Skip to contents

Sample-size comparison plot

Source: R/plots-diagnostics.R
cr_plot_sample_size.Rd

Draws, per group, the number of units a confirmatory study would need when powered on the observed effect beside the number needed when powered on the confidence bound nearer the null.

Usage

cr_plot_sample_size(
  sizes,
  label = NULL,
  observed = "n_observed",
  conservative = "n_conservative",
  available = NULL,
  colour_by = NULL,
  log_y = TRUE,
  title = NULL,
  subtitle = NULL,
  caption = NULL,
  x_lab = NA,
  y_lab = NULL
)

Arguments

sizes

A data frame with one row per group, carrying the two sample-size columns.

label

Name of the grouping column. NULL (default) picks the first of group, compound, treatment or label that is present.

observed, conservative

Names of the two sample-size columns.

available

Optional name of a column giving the units already available per group; folded into the axis labels in brackets.

colour_by

Column mapped to fill, or NULL for a single colour.

log_y

Draw the count axis on a log scale (default TRUE, because the two bars routinely differ by an order of magnitude).

title, subtitle, caption, x_lab, y_lab

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

Value

A ggplot object.

Details

The gap between the two bars is the point of the figure. Powering on the observed effect of a screen's top hit is circular - that group is the largest only because it was selected for being largest - so the conservative bar is the one a confirmatory design should be built on, and a conservative figure exists only where the interval excludes the null.

Value labels are drawn in ink above the bars, never inside them, and the number of units already available is folded into the axis label rather than plotted as a marker that collides with the labels on short bars.

See also

cr_plot_forest()

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

Examples

sizes <- data.frame(
  group = c("CompoundA", "CompoundB", "CompoundC"),
  n_observed = c(12, 84, 640),
  n_conservative = c(46, NA, NA),
  n_available = c(6, 6, 6)
)
cr_plot_sample_size(sizes, available = "n_available")