TimelyScholarFree Tools › Meta-Regression Bubble Plot
Fit a random-effects meta-regression and plot each study's effect size against a moderator, with bubbles sized by study weight and the fitted line with its confidence band.
FreeNo account neededYour data stays in your browser
| Slope | -0.0291 (95% CI -0.0432 to -0.0150)z = -4.04, p < 0.001 |
|---|---|
| Intercept | 0.2515 (SE 0.2491) |
| Residual τ² | 0.0763I² = 68.4% residual heterogeneity |
| Test for residual heterogeneity | QE(11) = 30.73, p = 0.001 |
| Variance explained (R²) | 75.6%Proportional reduction in τ² |
Bubble area reflects each study's weight; the shaded band is the 95% confidence interval for the regression line. Results match metafor's rma(..., mods = ~ x, method = "REML"). Associations across studies are observational and can be confounded.
Copy the columns from Excel or Google Sheets, with names in the first row, or load the example.
Estimates, tests and the figure update instantly in your browser.
Download the figure (SVG or PNG) and copy a results sentence.
Each bubble is one study; larger bubbles carry more weight in the analysis. The line shows how the expected effect changes with the moderator, and the shaded band is its 95% confidence interval.
The test of the moderator (QM) asks whether the slope differs from zero. R² estimates the share of between-study variance explained. Meta-regression findings are observational: they show associations across studies, not causal effects.
Odds ratios and risk ratios are analysed as logarithms so that effects are symmetrical around no effect. The axis can show the back-transformed values.
No. Across-study relationships can be confounded by other study characteristics (ecological bias).