Bubble Plot for Meta-Regression — bubble

Creates a bubble plot showing the relationship between a continuous moderator and the effect size, with point sizes proportional to study precision (inverse variance).

Usage

bubble_plot(
  object,
  mod,
  ci = TRUE,
  ci_level = 0.95,
  size_scale = 1,
  show_studies = FALSE,
  xlab = NULL,
  ylab = NULL,
  title = NULL,
  color_palette = c("#4292C6", "#08519C"),
  theme = ggplot2::theme_minimal()
)

Arguments

object: A bayesma_reg object.
mod: Character. Name of the moderator variable to plot. Must be a continuous moderator.
ci: Logical. Show confidence/credible band for the regression line (default: TRUE).
ci_level: Numeric. Credible interval level (default: 0.95).
size_scale: Numeric. Scaling factor for bubble sizes (default: 1).
show_studies: Logical. Label study points (default: FALSE).
xlab, ylab: Character. Axis labels. If NULL, uses variable names.
title: Character. Plot title.
color_palette: Character vector of length 2. Colors for points and regression line.
theme: A ggplot2 theme (default: theme_minimal()).

Value

A ggplot object.

Details

The bubble plot is a standard visualization for meta-regression with continuous moderators. Each study is represented by a circle:

Position: x = moderator value, y = effect size
Size: Proportional to study weight (1 / variance)
Line: Regression line showing the moderator effect
Band: Credible interval for the regression line

For centered moderators, the x-axis shows the centered values by default. Use centered = FALSE in the original meta_reg() call if you want the original scale.

Examples

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

# Basic bubble plot
bubble_plot(fit, mod = "year")

# Customized plot
bubble_plot(fit, mod = "dose",
            xlab = "Dose (mg)",
            ylab = "Log Odds Ratio",
            title = "Dose-Response Relationship")
} # }