Who builds this
The Forest Plot Generator is maintained by a team of methodologists who spend their working days on systematic reviews and quantitative evidence synthesis. We have pooled data for clinical, public health, and social science reviews, prepared figures for journal submission, and answered the methods questions reviewers raise. The tool exists because we wanted a fast, correct way to draw a forest plot without opening statistical software for every quick check, and we made it free because most of the people who need it are students and researchers, not labs with budgets.
The people behind it
Our guides are written and reviewed by named contributors who work on systematic reviews and meta-analyses across healthcare, public health, behavioural, and social science research. Each article carries the reviewer who stands behind it, and you can read their background on their profile.
- Andrew Collins, PhD, Independent Research Advisor
- Rachel Bennett, MPH, Evidence Synthesis Consultant
- James Walker, MSc, Research Specialist
- Olivia Parker, PhD, Academic Researcher and Evidence Synthesis Expert
- Christopher Evans, MPH, Systematic Review Consultant
How the pooling is calculated
Every analysis follows the standard methods described in the Cochrane Handbook for Systematic Reviews of Interventions and in Borenstein, Hedges, Higgins, and Rothstein, Introduction to Meta-Analysis. Ratio measures such as odds ratios, risk ratios, and hazard ratios are analysed on the natural-log scale, where their sampling distributions are symmetric, and back-transformed only for display. Fixed-effect pooling uses inverse-variance weights; for dichotomous outcomes you can also choose Mantel-Haenszel pooling with the Robins-Breslow-Greenland variance. Random-effects pooling estimates the between-study variance, tau-squared, with the DerSimonian-Laird method and widens the weights accordingly.
Heterogeneity is summarised with Cochran’s Q, the I-squared statistic, tau-squared, and H-squared, and a 95 percent prediction interval is reported for random-effects analyses with at least three studies and genuine between-study variance. Zero cells in binary data receive the standard 0.5 continuity correction, and standardised mean differences carry the Hedges small-sample correction. These are the same choices a reviewer would expect, which is why the output lines up with what a statistician reproduces by hand.
How we check the numbers
The engine is validated against established packages. We compare its pooled estimates, confidence intervals, and heterogeneity statistics with the output of the metafor package in R and statsmodels in Python on the same inputs, and they agree to the precision that matters for reporting. If you ever see a discrepancy you cannot explain, we want to hear about it.
References
- Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, Welch VA, editors. Cochrane Handbook for Systematic Reviews of Interventions. Cochrane.
- Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. Introduction to Meta-Analysis. Wiley.
- Viechtbauer W. Conducting meta-analyses in R with the metafor package. Journal of Statistical Software.
Learn the method
If you are new to evidence synthesis, our guides cover the ideas behind the tool in plain language: start with what a meta-analysis is, learn how to read a forest plot, and work through the choice between fixed-effect and random-effects models. When you are ready, the forest plot tool and the study pooling calculator put the method to work.