StatisticalAdvanced

Galbraith Plot

A radial scatter plot used in meta-analysis to assess heterogeneity — each study is plotted as precision versus standardized effect, with outliers from the regression line signalling between-study variability.

// 01 — The chart

What it looks like

Example — Meta-analysis of treatment efficacy8 studies
Precision (1/SE)z-score (effect / SE)0246802468ABCDEFGH (outlier)

A Galbraith plot of eight studies. Study H falls outside the 95% confidence band (dashed lines), indicating it contributes to heterogeneity. The solid line through the origin represents the pooled effect.

// 02 — Definition

What is a Galbraith plot?

A Galbraith plot (also called a radial plot) is a scatter plot used in meta-analysis to visually assess heterogeneity among studies. Each study is represented as a single point, with precision (the reciprocal of the standard error, 1/SE) on the x-axis and the standardized effect (z-score = effect size divided by SE) on the y-axis.

A regression line is drawn through the origin whose slope equals the pooled effect estimate. Studies that lie close to this line are consistent with the overall result. Two dashed lines at ±2 units from the regression line form a 95% confidence band. Studies falling outside this band are the primary contributors to between-study heterogeneity.

Unlike a forest plot which displays each study on its own row, the Galbraith plot condenses all studies into a single two-dimensional space. This makes it especially useful for meta-analyses with many studies, where a forest plot becomes unwieldy. It also avoids the visual dominance of imprecise studies that can occur in funnel plots.

Origin: The Galbraith plot was introduced by Rex Galbraith in 1988 as a “radial plot” for fission-track dating in geology. It was subsequently adopted by the medical statistics community for meta-analysis, where it provides a complementary view to the forest plot and funnel plot for detecting sources of heterogeneity.

// 03 — Anatomy

Parts of a Galbraith plot

ABCDE
A — Y-axis (z-score): Standardized effect (effect/SE), centering studies on a common scale regardless of metric
B — X-axis (precision): Inverse of the standard error (1/SE); larger studies appear further right with higher precision
C — Regression line: A line through the origin whose slope equals the pooled effect estimate from the meta-analysis
D — 95% confidence band: Dashed lines at ±2 from the regression line; studies outside this band are heterogeneous
E — Outlier study: A point outside the confidence band, indicating that study's result is inconsistent with the pooled estimate

// 04 — Usage

When to use it — and when not to

✓Use a Galbraith plot when…
  • Assessing heterogeneity among studies in a meta-analysis
  • Identifying specific studies that are outliers contributing to high I²
  • You have many studies (20+) and a forest plot is too dense to read
  • You want to complement a forest plot with a heterogeneity diagnostic
  • Comparing how well individual studies align with the pooled effect
  • Presenting meta-analysis results to a statistically literate audience
×Avoid a Galbraith plot when…
  • You have fewer than five studies — too few points to assess heterogeneity visually
  • Your audience is not familiar with meta-analysis concepts
  • You need to display individual study effect sizes and confidence intervals — use a forest plot
  • You want to detect publication bias — use a funnel plot instead
  • The analysis uses non-standard effect measures that can’t be standardized
  • You need a standalone visualization — it works best as a companion plot

// 05 — Reading guide

How to read a Galbraith plot

Follow these steps whenever you encounter a Galbraith plot in a meta-analysis.

1

Locate the regression line

The solid line passes through the origin. Its slope equals the pooled effect estimate. This is your reference — all studies should ideally cluster around it if there is no heterogeneity.

2

Identify the confidence band

Two dashed lines run parallel to the regression line at ±2 units on the y-axis. This band represents the 95% confidence region. Points inside the band are consistent with the pooled effect.

3

Check study positions along the x-axis

Studies further to the right are more precise (lower standard error, typically larger sample size). These studies contribute more weight to the pooled estimate. Studies on the left are less precise.

4

Spot outliers outside the band

Any study point falling outside the dashed confidence lines is an outlier. These studies have effects that differ significantly from the pooled estimate and are the main drivers of heterogeneity.

5

Assess overall scatter

If most points cluster tightly around the regression line, heterogeneity is low. If points are scattered widely, especially outside the band, heterogeneity is substantial and the pooled estimate may not be reliable without further investigation.

// 06 — Pitfalls

Common mistakes

Not forcing the regression line through the origin

Fix: The regression must pass through (0, 0). If an intercept is allowed, the slope no longer equals the pooled effect, and the confidence band becomes meaningless for heterogeneity detection.

Confusing it with a funnel plot

Fix: Both are scatter plots used in meta-analysis, but they serve different purposes. The funnel plot detects publication bias (effect vs. SE). The Galbraith plot detects heterogeneity (z-score vs. precision). Don’t swap them.

Using it as a standalone diagnostic

Fix: The Galbraith plot should complement — not replace — the forest plot and heterogeneity statistics (I², Q). It’s most useful for identifying which specific studies drive heterogeneity, not for quantifying it.

Ignoring scale transformations

Fix: For ratio measures (OR, RR), use log-transformed values when computing z-scores. Raw ratios violate the linearity assumption and distort the regression line.

Overinterpreting with few studies

Fix: With fewer than five or six studies, the scatter pattern is unreliable. A single outlier in a small set may look dramatic but could be due to chance. Reserve the plot for meta-analyses with enough data points.

// 07 — In the wild

Real-world examples

Cochrane meta-analyses of drug efficacy

When a Cochrane review includes 30+ trials of a pharmaceutical intervention, the forest plot becomes difficult to scan. A Galbraith plot supplements it by immediately flagging which trials fall outside the confidence band, helping reviewers decide whether to perform sensitivity analyses excluding those trials.

Epidemiological risk factor studies

Large-scale meta-analyses of observational data — such as the association between a dietary factor and cancer risk — often show high heterogeneity. A Galbraith plot helps identify whether the heterogeneity comes from a few outlier cohorts with unusual populations or methodologies.

Geological fission-track dating

The Galbraith plot originated in geology, where Rex Galbraith used it to display age estimates from fission-track counting. Each grain measurement is plotted as precision vs. standardized age, and outliers reveal grains with different thermal histories — the same principle now applied to clinical trials.

// 08 — Quick reference

Key facts

Also known asRadial plot
Best forDetecting heterogeneity sources in meta-analysis
Data typesEffect sizes with standard errors from multiple studies
Key elementsPrecision axis, z-score axis, regression line, confidence band
X-axis1/SE (precision)
Y-axisEffect/SE (z-score)
Common toolsR (metafor, meta), Stata (metafunnel), Python (statsmodels)
Common mistakesNon-origin intercept, confusion with funnel plot, too few studies

// 09 — Variations

Types of Galbraith plots

The Galbraith plot adapts to different meta-analytic contexts while maintaining its core structure of precision vs. standardized effect.

Standard Galbraith plot

The classic format with 1/SE on x-axis, z-score on y-axis, and a regression line through the origin with 95% confidence band.

Weighted Galbraith plot

Point sizes are proportional to study weight, making it easier to see which studies influence the pooled estimate most.

Color-coded subgroup plot

Studies are colored by subgroup (e.g., RCTs vs. observational) to reveal whether heterogeneity clusters within certain study types.

Adjusted Galbraith plot

Includes a second regression line from a random-effects model, allowing comparison between fixed- and random-effects pooled estimates.

// 10 — FAQs

Frequently asked questions

What is a galbraith plot?+

A Galbraith plot (also called a radial plot) is a scatter plot used in meta-analysis to visually assess heterogeneity among studies. Each study is represented as a single point, with precision (the reciprocal of the standard error, 1/SE) on the x-axis and the standardized effect (z-score = effect size divided by SE) on the y-axis.

When should you use a galbraith plot?+

Use a galbraith plot when assessing heterogeneity among studies in a meta-analysis. It also works well when identifying specific studies that are outliers contributing to high I², and when you have many studies (20+) and a forest plot is too dense to read.

When should you avoid a galbraith plot?+

Avoid a galbraith plot when you have fewer than five studies — too few points to assess heterogeneity visually. It is also a poor fit when your audience is not familiar with meta-analysis concepts, or when you need to display individual study effect sizes and confidence intervals — use a forest plot.

Is a galbraith plot suitable for dashboards?+

Yes — a galbraith plot can work well in dashboards as long as the panel is large enough for readers to perceive the encoded values, has a clear title, and includes the legend or axis labels needed to interpret it.

What category of chart is a galbraith plot?+

Galbraith Plot belongs to the Statistical family of charts. Charts in that family are designed to answer the same kind of question, so they often work as alternatives when one doesn't quite fit your data.

How do you read a galbraith plot?+

Start with the axis labels and legend, then look at the overall shape before zooming into individual marks. Compare prominent features against the rest of the data, and verify any conclusion against the underlying numbers when precision matters.