⚠️There's a newer version (0.12.2) of this package. Take me there.

ggstatsplot: ggplot2 Based Plots with Statistical Details

PackageStatusUsageGitHubReferences

Raison d’être

“What is to be sought in designs for the display of information is the clear portrayal of complexity. Not the complication of the simple; rather … the revelation of the complex.”
- Edward R. Tufte

ggstatsplot is an extension of ggplot2 package for creating graphics with details from statistical tests included in the plots themselves and targeted primarily at behavioral sciences community to provide a one-line code to produce information-rich plots. In a typical exploratory data analysis workflow, data visualization and statistical modeling are two different phases: visualization informs modeling, and modeling in its turn can suggest a different visualization method, and so on and so forth. The central idea of ggstatsplot is simple: combine these two phases into one in the form of graphics with statistical details, which makes data exploration simpler and faster.

Summary of available plots

It, therefore, produces a limited kinds of plots for the supported analyses:

FunctionPlotDescription
ggbetweenstatsviolin plotsfor comparisons between groups/conditions
ggwithinstatsviolin plotsfor comparisons within groups/conditions
gghistostatshistogramsfor distribution about numeric variable
ggdotplotstatsdot plots/chartsfor distribution about labeled numeric variable
ggpiestatspie chartsfor categorical data
ggbarstatsbar chartsfor categorical data
ggscatterstatsscatterplotsfor correlations between two variables
ggcorrmatcorrelation matricesfor correlations between multiple variables
ggcoefstatsdot-and-whisker plotsfor regression models

In addition to these basic plots, ggstatsplot also provides grouped_ versions (see below) that makes it easy to repeat the same analysis for any grouping variable.

Summary of types of statistical analyses

Currently, it supports only the most common types of statistical tests: parametric, nonparametric, robust, and bayesian versions of t-test/anova, correlation analyses, contingency table analysis, and regression analyses.

The table below summarizes all the different types of analyses currently supported in this package-

FunctionsDescriptionParametricNon-parametricRobustBayes Factor
ggbetweenstatsBetween group/condition comparisonsYesYesYesYes
ggwithinstatsWithin group/condition comparisonsYesYesYesYes
gghistostats, ggdotplotstatsDistribution of a numeric variableYesYesYesYes
ggcorrmatCorrelation matrixYesYesYesNo
ggscatterstatsCorrelation between two variablesYesYesYesYes
ggpiestats, ggbarstatsAssociation between categorical variablesYesNANAYes
ggpiestats, ggbarstatsEqual proportions for categorical variable levelsYesNANAYes
ggcoefstatsRegression model coefficientsYesNoYesNo

Statistical reporting

For all statistical tests reported in the plots, the default template abides by the APA gold standard for statistical reporting. For example, here are results from Yuen’s test for trimmed means (robust t-test):

Summary of statistical tests and effect sizes

Here is a summary table of all the statistical tests currently supported across various functions:

FunctionsTypeTestEffect size95% CI available?
ggbetweenstats (2 groups)ParametricStudent’s and Welch’s t-testCohen’s d, Hedge’s g
ggbetweenstats (> 2 groups)ParametricFisher’s and Welch’s one-way ANOVA
ggbetweenstats (2 groups)Non-parametricMann-Whitney U-testr
ggbetweenstats (> 2 groups)Non-parametricKruskal-Wallis Rank Sum Test
ggbetweenstats (2 groups)RobustYuen’s test for trimmed means
ggbetweenstats (> 2 groups)RobustHeteroscedastic one-way ANOVA for trimmed means
ggwithinstats (2 groups)ParametricStudent’s t-testCohen’s d, Hedge’s g
ggwithinstats (> 2 groups)ParametricFisher’s one-way repeated measures ANOVA
ggwithinstats (2 groups)Non-parametricWilcoxon signed-rank testr
ggwithinstats (> 2 groups)Non-parametricFriedman rank sum test
ggwithinstats (2 groups)RobustYuen’s test on trimmed means for dependent samples
ggwithinstats (> 2 groups)RobustHeteroscedastic one-way repeated measures ANOVA for trimmed means
ggpiestats and ggbarstats (unpaired)ParametricCramér’s V
ggpiestats and ggbarstats (paired)ParametricMcNemar’s testCohen’s g
ggpiestatsParametricOne-sample proportion testCramér’s V
ggscatterstats and ggcorrmatParametricPearson’s rr
ggscatterstats and ggcorrmatNon-parametric
ggscatterstatsand ggcorrmatRobustPercentage bend correlationr
gghistostats and ggdotplotstatsParametricOne-sample t-testCohen’s d, Hedge’s g
gghistostatsNon-parametricOne-sample Wilcoxon signed rank testr
gghistostats and ggdotplotstatsRobustOne-sample percentile bootstraprobust estimator
ggcoefstatsParametricRegression models

Installation

To get the latest, stable CRAN release (0.1.3):

utils::install.packages(pkgs = "ggstatsplot")

Note: If you are on a linux machine, you will need to have OpenGL libraries installed (specifically, libx11, mesa and Mesa OpenGL Utility library - glu) for the dependency package rgl to work.

You can get the development version of the package from GitHub (0.1.3.9000). To see what new changes (and bug fixes) have been made to the package since the last release on CRAN, you can check the detailed log of changes here: https://indrajeetpatil.github.io/ggstatsplot/news/index.html

If you are in hurry and want to reduce the time of installation, prefer-

# needed package to download from GitHub repo
utils::install.packages(pkgs = "remotes")

# downloading the package from GitHub
remotes::install_github(
  repo = "IndrajeetPatil/ggstatsplot", # package path on GitHub
  dependencies = FALSE, # assumes you have already installed needed packages
  quick = TRUE # skips docs, demos, and vignettes
)

If time is not a constraint-

remotes::install_github(
  repo = "IndrajeetPatil/ggstatsplot", # package path on GitHub
  dependencies = TRUE, # installs packages which ggstatsplot depends on
  upgrade_dependencies = TRUE # updates any out of date dependencies
)

If you are not using the RStudio IDE and you get an error related to “pandoc” you will either need to remove the argument build_vignettes = TRUE (to avoid building the vignettes) or install pandoc. If you have the rmarkdown R package installed then you can check if you have pandoc by running the following in R:

rmarkdown::pandoc_available()
#> [1] TRUE

Citation

If you want to cite this package in a scientific journal or in any other context, run the following code in your R console:

citation("ggstatsplot")

There is currently a publication in preparation corresponding to this package and the citation will be updated once it’s published.

Documentation and Examples

To see the detailed documentation for each function in the stable CRAN version of the package, see:

To see the documentation relevant for the development version of the package, see the dedicated website for ggstatplot, which is updated after every new commit: https://indrajeetpatil.github.io/ggstatsplot/.

Help

In R, documentation for any function can be accessed with the standard help command (e.g., ?ggbetweenstats).

Another handy tool to see arguments to any of the functions is args. For example-

args(name = ggstatsplot::specify_decimal_p)
#> function (x, k = 3, p.value = FALSE) 
#> NULL

In case you want to look at the function body for any of the functions, just type the name of the function without the parentheses:

# function to convert class of any object to `ggplot` class
ggstatsplot::ggplot_converter
#> function(plot) {
#>   cowplot::ggdraw() + cowplot::draw_grob(grid::grobTree(plot))
#> }
#> <bytecode: 0x000000002d4a7830>
#> <environment: namespace:ggstatsplot>

If you are not familiar either with what the namespace :: does or how to use pipe operator %>%, something this package and its documentation relies a lot on, you can check out these links-

Primary functions

Here are examples of the main functions currently supported in ggstatsplot.

Note: If you are reading this on GitHub repository, the documentation below is for the development version of the package. So you may see some features available here that are not currently present in the stable version of this package on CRAN. For documentation relevant for the CRAN version, see: https://CRAN.R-project.org/package=ggstatsplot/readme/README.html

ggbetweenstats

This function creates either a violin plot, a box plot, or a mix of two for between-group or between-condition comparisons with results from statistical tests in the subtitle. The simplest function call looks like this-

# loading needed libraries
library(ggstatsplot)

# for reproducibility
set.seed(123)

# plot
ggstatsplot::ggbetweenstats(
  data = iris,
  x = Species,
  y = Sepal.Length,
  messages = FALSE
) + # further modification outside of ggstatsplot
  ggplot2::coord_cartesian(ylim = c(3, 8)) +
  ggplot2::scale_y_continuous(breaks = seq(3, 8, by = 1))

Note that this function returns object of class ggplot and thus can be further modified using ggplot2 functions.

A number of other arguments can be specified to make this plot even more informative or change some of the default options. Additionally, this time we will use a grouping variable that has only two levels. The function will automatically switch from carrying out an ANOVA analysis to a t-test.

The type (of test) argument also accepts the following abbreviations: "p" (for parametric) or "np" (for nonparametric) or "r" (for robust) or "bf" (for Bayes Factor). Additionally, the type of plot to be displayed can also be modified ("box", "violin", or "boxviolin").

A number of other arguments can be specified to make this plot even more informative or change some of the default options.

library(ggplot2)

# for reproducibility
set.seed(123)

# let's leave out one of the factor levels and see if instead of anova, a t-test will be run
iris2 <- dplyr::filter(.data = iris, Species != "setosa")

# let's change the levels of our factors, a common routine in data analysis
# pipeline, to see if this function respects the new factor levels
iris2$Species <- factor(x = iris2$Species, levels = c("virginica", "versicolor"))

# plot
ggstatsplot::ggbetweenstats(
  data = iris2,
  x = Species,
  y = Sepal.Length,
  notch = TRUE, # show notched box plot
  mean.plotting = TRUE, # whether mean for each group is to be displayed
  mean.ci = TRUE, # whether to display confidence interval for means
  mean.label.size = 2.5, # size of the label for mean
  type = "parametric", # which type of test is to be run
  k = 3, # number of decimal places for statistical results
  outlier.tagging = TRUE, # whether outliers need to be tagged
  outlier.label = Sepal.Width, # variable to be used for the outlier tag
  outlier.label.color = "darkgreen", # changing the color for the text label
  xlab = "Type of Species", # label for the x-axis variable
  ylab = "Attribute: Sepal Length", # label for the y-axis variable
  title = "Dataset: Iris flower data set", # title text for the plot
  ggtheme = ggthemes::theme_fivethirtyeight(), # choosing a different theme
  ggstatsplot.layer = FALSE, # turn off ggstatsplot theme layer
  package = "wesanderson", # package from which color palette is to be taken
  palette = "Darjeeling1", # choosing a different color palette
  messages = FALSE
)

As can be seen from the plot, the function by default returns Bayes Factor for the test (here, Student’s t-test). If the null hypothesis can’t be rejected with the null hypothesis significance testing (NHST) approach, the Bayesian approach can help index evidence in favor of the null hypothesis (i.e.,

).

By default, natural logarithms are shown because Bayes Factor values can sometimes be pretty large. Having values on logarithmic scale also makes it easy to compare evidence in favor alternative () versus null () hypotheses (since ).

Additionally, there is also a grouped_ variant of this function that makes it easy to repeat the same operation across a single grouping variable:

# for reproducibility
set.seed(123)

# plot
ggstatsplot::grouped_ggbetweenstats(
  data = dplyr::filter(
    .data = ggstatsplot::movies_long,
    genre %in% c("Action", "Action Comedy", "Action Drama", "Comedy")
  ),
  x = mpaa,
  y = length,
  grouping.var = genre, # grouping variable
  pairwise.comparisons = TRUE, # display significant pairwise comparisons
  pairwise.annotation = "p.value", # how do you want to annotate the pairwise comparisons
  p.adjust.method = "bonferroni", # method for adjusting p-values for multiple comparisons
  conf.level = 0.99, # changing confidence level to 99%
  ggplot.component = list( # adding new components to `ggstatsplot` default
    ggplot2::scale_y_continuous(sec.axis = ggplot2::dup_axis())
  ),
  k = 3,
  title.prefix = "Movie genre",
  caption = substitute(paste(
    italic("Source"),
    ":IMDb (Internet Movie Database)"
  )),
  palette = "default_jama",
  package = "ggsci",
  messages = FALSE,
  nrow = 2,
  title.text = "Differences in movie length by mpaa ratings for different genres"
)

Summary of tests

Following (between-subjects) tests are carried out for each type of analyses-

TypeNo. of groupsTest
Parametric> 2Fisher’s or Welch’s one-way ANOVA
Non-parametric> 2Kruskal–Wallis one-way ANOVA
Robust> 2Heteroscedastic one-way ANOVA for trimmed means
Bayes Factor> 2Fisher’s ANOVA
Parametric2Student’s or Welch’s t-test
Non-parametric2Mann–Whitney U test
Robust2Yuen’s test for trimmed means
Bayes Factor2Student’s t-test

The omnibus effect in one-way ANOVA design can also be followed up with more focal pairwise comparison tests. Here is a summary of multiple pairwise comparison tests supported in ggbetweenstats-

TypeEqual variance?Testp-value adjustment?
ParametricNoGames-Howell testYes
ParametricYesStudent’s t-testYes
Non-parametricNoDwass-Steel-Crichtlow-Fligner testYes
RobustNoYuen’s trimmed means testYes
Bayes FactorNoNoNo
Bayes FactorYesNoNo

For more, see the ggbetweenstats vignette: https://indrajeetpatil.github.io/ggstatsplot/articles/web_only/ggbetweenstats.html

ggwithinstats

ggbetweenstats function has an identical twin function ggwithinstats for repeated measures designs that behaves in the same fashion with a few minor tweaks introduced to properly visualize the repeated measures design. As can be seen from an example below, the only difference between the plot structure is that now the group means are connected by paths to highlight the fact that these data are paired with each other.

# for reproducibility and data
set.seed(123)
library(WRS2)

# plot
ggstatsplot::ggwithinstats(
  data = WRS2::WineTasting,
  x = Wine,
  y = Taste,
  sort = "descending", # ordering groups along the x-axis based on
  sort.fun = median, # values of `y` variable
  pairwise.comparisons = TRUE,
  pairwise.display = "s",
  pairwise.annotation = "p",
  title = "Wine tasting",
  caption = "Data from: `WRS2` R package",
  ggtheme = ggthemes::theme_fivethirtyeight(),
  ggstatsplot.layer = FALSE,
  messages = FALSE
)

As with the ggbetweenstats, this function also has a grouped_ variant that makes repeating the same analysis across a single grouping variable quicker. We will see an example with only repeated measurements-

# common setup
set.seed(123)

# getting data in tidy format
data_bugs <- ggstatsplot::bugs_long %>%
  dplyr::filter(.data = ., region %in% c("Europe", "North America"))

# plot
ggstatsplot::grouped_ggwithinstats(
  data = dplyr::filter(data_bugs, condition %in% c("LDLF", "LDHF")),
  x = condition,
  y = desire,
  xlab = "Condition",
  ylab = "Desire to kill an artrhopod",
  grouping.var = region,
  outlier.tagging = TRUE,
  outlier.label = education,
  ggtheme = hrbrthemes::theme_ipsum_tw(),
  ggstatsplot.layer = FALSE,
  messages = FALSE
)

Summary of tests

Following (within-subjects) tests are carried out for each type of analyses-

TypeNo. of groupsTest
Parametric> 2One-way repeated measures ANOVA
Non-parametric> 2Friedman test
Robust> 2Heteroscedastic one-way repeated measures ANOVA for trimmed means
Bayes Factor> 2One-way repeated measures ANOVA
Parametric2Student’s t-test
Non-parametric2Wilcoxon signed-rank test
Robust2Yuen’s test on trimmed means for dependent samples
Bayes Factor2Student’s t-test

The omnibus effect in one-way ANOVA design can also be followed up with more focal pairwise comparison tests. Here is a summary of multiple pairwise comparison tests supported in ggwithinstats-

TypeTestp-value adjustment?
ParametricStudent’s t-testYes
Non-parametricDurbin-Conover testYes
RobustYuen’s trimmed means testYes
Bayes FactorNoNo

For more, see the ggwithinstats vignette: https://indrajeetpatil.github.io/ggstatsplot/articles/web_only/ggwithinstats.html

ggscatterstats

This function creates a scatterplot with marginal distributions overlaid on the axes (from ggExtra::ggMarginal) and results from statistical tests in the subtitle:

ggstatsplot::ggscatterstats(
  data = ggplot2::msleep,
  x = sleep_rem,
  y = awake,
  xlab = "REM sleep (in hours)",
  ylab = "Amount of time spent awake (in hours)",
  title = "Understanding mammalian sleep",
  messages = FALSE
)

The available marginal distributions are-

  • histograms
  • boxplots
  • density
  • violin
  • densigram (density + histogram)

Number of other arguments can be specified to modify this basic plot-

# for reproducibility
set.seed(123)

# plot
ggstatsplot::ggscatterstats(
  data = dplyr::filter(.data = ggstatsplot::movies_long, genre == "Action"),
  x = budget,
  y = rating,
  type = "robust", # type of test that needs to be run
  conf.level = 0.99, # confidence level
  xlab = "Movie budget (in million/ US$)", # label for x axis
  ylab = "IMDB rating", # label for y axis
  label.var = "title", # variable for labeling data points
  label.expression = "rating < 5 & budget > 100", # expression that decides which points to label
  line.color = "yellow", # changing regression line color line
  title = "Movie budget and IMDB rating (action)", # title text for the plot
  caption = expression( # caption text for the plot
    paste(italic("Note"), ": IMDB stands for Internet Movie DataBase")
  ),
  ggtheme = hrbrthemes::theme_ipsum_ps(), # choosing a different theme
  ggstatsplot.layer = FALSE, # turn off ggstatsplot theme layer
  marginal.type = "density", # type of marginal distribution to be displayed
  xfill = "#0072B2", # color fill for x-axis marginal distribution
  yfill = "#009E73", # color fill for y-axis marginal distribution
  xalpha = 0.6, # transparency for x-axis marginal distribution
  yalpha = 0.6, # transparency for y-axis marginal distribution
  centrality.para = "median", # central tendency lines to be displayed
  messages = FALSE # turn off messages and notes
)

Additionally, there is also a grouped_ variant of this function that makes it easy to repeat the same operation across a single grouping variable. Also, note that, as opposed to the other functions, this function does not return a ggplot object and any modification you want to make can be made in advance using ggplot.component argument (available for all functions, but especially useful for this particular function):

# for reproducibility
set.seed(123)

# plot
ggstatsplot::grouped_ggscatterstats(
  data = dplyr::filter(
    .data = ggstatsplot::movies_long,
    genre %in% c("Action", "Action Comedy", "Action Drama", "Comedy")
  ),
  x = rating,
  y = length,
  label.var = title,
  label.expression = length > 200,
  conf.level = 0.99,
  k = 3, # no. of decimal places in the results
  xfill = "#E69F00",
  yfill = "#8b3058",
  xlab = "IMDB rating",
  grouping.var = genre, # grouping variable
  title.prefix = "Movie genre",
  ggtheme = ggplot2::theme_grey(),
  ggplot.component = list(
    ggplot2::scale_x_continuous(breaks = seq(2, 9, 1), limits = (c(2, 9)))
  ),
  messages = FALSE,
  nrow = 2,
  title.text = "Relationship between movie length by IMDB ratings for different genres"
)

Summary of tests

Following tests are carried out for each type of analyses. Additionally, the correlation coefficients (and their confidence intervals) are used as effect sizes-

TypeTestCI?
ParametricPearson’s correlation coefficientYes
Non-parametricSpearman’s rank correlation coefficientYes
RobustPercentage bend correlation coefficientYes
Bayes FactorPearson’s correlation coefficientNo

For more, see the ggscatterstats vignette: https://indrajeetpatil.github.io/ggstatsplot/articles/web_only/ggscatterstats.html

ggpiestats

This function creates a pie chart for categorical or nominal variables with results from contingency table analysis (Pearson’s test for between-subjects design and McNemar’s test for within-subjects design) included in the subtitle of the plot. If only one categorical variable is entered, results from one-sample proportion test (i.e., a goodness of fit test) will be displayed as a subtitle.

Here is an example of a case where the theoretical question is about proportions for different levels of a single nominal variable:

# for reproducibility
set.seed(123)

# plot
ggstatsplot::ggpiestats(
  data = ggplot2::msleep,
  x = vore,
  title = "Composition of vore types among mammals",
  messages = FALSE
)

This function can also be used to study an interaction between two categorical variables:

# for reproducibility
set.seed(123)

# plot
ggstatsplot::ggpiestats(
  data = mtcars,
  x = am,
  y = cyl,
  conf.level = 0.99, # confidence interval for effect size measure
  title = "Dataset: Motor Trend Car Road Tests", # title for the plot
  stat.title = "interaction: ", # title for the results
  legend.title = "Transmission", # title for the legend
  factor.levels = c("1 = manual", "0 = automatic"), # renaming the factor level names (`x`)
  facet.wrap.name = "No. of cylinders", # name for the facetting variable
  slice.label = "counts", # show counts data instead of percentages
  package = "ggsci", # package from which color palette is to be taken
  palette = "default_jama", # choosing a different color palette
  caption = substitute( # text for the caption
    paste(italic("Source"), ": 1974 Motor Trend US magazine")
  ),
  messages = FALSE # turn off messages and notes
)

In case of repeated measures designs, setting paired = TRUE will produce results from McNemar’s test-

# for reproducibility
set.seed(123)

# data
survey.data <- data.frame(
  `1st survey` = c("Approve", "Approve", "Disapprove", "Disapprove"),
  `2nd survey` = c("Approve", "Disapprove", "Approve", "Disapprove"),
  `Counts` = c(794, 150, 86, 570),
  check.names = FALSE
)

# plot
ggstatsplot::ggpiestats(
  data = survey.data,
  x = `1st survey`,
  y = `2nd survey`,
  counts = Counts,
  paired = TRUE, # within-subjects design
  conf.level = 0.99, # confidence interval for effect size measure
  package = "wesanderson",
  palette = "Royal1"
)
#> Note: 99% CI for effect size estimate was computed with 100 bootstrap samples.
#> # A tibble: 2 x 11
#>   `2nd survey` counts  perc N         Approve Disapprove statistic   p.value
#>   <fct>         <int> <dbl> <chr>     <chr>   <chr>          <dbl>     <dbl>
#> 1 Disapprove      720   45  (n = 720) 20.83%  79.17%          245  3.20e- 55
#> 2 Approve         880   55. (n = 880) 90.23%  9.77%           570. 6.80e-126
#>   parameter method                                   significance
#>       <dbl> <chr>                                    <chr>       
#> 1         1 Chi-squared test for given probabilities ***         
#> 2         1 Chi-squared test for given probabilities ***

Additionally, there is also a grouped_ variant of this function that makes it easy to repeat the same operation across a single grouping variable:

# for reproducibility
set.seed(123)

# plot
ggstatsplot::grouped_ggpiestats(
  dplyr::filter(
    .data = ggstatsplot::movies_long,
    genre %in% c("Action", "Action Comedy", "Action Drama", "Comedy")
  ),
  x = mpaa,
  grouping.var = genre, # grouping variable
  title.prefix = "Movie genre", # prefix for the facetted title
  label.text.size = 3, # text size for slice labels
  slice.label = "both", # show both counts and percentage data
  perc.k = 1, # no. of decimal places for percentages
  messages = FALSE,
  nrow = 2,
  title.text = "Composition of MPAA ratings for different genres"
)

Summary of tests

Following tests are carried out for each type of analyses-

Type of dataDesignTest
Unpaired contingency tablePearson’s test
Paired contingency tableMcNemar’s test
Frequency contingency tableGoodness of fit ()

Following effect sizes (and confidence intervals/CI) are available for each type of test-

TypeEffect sizeCI?
Pearson’s chi-squared testCramér’s VYes
McNemar’s testCohen’s gYes
Goodness of fitCramér’s VYes

For more, see the ggpiestats vignette: https://indrajeetpatil.github.io/ggstatsplot/articles/web_only/ggpiestats.html

ggbarstats

In case you are not a fan of pie charts (for very good reasons), you can alternatively use ggbarstats function which has a similar syntax-

# for reproducibility
set.seed(123)

# plot
ggstatsplot::ggbarstats(
  data = ggstatsplot::movies_long,
  x = mpaa,
  y = genre,
  sampling.plan = "jointMulti",
  title = "MPAA Ratings by Genre",
  xlab = "movie genre",
  perc.k = 1,
  x.axis.orientation = "slant",
  ggtheme = hrbrthemes::theme_modern_rc(),
  ggstatsplot.layer = FALSE,
  ggplot.component = ggplot2::theme(axis.text.x = ggplot2::element_text(face = "italic")),
  palette = "Set2",
  messages = FALSE
)

Note that p-values for results from one-sample proportion tests are displayed for each bar in the form of asterisks with the following convention:

  • :
  • :
  • :
  • :

And, needless to say, there is also a grouped_ variant of this function-

# setup
set.seed(123)

# smaller dataset
df <- dplyr::filter(
  .data = forcats::gss_cat,
  race %in% c("Black", "White"),
  relig %in% c("Protestant", "Catholic", "None"),
  !partyid %in% c("No answer", "Don't know", "Other party")
)

# plot
ggstatsplot::grouped_ggbarstats(
  data = df,
  x = relig,
  y = partyid,
  grouping.var = race,
  title.prefix = "Race",
  xlab = "Party affiliation",
  ggtheme = ggthemes::theme_tufte(base_size = 12),
  ggstatsplot.layer = FALSE,
  messages = FALSE,
  title.text = "Race, religion, and political affiliation",
  nrow = 2
)

Summary of tests

This is identical to the ggpiestats function summary of tests.

gghistostats

To visualize the distribution of a single variable and check if its mean is significantly different from a specified value with a one-sample test, gghistostats can be used. We will again demonstrate here that the outputs are ggplot objects and can be further modified using any of the ggplot extension packages:

# for reproducibility
set.seed(123)
library(firatheme)

# plot
ggstatsplot::gghistostats(
  data = ToothGrowth, # dataframe from which variable is to be taken
  x = len, # numeric variable whose distribution is of interest
  xlab = "Tooth length", # `x`-axis label
  title = "Distribution of Tooth Length", # title for the plot
  fill.gradient = TRUE, # use color gradient
  test.value = 10, # the comparison value for one-sample test
  test.value.line = TRUE, # display a vertical line at test value
  type = "bayes", # bayes factor for one sample t-test
  bf.prior = 0.8, # prior width for calculating the bayes factor
  messages = FALSE # turn off the messages
) + # modification outside of `ggstatsplot` using extension packages
  theme_fira() +
  scale_colour_fira()

The aesthetic defaults can be easily modified internally as well-

# for reproducibility
set.seed(123)

# plot
ggstatsplot::gghistostats(
  data = iris, # dataframe from which variable is to be taken
  x = Sepal.Length, # numeric variable whose distribution is of interest
  title = "Distribution of Iris sepal length", # title for the plot
  caption = substitute(paste(italic("Source:"), "Ronald Fisher's Iris data set")),
  type = "parametric", # one sample t-test
  conf.level = 0.99, # changing confidence level for effect size
  bar.measure = "mix", # what does the bar length denote
  test.value = 5, # default value is 0
  test.value.line = TRUE, # display a vertical line at test value
  test.value.color = "#0072B2", # color for the line for test value
  centrality.para = "mean", # which measure of central tendency is to be plotted
  centrality.color = "darkred", # decides color for central tendency line
  binwidth = 0.10, # binwidth value (experiment)
  bf.prior = 0.8, # prior width for computing bayes factor
  messages = FALSE, # turn off the messages
  ggtheme = hrbrthemes::theme_ipsum_tw(), # choosing a different theme
  ggstatsplot.layer = FALSE # turn off ggstatsplot theme layer
)

As can be seen from the plot, bayes factor can be attached (bf.message = TRUE) to assess evidence in favor of the null hypothesis.

Additionally, there is also a grouped_ variant of this function that makes it easy to repeat the same operation across a single grouping variable:

# for reproducibility
set.seed(123)

# plot
ggstatsplot::grouped_gghistostats(
  data = dplyr::filter(
    .data = ggstatsplot::movies_long,
    genre %in% c("Action", "Action Comedy", "Action Drama", "Comedy")
  ),
  x = budget,
  xlab = "Movies budget (in million US$)",
  type = "robust", # use robust location measure
  grouping.var = genre, # grouping variable
  normal.curve = TRUE, # superimpose a normal distribution curve
  normal.curve.color = "red",
  title.prefix = "Movie genre",
  ggtheme = ggthemes::theme_tufte(),
  ggplot.component = list( # modify the defaults from `ggstatsplot` for each plot
    ggplot2::scale_x_continuous(breaks = seq(0, 200, 50), limits = (c(0, 200)))
  ),
  messages = FALSE,
  nrow = 2,
  title.text = "Movies budgets for different genres"
)

Summary of tests

Following tests are carried out for each type of analyses-

TypeTest
ParametricOne-sample Student’s t-test
Non-parametricOne-sample Wilcoxon test
RobustOne-sample percentile bootstrap
Bayes FactorOne-sample Student’s t-test

Following effect sizes (and confidence intervals/CI) are available for each type of test-

TypeEffect sizeCI?
ParametricCohen’s d, Hedge’s g (central-and noncentral-t distribution based)Yes
Non-parametricr (computed as )Yes
Robust (Robust location measure)Yes
Bayes FactorNoNo

For more, including information about the variant of this function grouped_gghistostats, see the gghistostats vignette: https://indrajeetpatil.github.io/ggstatsplot/articles/web_only/gghistostats.html

ggdotplotstats

This function is similar to gghistostats, but is intended to be used when the numeric variable also has a label.

# for reproducibility
set.seed(123)

# plot
ggdotplotstats(
  data = dplyr::filter(.data = gapminder::gapminder, continent == "Asia"),
  y = country,
  x = lifeExp,
  test.value = 55,
  test.value.line = TRUE,
  test.line.labeller = TRUE,
  test.value.color = "red",
  centrality.para = "median",
  centrality.k = 0,
  title = "Distribution of life expectancy in Asian continent",
  xlab = "Life expectancy",
  messages = FALSE,
  caption = substitute(
    paste(
      italic("Source"),
      ": Gapminder dataset from https://www.gapminder.org/"
    )
  )
)

As with the rest of the functions in this package, there is also a grouped_ variant of this function to facilitate looping the same operation for all levels of a single grouping variable.

# for reproducibility
set.seed(123)

# removing factor level with very few no. of observations
df <- dplyr::filter(.data = ggplot2::mpg, cyl %in% c("4", "6"))

# plot
ggstatsplot::grouped_ggdotplotstats(
  data = df,
  x = cty,
  y = manufacturer,
  xlab = "city miles per gallon",
  ylab = "car manufacturer",
  type = "nonparametric", # non-parametric test
  grouping.var = cyl, # grouping variable
  test.value = 15.5,
  title.prefix = "cylinder count",
  point.color = "red",
  point.size = 5,
  point.shape = 13,
  test.value.line = TRUE,
  ggtheme = ggthemes::theme_par(),
  messages = FALSE,
  title.text = "Fuel economy data"
)

Summary of tests

This is identical to summary of tests for gghistostats.

ggcorrmat

ggcorrmat makes a correlalogram (a matrix of correlation coefficients) with minimal amount of code. Just sticking to the defaults itself produces publication-ready correlation matrices. But, for the sake of exploring the available options, let’s change some of the defaults. For example, multiple aesthetics-related arguments can be modified to change the appearance of the correlation matrix.

# for reproducibility
set.seed(123)

# as a default this function outputs a correlalogram plot
ggstatsplot::ggcorrmat(
  data = ggplot2::msleep,
  corr.method = "robust", # correlation method
  sig.level = 0.001, # threshold of significance
  p.adjust.method = "holm", # p-value adjustment method for multiple comparisons
  cor.vars = c(sleep_rem, awake:bodywt), # a range of variables can be selected
  cor.vars.names = c(
    "REM sleep", # variable names
    "time awake",
    "brain weight",
    "body weight"
  ),
  matrix.type = "upper", # type of visualization matrix
  colors = c("#B2182B", "white", "#4D4D4D"),
  title = "Correlalogram for mammals sleep dataset",
  subtitle = "sleep units: hours; weight units: kilograms"
)

Note that if there are NAs present in the selected variables, the legend will display minimum, median, and maximum number of pairs used for correlation tests.

Alternatively, you can use it just to get the correlation matrices and their corresponding p-values (in a tibble format).

# for reproducibility
set.seed(123)

# show four digits in a tibble
options(pillar.sigfig = 4)

# getting the correlation coefficient matrix
ggstatsplot::ggcorrmat(
  data = iris, # all numeric variables from data will be used
  corr.method = "robust",
  output = "correlations", # specifying the needed output ("r" or "corr" will also work)
  digits = 3 # number of digits to be dispayed for correlation coefficient
)
#> # A tibble: 4 x 5
#>   variable     Sepal.Length Sepal.Width Petal.Length Petal.Width
#>   <chr>               <dbl>       <dbl>        <dbl>       <dbl>
#> 1 Sepal.Length        1          -0.143        0.878       0.837
#> 2 Sepal.Width        -0.143       1           -0.426      -0.373
#> 3 Petal.Length        0.878      -0.426        1           0.966
#> 4 Petal.Width         0.837      -0.373        0.966       1

# getting the p-value matrix
ggstatsplot::ggcorrmat(
  data = ggplot2::msleep,
  cor.vars = sleep_total:bodywt,
  corr.method = "robust",
  output = "p.values", # only "p" or "p-values" will also work
  p.adjust.method = "holm"
)
#> # A tibble: 6 x 7
#>   variable    sleep_total sleep_rem sleep_cycle     awake   brainwt    bodywt
#>   <chr>             <dbl>     <dbl>       <dbl>     <dbl>     <dbl>     <dbl>
#> 1 sleep_total   0.        5.291e-12   9.138e- 3 0.        3.170e- 5 2.568e- 6
#> 2 sleep_rem     4.070e-13 0.          1.978e- 2 5.291e-12 9.698e- 3 3.762e- 3
#> 3 sleep_cycle   2.285e- 3 1.978e- 2   0.        9.138e- 3 1.637e- 9 1.696e- 5
#> 4 awake         0.        4.070e-13   2.285e- 3 0.        3.170e- 5 2.568e- 6
#> 5 brainwt       4.528e- 6 4.849e- 3   1.488e-10 4.528e- 6 0.        4.509e-17
#> 6 bodywt        2.568e- 7 7.524e- 4   2.120e- 6 2.568e- 7 3.221e-18 0.

# getting the confidence intervals for correlations
ggstatsplot::ggcorrmat(
  data = ggplot2::msleep,
  cor.vars = sleep_total:bodywt,
  corr.method = "spearman",
  output = "ci",
  p.adjust.method = "holm"
)
#> Note: In the correlation matrix,
#> the upper triangle: p-values adjusted for multiple comparisons
#> the lower triangle: unadjusted p-values.
#> # A tibble: 15 x 7
#>    pair                          r   lower    upper         p lower.adj
#>    <chr>                     <dbl>   <dbl>    <dbl>     <dbl>     <dbl>
#>  1 sleep_total-sleep_rem    0.7641  0.6344  0.8520  7.806e-13   0.5632 
#>  2 sleep_total-sleep_cycle -0.4888 -0.7155 -0.1689  4.530e- 3  -0.7609 
#>  3 sleep_total-awake       -1      -1      -1       0.         -1      
#>  4 sleep_total-brainwt     -0.5935 -0.7408 -0.3918  1.426e- 6  -0.7852 
#>  5 sleep_total-bodywt      -0.5346 -0.6727 -0.3605  1.931e- 7  -0.7121 
#>  6 sleep_rem-sleep_cycle   -0.3344 -0.6118  0.01617 6.1

Copy Link

Version

Down Chevron

Install

install.packages('ggstatsplot')

Monthly Downloads

11,029

Version

0.1.4

License

GPL-3 | file LICENSE

Issues

Pull Requests

Stars

Forks

Maintainer

Last Published

December 18th, 2019

Functions in ggstatsplot (0.1.4)