Forest plot layer (estimate + interval)

Draws an estimate point with its confidence interval per row – the standard meta-analysis / effect-size panel. This is a syntax-sugar composite mark combining mark_errorbar() and mark_point(), plus a vertical reference rule when ref is supplied.

Usage

mark_forest(
  plot,
  mapping = NULL,
  data = NULL,
  ref = NULL,
  point_size = 2,
  bar_width = 0.4,
  line_color = ._MARK_STYLE$soft,
  line_width = ._MARK_STYLE$lw_thin,
  ...
)

Arguments

plot

A plotit object

mapping

Optional new aesthetics (must include x, y, xmin, xmax)

data

Optional data for this layer

ref

Reference value for the null-effect rule (e.g. 0 for differences, 1 for ratios). NULL (default) draws no rule.

point_size

Size of the estimate points (default 2).

bar_width

Width of the interval bars as a fraction of the categorical slot (default 0.4).

line_color

Colour for the interval bars and reference rule (default ._MARK_STYLE$soft = "grey50").

line_width

Stroke width for interval bars (default 0.5).

...

Other arguments passed to the estimate point layer

Value

Modified plotit object

Details

Equivalent expansion:

  p |> mark_errorbar(width = 0.3) |>
       mark_point(size = 2) |>
       mark_rule(xintercept = ref, linetype = "dashed")

Each row needs y (the study/category label position), x (the estimate) and xmin/xmax (the interval); map them through encode(). This is the standard horizontal forest; for a vertical forest, flip the whole plot afterwards with project_cartesian(flip = TRUE).

References

tidyplots: add_ci95_errorbar() + add_mean_dot() + add_reference_lines()

Vega-Lite: point + errorbar layer composition

Examples

studies <- data.frame(
  trial = paste0("Trial ", 1:5),
  es    = c(0.42, 0.31, 0.55, 0.20, 0.48),
  lo    = c(0.10, -0.05, 0.30, -0.10, 0.22),
  hi    = c(0.74, 0.67, 0.80, 0.50, 0.74)
)
studies |>
  plotit(encode(x = es, y = trial, xmin = lo, xmax = hi)) |>
  mark_forest(ref = 0) |>
  project_cartesian(flip = TRUE)