Violin + boxplot of a numeric variable by a grouping factor — plot_numeric_by

Create a violin plot with an overlaid boxplot showing the distribution of a numeric variable split by a factor. The function expects the grouping column to be a factor and the outcome to be numeric.

Usage

plot_numeric_by_3plusgroups(
  yvar,
  group,
  d,
  yvar_label = NULL,
  group_label = NULL,
  title = NULL,
  group_colors = NULL,
  show_effect_size = TRUE,
  n_boot = 1000,
  xaxis_labels_nchar_wrap = 20,
  pairwise_wilcoxon_test_letters = TRUE
)

Arguments

yvar: Character scalar. Column name (in d) of the numeric variable to plot. Must be length 1 and refer to a numeric column in d.
group: Character scalar. Column name (in d) of the grouping factor. Must be length 1 and refer to a factor column in d.
d: A data.frame containing the variables.
yvar_label: Optional character scalar for the y-axis label. Defaults to yvar. If supplied, must be length 1.
group_label: Optional character scalar for the x-axis label. Defaults to group. If supplied, must be length 1.
title: Optional character scalar for the plot title. Defaults to "<yvar> by <group>". If supplied, must be length 1.
group_colors: Optional character vector of colours to use for the grouping levels. Must be length equal to nlevels(d[[group]]) when supplied. Use colour names or hex codes.
show_effect_size: Logical scalar. If TRUE the subtitle will include epsilon-squared and p-value.
n_boot: Integer-like numeric scalar. Number of bootstrap samples for estimating CI of epsilon-squared. Default is 1000. Must be a single numeric value >= 10.
xaxis_labels_nchar_wrap: Numeric scalar. Maximum line width (number of characters) used to wrap group labels on the x-axis before appending the sample size. Defaults to 20. Must be a single positive numeric value.
pairwise_wilcoxon_test_letters: Logical scalar. If TRUE (default), run pairwise Wilcoxon tests, compute compact-letter group labels and display them above each group. If FALSE, skip tests and do not show letters.

Value

A list with components:

ggplot: The ggplot2 object produced by the function.
stats: A data frame with effect size results (epsilon-squared estimate, CI, p-value, sample size, etc.).
letters: A named character vector of compact-letter grouping codes generated by pairwise Wilcoxon tests, or NULL if pairwise_wilcoxon_test_letters was FALSE.

Examples

data(mtcars)
mtcars$gear <- factor(mtcars$gear)
p <- plot_numeric_by_3plusgroups(
  yvar = "mpg",
  group = "gear",
  d = mtcars
)
p$ggplot