Skip to contents

Forest plot of effect sizes with confidence intervals

Source: R/plots-screen.R
cr_plot_forest.Rd

Draws one row per contrast: the point estimate with its confidence interval, a reference line at the null, and the rows ordered by the estimate rather than alphabetically, so the ordering is computed from the data and cannot go stale.

Usage

cr_plot_forest(
  effects,
  estimate = "estimate",
  ci_low = "ci_low",
  ci_high = "ci_high",
  label = NULL,
  facet_by = NULL,
  colour_by = NULL,
  method = "cohens_d",
  reference = 0,
  order_by_estimate = TRUE,
  descending = FALSE,
  title = NULL,
  subtitle = NULL,
  x_lab = NULL,
  y_lab = NA
)

Arguments

effects

A data frame of effect sizes (for example the output of an effect-size grid), a cr_result, or a list of cr_results.

estimate, ci_low, ci_high

Names of the estimate and interval columns.

label

Name of the column identifying each row. NULL (default) picks the first of group, compound, treatment, contrast, label or term that is present.

facet_by

One or two column names to facet on, or NULL.

colour_by

Column mapped to colour, fill and shape, or NULL.

method

Effect-size method to keep when effects carries a method column with several methods (default "cohens_d").

reference

Position of the null reference line (default 0).

order_by_estimate

Order rows by the estimate (default TRUE). FALSE keeps the order of label.

descending

Order largest estimate at the top (default FALSE, which puts the most negative - typically the strongest protection - at the top).

title, subtitle, x_lab, y_lab

Plot labels. NULL uses a computed default; NA omits the label.

Value

A ggplot object.

Details

The default subtitle states which intervals exclude the reference. It is computed from the interval bounds for the same reason: the equivalent sentence written by hand in the source pipeline stopped matching the data the first time the data changed.

See also

cr_plot_screen(), cr_plot_effect_sizes()

Other screen figures: cr_plot_qc_gate(), cr_plot_sample_size(), cr_plot_screen(), cr_plot_specificity(), cr_save_plot()

Examples

effects <- data.frame(
  group = c("CompoundA", "CompoundB", "CompoundC"),
  estimate = c(-1.2, -0.35, 0.1),
  ci_low = c(-1.9, -0.95, -0.4),
  ci_high = c(-0.5, 0.25, 0.6)
)
cr_plot_forest(effects)