Meta-analysis

A meta-analysis is a study that statistically combines the results of many separate studies on the same question. Pooling the data gives a clearer, more reliable estimate than any single study can on its own.

Definition from Shrinktionary, reviewed by Shariq Refai, MD, MBA, FAPA (last reviewed 2026-06-26). Shrinktionary is the network's language layer and carries the canonical short definition; this page adds the research-context detail below. Full definition on Shrinktionary →

Each study on its own gives a rough answer. A meta-analysis pools those answers and gives more weight to the larger and more precise studies. The result is a single estimate of how well a treatment works, plus a measure of how much the studies agreed or disagreed with each other.

Most headline claims about anxiety treatment, such as "CBT has a moderate to large effect," come from meta-analyses. Pooling studies gives a steadier answer than any one trial. It can also expose problems, such as publication bias, where studies with disappointing results never get published.

Ten trials each test an SSRI for generalized anxiety disorder. Some show a large benefit, some a small one. A meta-analysis combines all ten into one effect size, so you can see the average across the whole set rather than picking one trial.

A meta-analysis can look precise even when it isn't. If it pools weak studies, the result is a tidy number built on shaky data. Strong disagreement between the studies, called heterogeneity, is a warning sign. And different inclusion rules produce different answers, so two meta-analyses of the same treatment won't always match.