Subgroup Analysis and Meta-Regression Calculator with Bubble Plot
High I² in your meta-analysis? Find out why. Add one or two study-level moderators (for example latitude, dose, year, region or study design) and run a subgroup analysis with a test for subgroup differences, or a mixed-effects meta-regression with coefficients, the omnibus QM test, residual heterogeneity, R² and a bubble plot.
Type the moderator names in the last two column headers. Numbers are treated as continuous, text as categories (the first category in alphabetical order is the reference).
- PriceFree, no login or sign-up needed
- Your dataProcessed in your browser, never uploaded
- Works onPhone, tablet or computer, any modern browser
- Made byDr. Muhammad Aammar Tufail, Codanics
- Published17 November 2026
- Last updated16 November 2026
📚 How to cite this tool
This tool is free. If it helped your thesis, paper, report or assignment, please cite it. It keeps the tools free and helps other students find them.
Tufail, M. A. (2026). Subgroup analysis and meta-regression calculator with bubble plot [Online tool]. Codanics. Retrieved 16 November 2026, from https://codanics.com/tools/meta-regression-calculator/
In-text citation: (Tufail, 2026) · Narrative: Tufail (2026)
The access date is today’s date, filled in automatically. Author: Muhammad Aammar Tufail. Published 17 November 2026.
How to use the subgroup analysis and meta-regression calculator
- Choose the type of data and the effect measure, then enter one row per study. Type the names of up to two moderators in the last two column headers and their values for each study (numbers such as latitude or dose, or categories such as region or design).
- Or paste/upload your extraction sheet. Load example opens the classic BCG vaccine trials with latitude and allocation method.
- Pick the analysis: meta-regression with all moderators, or a subgroup analysis by moderator 1 or 2. REML and the standard tests are good defaults; Knapp–Hartung is safer with few studies.
- Click Run analysis to get the coefficients, QM test, residual heterogeneity, R² and a bubble plot, or subgroup estimates with the test for subgroup differences and a forest plot.
- Copy the results paragraph and download the plots and the coefficient table.
Subgroup analysis vs meta-regression
When studies disagree more than chance would explain (high I² or τ²), you can ask whether study characteristics, called moderators or effect modifiers, explain the differences. A subgroup analysis pools the studies separately within categories (for example region or study design) and tests whether the subgroup estimates differ. A meta-regression generalises this to continuous moderators (dose, latitude, year, mean age) and to several moderators at once, using a weighted regression in which each study’s weight is 1/(vᵢ + τ²).
Methods and formulas
Categorical moderators are dummy coded, with the first category in alphabetical order as the reference (as R’s factor() does). For ratio measures the coefficients are on the log scale; exponentiate them to get the ratio of risk ratios or odds ratios.
Worked example: BCG vaccine and latitude
The thirteen BCG trials (Colditz et al., 1994) are very heterogeneous (I² = 92%). A meta-regression of the log risk ratio on absolute latitude (REML) gives a slope of −0.0291 per degree (SE 0.0072, p < .001): the vaccine works better further from the equator. Latitude explains about 76% of the between-trial variance (τ² drops from 0.313 to 0.076; QM(1) = 16.36). A subgroup analysis by allocation method finds no clear difference. We checked every number against R: rma(yi, vi, mods = ~ ablat, data = dat), mods = ~ ablat + year with test = "knha", mods = ~ factor(alloc) and separate subgroup models, in metafor 5.2. Estimates agree to 5–6 significant digits (metafor stops its REML iterations a little earlier than this tool).
Frequently asked questions
When should I do a meta-regression instead of a subgroup analysis?
Use a subgroup analysis for a categorical characteristic (region, design, risk of bias). Use meta-regression for continuous characteristics (dose, year, mean age, latitude) or when you want to look at two moderators together. A meta-regression with one categorical moderator and a common τ² is the same as a subgroup analysis that assumes a common τ².
What is R² in meta-regression?
It is the proportion of the between-study variance (τ²) explained by the moderators: (τ² without moderators − residual τ²) / τ² without moderators. It can be unstable with few studies.
Which test for subgroup differences does this tool report?
Two: the RevMan/meta approach that pools each subgroup with its own τ² and compares them with a Q test, and the mixed-effects model with a common τ² (metafor’s QM test). They often agree; if they don’t, report the one your protocol specifies.
How many studies do I need?
The Cochrane Handbook advises against meta-regression with fewer than about 10 studies per moderator. With fewer studies, results are exploratory and the Knapp–Hartung adjustment is recommended.
Can I use the results to say the moderator causes the difference?
No. Meta-regression is observational: studies differ in many ways at once, so an association with a moderator may be confounded by other study characteristics.
Is my data uploaded?
No. All calculations run in your browser.
Do I need to log in or create an account to use the subgroup analysis and meta-regression calculator?
No. There is no login, no sign-up and no email required. Open the page and start using it straight away, for free, on any device.
How do I cite this tool in my thesis or paper?
Use the citation box on this page. In APA 7th edition: Tufail, M. A. (2026). Subgroup analysis and meta-regression calculator with bubble plot [Online tool]. Codanics. Retrieved [date you used it], from https://codanics.com/tools/meta-regression-calculator/. BibTeX, RIS (EndNote, Mendeley, Zotero), MLA, Harvard, Chicago and Vancouver formats are also available.
Free guides and courses from Codanics
More free tools
See all free Codanics tools → · Codanics on YouTube · WhatsApp channel

