigial arguments:
facet.proptest arguments were difficult to remember and confusing and are replaced by a common
proportion.test argument. Additionally, the following arguments have all been removed and replaced by
data.label. These plethora of options was a headache to remember.
gghistostats loses the following arguments:
high.color. It made no sense to add a color gradient to this plot when the Y-axis already displayed the information about what the bar represented.
ggscatterstats loses the following arguments:
package. Since this function requires only two colors, it didn’t make much sense to use color palettes to specify this. They can be instead specified using
yfill. You can always use
paletteer::paletteer_d to get a vector of color values and then provide values of your choosing to
Removes sorting options in
ggwithinstats functions. This is something the users can easily do before entering the data in these functions.
ggcorrmat was never supposed to work with Kendall’s correlation coefficient but it accidentally did. This is no longer the case.
ggstatsplot now has a logo, thanks to Sarah! :)
theme_ggstatsplot changes slightly. The biggest change is that the title and the subtitle for plots are now aligned to the left of the plot. This change also forced the legend for
ggpiestats to be displayed on the right side of the plot rather than at the bottom.
More models supported in
Following functions are now re-exported from
normality_message. A few other internal data wrangling functions now reside in
To have a more manageable length of function arguments, additional aesthetic specifications for any given geom can be provided via a dedicated
*.args argument. For example, all aesthetic arguments for
geom_vline can be provided via
ggstatsplot continues with its conscious uncoupling that started in
0.1.0 release: The following functions have now been moved to
bf_meta_message and follow a more logical nomenclature. For the same reason,
lm_effsize_ci function is also no longer exported and lives in the
The summary caption no longer displays log-likelihood value because it tends to be not available for a number of regression model objects and so the caption was unnecessarily being skipped.
Supports robust and Bayes Factors for random-effects meta-analysis.
New dataset included:
More models supported in
Removed vestigial arguments from
continuity, etc.) and
ggwithinstatsno longer produce error with variables with pattern
pairwise_phas been reintroduced as a number of users found it useful to call the function from
ggstatsplotitself rather than using
[ instead of
( to display confidence intervals. Additionally,
denoted sample mean, but was confused with population mean by some users. So these functions instead display
More models supported in
Adapts to the new syntax provided in
ggcoefstatsfunction have been removed. The users can instead provide all such arguments in a list to
ggwithinstats mean labels respect
k argument (#331).
More models supported in
only.significant argument to only display display stats labels for significant effects. This can be helpful when a large number of regression coefficients are to be displayed in a single plot.
ggwithinstatspairwise comparisons were adjusted or not for multiple comparisons.
ggstatsplot is undergoing conscious uncoupling whereby all the statistical processing functions that make stats subtitles are being moved to a new package called
statsExpressions. This new package will act as a backend that handles all things statistical processing. This will not affect the end users of
ggstatsplot unless you have been using the helper functions.
Additionally, multiple pairwise comparison tests are being moved to an independent package called
This uncoupling is designed to achieve two things:
Make the code base of more manageable size in
ggstatsplot, which will make package development a bit easier.
Make the workflow more customizable since now you can prepare your own plots and then use
statsExpressions to display results in the plot rather than relying on
ggstatsplot default plots which are heavily opinionated and not appealing to everyone.
All helper functions
bf_* have been moved to the new
To be consistent with all the other
bf_contingency_tab now use the arguments
y instead of
Major refactoring to reduce the codesize and to rely fully on
There was confusion about what did the red point in
ggbetweenstats plots represents. Now the label also contains μ to highlight that what is being displayed is a mean value.
To be consistent with the rest of the functions,
ggbarstats now uses the following aliases for arguments:
condition. This change is backward-compatible and should not pose any problems for scripts that used
condition arguments in these functions.
Most subtitle expressions now report details about the design. In case of between-subjects design, this will be n_obs, while in case of repeated measures design, this will be n_pairs.
pairwise.annotation now defaults to
"p.value" rather than
ggwithinstats (and their
grouped_ variants) functions. This was done because the asterisk conventions are not consistent across various scientific disciplines.
New dataset included:
bugs_long, for repeated measures designs with
NAs present in the data.
ggstatsplot now uses
rcompanion to compute Spearman’s rho and Kendall’s W. Therefore,
DescTools is removed from dependencies.
ggcoefstats supports following objects:
ggcoefstats now respects
conf.int. It internally always defaulted to
conf.int = TRUE in
broom::tidy irrespective of what was specified by the user.
It was painfully confusing for a lot of users what exactly the asterisks in each facet of
ggpiestats signified. So instead now
ggpiestats displays more detailed results from a goodness of fit (gof) test. No such change is made for
ggbarstats because there is no space to include more details above the bar.
conf.type arguments for
p.kr argument removed because
ggcoefstats will begin to rely on
parameters instead of
sjstats package to compute p-values for some regression models.
ggwithinstatscaption, displayed by default, was incorrect. This has been fixed. This stemmed from a line of code which should have been
paired = TRUE, but was instead
paired = FALSE.
ggcoefstats defaults to
bf.message = TRUE to be consistent with the rest of the functions in the package.
ggcoefstats supports the following class of objects:
bf_ttest is introduced as a general function. The previously exported
bf_two_sample_ttest become its aliases.
bf_meta_message syntax changes to adapt to updates made to
metaBMA package (thanks to #259).
The vestigial arguments
ggcorrmat have been removed. The margins can be adjusted using
gghistostats no longer allows
data argument to be
NULL. This is to make this function’s syntax consistent with rest of the functions in this package (none of which allow
data to be
NULL). This also removes confusion that arose for some users when
data couldn’t be
NULL for its
grouped_ cousin (
outlier_df function is no longer exported since it was always meant to be an internal function and was accidently exported during initial release and was retained for a while for backward compatibility.
Instead of having two separate functions that dealt with repeated measures (
subtitle_friedman_nonparametric) and between-subjects (
subtitle_kw_nonparametric), a single function
subtitle_anova_nonparametric handles both of these designs with the
paired argument determining which test is run.
All functions that supported Bayes Factor analysis (
type = "bf") will only return BF value and the scale used. Previously, this was a mix of parametric statistics and BF, which was confusing and often times misleading since these two types of analyses relied on different tests.
The default for
bf.message has been changed from
TRUE. This is to make the Bayes Factor analysis more visible to the user.
ggscatterstatsreturns only plot (without any statistical details) when the specified model is not linear (i.e., either when
methodargument is not
y ~ x).
ggwithinstats (and its
grouped_ variant) are introduced as a counterpart to
ggbetweenstats to handle repeated measures designs.
For repeated measures ANOVA,
subtitle_anova_nonparametric now returns confidence intervals for Kendall’s W.
All functions get
return argument that can be used to return either
"caption". This makes it unnecessary to remember which subtitle function is to be used where. As a result, in the next release, all subtitle making functions will not be exported and are encouraged not be used either by other developers or by users.
subtitle_anova_parametric gain a new argument
paired to support repeated measures designs.
ggcoefstats can support following new model objects:
bf.message argument to display a caption containing results from Bayesian random-effects meta-analysis. It therefore gains a new dependency:
ggcatstats will now display Cramer’s V as effect size for one-sample proportion tests.
All functions gain
stat.title argument (
NULL by default) that can be used to prefix the subtitle with a string of interest. This is possibly useful for specifying the details of the statistical test.
pairwise_p() function no longer outputs
conf.high columns when parametric post hoc tests are run. This is because these values were accurate only when no p-value adjustment was carried out.
Instead of using the internal function
ggscatterstats instead used
SpearmanRho function from
DescTools package. This was done to reduce number of custom internal functions used to compute CIs for various effect sizes.
ggstatsplot therefore gains
DescTools as a dependency.
sampling.plan argument default for
ggbarstats function has been changed from
"jointMulti" to be consistent with its sister function
ggcoefstats can support following new model objects:
VR_dilemma dataset for toying around with within-subjects design.
subtitle_t_onesample supports both Cohen’s d and Hedge’s g as effect sizes and also produces their confidence intervals. Additionally, non-central variants of these effect sizes are also supported. Thus,
gghistostats and its
grouped_ variant gets two new arguments:
ggpiestats used to display odds ratio as effect size for paired designs (McNemar test). But this was only working when the analysis was a 2 x 2 contingency table. It now instead displays Cohen’s G as effect size, which generalizes to any kind of design.
The internal function
outlier_df to add a column specifying outlier status of any given data point is now exported.
ggstatsplot previously relied on an internal function
chisq_v_ci to compute confidence intervals for Cramer’s V using bootstrapping but it was pretty slow. It now instead relies on
rcompanion package to compute confidence intervals for V.
ggstatsplot, therefore, gains a new dependency.
subtitle_t_onesample now computes effect size r and its confidence intervals as
(with the help of
rcompanion package), instead of using Spearman correlation.
subtitle_t_onesampleno longer has
dataas the optional argument. This was done to be consistent with other subtitle helper functions.
ggbarstats (and its
grouped_ variant) introduced for making bar charts (thanks to #78).
ggcoefstats also displays a caption with model summary when meta-analysis is required.
gghistostats and its
grouped_ variant has a new argument
normal.curve to superpose a normal distribution curve on top of the histogram (#138).
ggcoefstats can support following new regression model objects:
New function to convert plots which are not of
ggplot class to
ggplot class objects.
Instead of using
effsize to compute Cohen’s d and Hedge’s g,
ggstatsplot now relies on a new (#159) internal function
effect_t_parametric to compute them. This removes
effsize from dependencies.
To be consistent with other functions in this package, both
results.subtitle which can be set to
FALSE if statistical analysis is not required, in which case
subtitle argument can be used to provide alternative subtitle.
ggbetweenstatsnow defaults to using noncentral-t distribution for computing Cohen’s d and Hedge’s g. To get variants with central-t distribution, use
effsize.noncentral = FALSE.
grouped_ functions had argument
title.prefix that defaulted to
"Group". It now instead defaults to
NULL, in which case the prefix will variable name for
To accommodate non-parametric tests,
subtitle_template function can now work with
parameter = NULL.
ggbetweenstats, details contained in the subtitle for non-parametric test are modified. It now uses Spearman’s rho-based effect size estimates. This removes
coin from dependencies.
ggbetweenstats and its
grouped_ variant gain a new argument
axes.range.restrict (which defaults to
FALSE). This restricts
y-axes limits to minimum and maximum of
y variable. This is what these functions were doing by default in the past versions, which created issues for additional ggplot components using the
All bayes factor related subtitle and captions replace
ggcoefstats passes dots (
augment method from
The helper function
bf_extractor no longer provides option to extract information about posterior distribution because these details were incorrect. The
posterior = TRUE details were not used anywhere in the package so nothing about the results changes.
ggcorrmat didn’t output pair names when
output == "ci" was used. This is fixed.
meta.analytic.effect that can be used to carry out meta-analysis on regression estimates. This especially useful when a dataframe with regression estimates and standard error is available from prior analyses. The
subtitle is prepared with the new function
subtitle_meta_ggcoefstats which is also exported.
ggdotplotstats (and their
grouped_ variants) all gain a new
ggplot.component argument. This argument will primarily be helpful to change the individual plots in a
ggcoefstats can support following new regression model objects:
ggcoefstats didn’t work when
statistic argument was set to
NULL. This was not expected behavior. This has been fixed. Now, if
statistic is not specified, only the dot-and-whiskers will be shown without any labels.
subtitle_t_parametric was producing incorrect sample size information when
paired = TRUE and the data contained
NAs. This has been fixed.
ggscatterstats and its
grouped_ variant accept both character and bare exressions as input to arguments
To be consistent with rest of the functions in the package, both Pearson’s r, Spearman’s rho, and robust percentage bend correlations also display information about statistic associated with these tests.
ggscatterstats, by default, showed jittered data points (because it relied on
position_jitter defaults). This could be visually inaccurate and, therefore,
ggscatterstats now displays points without any jitter. The user can introduce jitter if they wish to using
point.height.jitter arguments. For similar reasons, for
ggbetweenstats and its
point.jitter.height default has been changed from
0 (no vertical jitter, i.e.).
Confidence interval for Kendall’s W is now computed using
stats::kruskal.test. As a result,
PMCMRplus removed from dependencies.
ggcoefstats gains a
caption argument. If
caption.summary is set to
TRUE, the specified caption will be added on top of the
ggcoefstats was showing wrong confidence intervals for
merMod class objects due to a bug in the
broom.mixed package (https://github.com/bbolker/broom.mixed/issues/30#issuecomment-428385005). This was fixed in
broom.mixed and so
ggcoefstats should no longer have any issues.
specify_decimal_p has been modified because it produced incorrect results when
k < 3 and
p.value = TRUE (e.g.,
0.002 was printed as
ggpiestats produced incorrect results if some levels of the factor had been filtered out prior to using this function. It now drops unused levels and produces correct results.
gghistostats wasn’t filtering out
NAs properly. This has been fixed.
ggdotplotstats for creating a dot plot/chart for labelled numeric data.
All primary functions gain
conf.level argument to control confidence level for effect size measures.
As per APA guidelines, all results show results with two decimal places. That is, the default value for
k argument for all functions has been changed from
All helper functions for the
ggbetweenstats subtitles have been renamed to remove
_ggbetween_ from their names as this was becoming confusing for the user. Some of these functions work both with the between- and within-subjects designs, so having
_ggbetween_ in their names made users suspect if they could use these functions for within-subjects designs.
ggstatsplot now depends on
R 3.5.0. This is because some of its dependencies require 3.5.0 to work (e.g.,
ggbetweenstats now supports multiple pairwise comparison tests (parametric, nonparametric, and robust variants). It gains a new dependency
ggbetweenstats now supports eta-squared and omega-squared effect sizes for anova models. This function gains a new argument
Following functions are now reexported from the
groupedstats package to avoid repeating the same code in two packages:
groupedstats is now added as a dependency.
gghistostats can now show both counts and proportions information on the same plot when
bar.measure argument is set to
ggcoefstats works with tidy dataframes.
The helper function
untable has been deprecated in light of
tidyr::uncount, which does exactly what
untable was doing. The author wasn’t aware of this function when
untable was written.
All vignettes have been removed from
CRAN to reduce the size of the package. They are now available on the package website: https://indrajeetpatil.github.io/ggstatsplot/articles/.
subtitle_t_robust function can now handle dependent samples and gains
ggpiestats, and their
grouped_ variant support bayes factor tests and gain new arguments relevant to this test.
Effect size and their confidence intervals now available for Kruskal-Wallis test.
Minor stylistic changes to how symbols for partial-eta-/omega-squared were being displayed in subtitles.
ggbetweenstats supports bayes factor tests for anova designs.
ggpiestats (and its
grouped_ version) gain
slice.label argument that decides what information needs to be displayed as a label on the slices of the pie chart:
"percentage" (which has been the default thus far),
ggcorrmat can work with
cor.vars = NULL. In such case, all numeric variables from the provided dataframe will be used for computing the correlation matrix.
Given the constant changes to the default behavior of functions, the lifecycle badge has been changed from
When the number of colors needed by a function exceeds the number of colors contained in a given palette, informative message is displayed to the user (with the new internal function
Several users had requested an easier way to turn off subtitles with results from tests (which was already implemented in
gghistostats with the argument
ggbetweenstats also gains two new arguments to do this:
New dataset added:
More tests added and the code coverage has now jumped to over 75%.
To avoid code repetition, there is a now a function that produces a generic message any time confidence intervals for effect size estimate are computed using bootstrapping.
The package now exports all functions used to create text expressions with results. This makes it easy for people to use these results in their own plots at any location they want (and not just in
subtitle, the current default for
p.adjust.method argument which allows p-values for correlations to be corrected for multiple comparisons.
label.expression arguments to attach labels to points.
gghistostats now defaults to not showing (redundant) color gradient (
fill.gradient = FALSE) and shows both
"proportion" data. It also gains a new argument
bar.fill that can be used to fill bars with a uniform color.
ggpiestats now support all palettes contained in the
paletteer package. This helps avoid situations where people had large number of groups (> 12) and there were not enough colors in any of the
bf.message argument to display bayes factors in favor of the null (currently works only for parametric t-test).
gghistostats function no longer has
line.labeller.y argument; this position is automatically determined now.
legend.title.margin function has been deprecated since
ggplot2 3.0.0 has improved on the margin issues from previous versions. All functions that wrapped around this function now lose the relevant arguments (
ggstatsplot.theme has been changed to
ggcorrmat function to be consistent across functions.
conf.type arguments for
ggbetweenstats have been deprecated. No other function in the package allowed changing confidence interval or their type for effect size estimation. These arguments were relevant only for
robust tests anyway.
type has been changed to
matrix.type because for all other functions
type argument specifies the type of the test, while for this function it specified the display of the visualization matrix. This will make the syntax more consistent across functions.
ggscatterstats gains new arguments to specify aesthetics for geom point (
point.alpha). To be consistent with this naming schema, the
height.jitter arguments have been renamed to
gghistostats: To be compatible with
JASP, natural logarithm of Bayes Factors is displayed, and not base 10 logarithm.
formula arguments to modify smoothing functions.
ggcorrmat can now show
robust correlation coefficients in the matrix plot.
binwidth value, if not specified, is computed with
(max-min)/sqrt(n). This is basically to get rid of the warnings ggplot2 produces. Thanks to Chuck Powell’s PR (#43).
ggcoefstats gains a new argument
partial and can display eta-squared and omega-squared effect sizes for anovas, in addition to the prior partial variants of these effect sizes.
perc.k argument to show desired number of decimal places in percentage labels.
grouped_ggpiestatswasn’t working when only
mainvariable was provided with
countsdata. Fixed that.
For the sake of consistency,
theme_mprl is now called
theme_mprl function will still be around and will not be deprecated, so feel free to use either or both of them since they are identical.
ggcoefstats no longer has arguments
ran_params because only fixed effects are shown for mixed-effects models.
ggpiestats can now handle within-subjects designs (McNemar test results will be displayed).
ggbetweenstats was producing wrong axes labels when
sample.size.label was set to
TRUE and user had reordered factor levels before using this function. The new version fixes this.
ggcoefstats wasn’t producing partial omega-squared for
aovlist objects. Fixed that with new version of
Removed the trailing comma from the robust correlation analyses.
gghistostats has a new argument to remove color fill gradient.
ggbetweenstats takes new argument
mean.ci to show confidence intervals for the mean values.
lmer models, p-values are now computed using
sjstats::p_value. This removes
lmerTest package from dependencies.
sjstats no longer suggests
apaTables package to compute confidence intervals for partial eta- and omega-squared. Therefore,
MBESS are removed from dependencies.
densigram with the development version of
ggExtra. It additionally gains few extra arguments to change aesthetics of marginals (alpha, size, etc.).
ggcoefstats for displaying model coefficients.
All functions now have
ggtheme argument that can be used to change the default theme, which has now been changed from
The robust correlation is no longer
MASS::rlm, but percentage bend correlation, as implemented in
WRS2::pbcor. This was done to be consistent across different functions.
ggcorrmat also uses percentage bend correlation as the robust correlation measure. This also means that
ggstatsplot no longer imports
data argument is no longer
NULL for all functions, except
gghistostats. In other words, the user must provide a dataframe from which variables or formulas should be selected.
All subtitles containing results now also show sample size information (n). To adjust for the inflated length of the subtitle, the default subtitle text size has been changed from
Switched back to Shapiro-Wilk test of normality to remove
nortest from imports.
ggpiestats now display sample sizes for each level of the groping factor by default. This behavior can be turned off by setting
Three new datasets added:
Added confidence interval for effect size for robust ANOVA.
The 95% CI for Cramer’V computed using
boot::boot. Therefore, the package no longer imports
To be consistent across correlations covered, all correlations now show estimates for correlation coefficients, confidence intervals for the estimate, and p-values. Therefore, t-values and regression coefficients are no longer displayed for Pearson’s r.
legend.title.margin arguments for
ggcorrmat now default to
ggplot2 3.0.0 has better legend title margins.
ggpiestats now sorts the summary dataframes not by percentages but by the levels of
main variable. This was done to have the same legends across different levels of a grouping variable in
To remove cluttered display of results in the subtitle,
ggpiestats no longer shows titles for the tests run (these were “Proportion test” and “Chi-Square test”). From the pie charts, it should be obvious to the user or reader what test was run.
gghistostats also allows running robust version of one-sample test now (One-sample percentile bootstrap).
ggbetweenstats function can now show notched box plots. Two new arguments
notchwidth control its behavior. The defaults are still standard box plots.
Removed warnings that were appearing when
outlier.label argument was of
The default color palette used for all plots is colorblind friendly.
density as a value measure for bar heights to show proportions and density. New argument
bar.measure controls this behavior.
grouped_ variants of functions
ggpiestats introduced to create multiple plots for different levels of a grouping variable.
To be internally consistent, all functions in
ggstatsplot use the spelling
color, rather than
colour in some functions, while
color in others.
Removed the redundant argument
gghistostats function. This argument was relevant for the first avatar of this function, but is no longer playing any role.
To be internally consistent, the argument
ggcorrmat have been changed to
Added a new argument to
ggstatsplot.theme function to control if
ggstatsplot::theme_mprl is to be overlaid on top of the selected
ggtheme (ggplot2 theme, i.e.).
Two new arguments added to
gghistostats to allow user to change colorbar gradient. Defaults are colorblind friendly.
ggcorrmat have a new argument
legend.title.margin to control margin adjustment between the title and the colorbar.
The vertical lines denoting test values and centrality parameters can be tagged with text labels with a new argument
ggscatterstatswas not working properly. Choosing
"median"didn’t show median, but the mean. This is fixed now.
Bayesian test added to
gghistostats and two new arguments to also display a vertical line for
Vignette added for
Added new function
grouped_gghistostats to facilitate applying
gghistostats for multiple levels of a grouping factor.
ggbetweenstats has a new argument
outlier.coef to adjust threshold used to detect outliers. Removed bug from the same function when
outlier.label argument is of factor/character type.
grouped_proptest are now deprecated. They were exported in the first release by mistake.
gghistostats no longer displays both density and count since the density information was redundant. The
density.plot argument has also been deprecated.
intercept has now been changed to
centrality.para. This was due to possible confusion about interpretation of these lines; they show central tendency measures and not intercept for the linear model. Thus the change.
The default for
effsize.type = "biased" effect size for
ggbetweenstats in case of ANOVA is partial omega-squared, and not omega-squared. Additionally, both partial eta- and omega-squared are not computed using bootstrapping with (default) 100 bootstrap samples.
More examples added to the
95% confidence intervals for Spearman’s rho are now computed using
RVAideMemoire package is thus removed from dependencies.
95% confidence intervals for partial eta- and omega-squared for
ggbetweenstats function are now computed using
sjstats package, which allows bootstrapping.
userfriendlyscience packages are thus removed from dependencies.