What is meta-analysis?
Meta-analysis is the statistical step that combines the results of several studies into one pooled effect — more precise than any single study. It’s the optional capstone of a systematic review: a systematic review finds and appraises the evidence; a meta-analysis, when the studies are similar enough, pools it. The art is knowing when pooling clarifies and when it misleads.
How pooling works
Each study contributes its effect estimate (a mean difference, odds ratio, correlation, etc.), and the meta-analysis takes a weighted average — usually weighting by the inverse of each study’s variance, so larger, more precise studies count more. The output is a summary effect with a tighter confidence interval. Express the per-study effects on a common scale first; the Effect Size Converter helps when studies report different metrics.
The forest plot
The forest plot is the signature figure. Each study is a row showing its effect and confidence interval, with the box size reflecting its weight; a diamond at the bottom is the pooled effect. One glance tells you how consistent the studies are and where the overall estimate sits relative to the line of no effect.
Fixed-effect vs random-effects
- Fixed-effect assumes every study estimates the same true effect; differences are chance alone.
- Random-effects assumes the true effect varies across studies and estimates the average of that distribution — wider intervals, and usually the more realistic default, since real studies differ in population and method.
Heterogeneity — and when not to pool
Heterogeneity is real variation in the effect across studies, beyond chance. I² reports the share of total variation due to heterogeneity as a percentage — high I² is a warning that a single pooled number may paper over genuine differences. When studies are too clinically or methodologically diverse, or heterogeneity is high with no explanation, don’t pool: a structured narrative synthesis (e.g. SWiM) is more honest than a misleading average. Also check for publication bias — small studies with null results often go unpublished, which can inflate a pooled effect (a funnel plot helps spot it). All of this connects back to risk of bias and GRADE: a precise pooled estimate built on high-risk studies is still low-certainty evidence.
Get the free Systematic Review toolkit
A data-extraction form, a synthesis-planning worksheet, and a PRISMA checklist from The Dissertation Literature Review Sprint. We’ll email you the download link.
Frequently asked questions
What is a meta-analysis?
The statistical pooling of several studies’ results into one weighted summary effect — more precise than any single study.
What is a forest plot?
The standard figure: each study’s effect + CI as a row, with a diamond for the pooled effect.
Fixed vs random effects?
Fixed assumes one true effect; random allows it to vary across studies (wider intervals, usual default).
What is I²?
The percentage of total variation due to heterogeneity rather than chance — high I² warns against a single pooled number.
When should you not pool?
When studies are too diverse, heterogeneity is high without explanation, or outcomes aren’t comparable — use a narrative synthesis instead.