Plot Method for bayesma_coef_evidence Objects — metareg_mod

Creates a forest-style plot for meta-regression coefficients using posterior density slabs, consistent with the forest visual style.

Usage

metareg_mod_plot(
  x,
  model = NULL,
  include_intercept = FALSE,
  show_null_range = TRUE,
  null_value = 0,
  output = c("density", "pointinterval"),
  split_color_by_null = FALSE,
  color_posterior = "dodgerblue",
  color_posterior_outline = "blue",
  color_favours_positive = "dodgerblue",
  color_favours_negative = "firebrick",
  color_null_range = "grey50",
  color_pointinterval = "blue",
  xlim = NULL,
  x_breaks = NULL,
  xlab = "Coefficient Estimate",
  title = "Meta-Regression Coefficients",
  subtitle = NULL,
  add_table = TRUE,
  table_width = 0.4,
  font = NULL,
  ...
)

Arguments

x: A bayesma_coef_evidence object (created by coefficient_evidence()).
model: The bayesma_metareg model. Required to extract posterior draws for density slabs.
include_intercept: Logical. Include intercept in plot (default: FALSE).
show_null_range: Logical. Show null range region if available (default: TRUE).
null_value: Numeric. Value for the null reference line (default: 0).
output: Character. Either "density" (default) for density slabs or "pointinterval" for point + interval display.
split_color_by_null: Logical. If TRUE, posterior densities are split and coloured based on whether values fall above or below the null value (default: FALSE).
color_posterior: Color for coefficient posterior densities (default: "dodgerblue").
color_posterior_outline: Color for posterior outlines ( default: "blue").
color_favours_positive: Colour for density regions > null when split_color_by_null = TRUE (default: "dodgerblue").
color_favours_negative: Colour for density regions < null when split_color_by_null = TRUE (default: "firebrick").
color_null_range: Color for null range shading (default: "grey50").
color_pointinterval: Color for point intervals when output = "pointinterval" (default: "blue").
xlim: Numeric vector of length 2 specifying x-axis limits. Default is NULL (auto-scaled).
x_breaks: Numeric vector specifying custom x-axis break points. Default is NULL (auto).
xlab: Character. X-axis label (default: "Coefficient Estimate").
title: Character. Plot title (default: "Meta-Regression Coefficients").
subtitle: Character. Plot subtitle. Default shows null range if provided.
add_table: Logical. Add a table with coefficient summaries on the right (default: TRUE).
table_width: Numeric. Relative width of the table vs plot (default: 0.4).
font: Character. Font family for text elements (default: NULL).
...: Additional arguments (unused).

Value

A ggplot object (or patchwork object if add_table = TRUE).

Details

This function creates a forest-style visualisation for meta-regression coefficients that matches the aesthetic of forest(). Each coefficient is displayed as a posterior density slab (or point interval), allowing visualisation of the full posterior distribution rather than just point estimates and credible intervals

When split_color_by_null = TRUE, the density is split at the null value and coloured to show the proportion of the posterior on each side.

Examples

if (FALSE) { # \dontrun{
fit <- meta_reg(data, studyvar = "author", yi = "yi", vi = "vi",
                mods = ~ year + quality)

# Get evidence summary
ev <- coefficient_evidence(fit, null_range = c(-0.1, 0.1))

# Basic forest-style plot
metareg_mod_plot(ev, model = fit)

# With split coloring
metareg_mod_plot(ev, model = fit, split_color_by_null = TRUE)

# Point interval style
metareg_mod_plot(ev, model = fit, output = "pointinterval")
} # }