!! A boxplot with additional components

A boxplot (in the style of Tukey) with additional components: mean and its confidence intervals as well as jittered points.

gg_boxplot_plus(
  formula,
  data = NULL,
  cld = NULL,
  xlab = NULL,
  ylab = NULL,
  legend_title = NULL,
  sort_groups = c("no", "yes", "ascending", "descending", TRUE, FALSE),
  sort_fun = median,
  add_points = TRUE,
  notch = FALSE,
  varwidth = FALSE,
  add_mean_ci = TRUE,
  conf_level = 0.95,
  ci_boot_reps = 999,
  cld_color = "black",
  cld_y_adj = 0,
  cld_y_mult = 0.06,
  ci_x_adj = -0.3,
  points_x_adj = 0.3,
  gr_sep = "|",
  x_rotate = c(0, 30, 60, 90),
  ...
)

Arguments

formula	a formula with two variable names to analyze. First one is numeric, second one is a factor, e.g. y ~ group.
data	a data frame with data.
cld	a data frame with cld results (object of class cld_object).
xlab	(character) Title for the x-axis.
ylab	(character) Title the y-axis.
legend_title	(character) Title for the legend.
sort_groups	("no"|"yes"|"ascending"|"descending") Sort groups by position of either median or other statistic indicated in sort_fun.
sort_fun	A function that calculates one numeric statistic (name without quotes). May be median, mean, sd, var, IQR, or similar.
add_points	(TRUE|FALSE) If TRUE, jittered point are added to the left of the boxplot.
notch	(TRUE|FALSE) if FALSE (default) make a standard box plot. If TRUE, notched boxplot is drawn. If the notches of two plots do not overlap this is a ‘strong evidence’ that the two medians statistically signigicantly differ (Chambers et al, 1983, p. 62).
varwidth	(logical) If FALSE (default) make a standard box plot. If TRUE, boxes are drawn with widths proportional to the square-roots of the number of observations in the groups.
add_mean_ci	(TRUE|FALSE) If TRUE, mean with its confidence interval are added to the right of the boxplot.
conf_level	(numeric) Confidence level for confidence interval. Number from 0 to 1. Default is 0.95.
ci_boot_reps	(numeric) Number of bootstrap repetitions for mean confidence interval calculation.
cld_color	(character) Name of color for cld letters.
cld_y_adj	(numeric) cly y position adjustment factor (additive) for cld letters.
cld_y_mult	(numeric) cly y position adjustment factor (multiplicative).
ci_x_adj	(numeric) x position correction factor (additive) for mean confidence interval lines/dots.
points_x_adj	(numeric) x position correction factor (additive) for jittered points.
gr_sep	(character) Separator used if move than one grouping variabe is chosen. Default is "|".
x_rotate	(character|numeric) An angle to rotete x axis tick labels. Supported values are 0, 30, 60, and 90. If theme is added, this parameter has no effect.
...	arguments to sort_fun.

Value

A ggplot2 plot object.

Details

The boxplot compactly displays the distribution of a continuous numeric variable. It visualises five summary statistics (including the first quartile, the median, and the third quartile) as well as all "outlying" points individually.

The plot is based on ggplot2, thus ggplot2 elements can be added to modify the plot.

References

Chambers, J. M., Cleveland, W. S., Kleiner, B. and Tukey, P. A. (1983) Graphical Methods for Data Analysis. Wadsworth & Brooks/Cole.

See also

https://en.wikipedia.org/wiki/Box_plot

Examples

library(biostat)

# Example 1a
gg_boxplot_plus(~decrease, OrchardSprays)

# Example 1b
gg_boxplot_plus(decrease ~ treatment, OrchardSprays)
gg_boxplot_plus(~ decrease | treatment, OrchardSprays)


# Example 2
gg_boxplot_plus(decrease ~ treatment, OrchardSprays,
  sort_groups = "descending"
)


# Example 3a

res <- posthoc_anova(weight ~ Diet, data = ChickWeight)
cld_result <- make_cld(res)

gg_boxplot_plus(weight ~ Diet,
  data = ChickWeight,
  cld = cld_result
)

# Example 3b

gg_boxplot_plus(weight ~ Diet,
  data = ChickWeight,
  cld = cld_result,
  sort_groups = "descending",
  sort_fun = mean
)

# Example 3c: do simple transformations

gg_boxplot_plus(log(weight) ~ Diet,
  data = ChickWeight,
  sort_groups = "descending",
  sort_fun = mean
)

# Example 3d: facetting

gg_boxplot_plus(weight ~ as.factor(Time), data = ChickWeight) +
  facet_wrap("Diet")

# Example 4

res2 <- posthoc_anova(decrease ~ treatment, data = OrchardSprays)
cld_result2 <- make_cld(res2)

gg_boxplot_plus(decrease ~ treatment,
  data = OrchardSprays,
  cld = cld_result2,
  sort_groups = "descending",
  sort_fun = mean
)

gg_boxplot_plus(decrease ~ treatment,
  data = OrchardSprays,
  cld = cld_result2,
  sort_groups = "ascending",
  sort_fun = IQR
)

# Example 5

gg_boxplot_plus(~ weight | Diet + Chick, data = ChickWeight)