Meta-analysis is the statistical part: taking the results of several studies that asked a comparable question and combining them into a single estimate with a confidence interval around it. It usually sits inside a systematic review, and it is optional. This page is about what pooling buys you, what it costs in assumptions, and the two situations where a pooled number is worse than no number at all.
What pooling actually buys
Power and precision. Several small studies, each too small to detect an effect on its own, can together produce an estimate precise enough to be useful, and the confidence interval around the pooled figure is usually much narrower than around any single study. That is the entire case for it, and it is a strong one when the studies are genuinely asking the same question of the same kind of people.
The condition every meta analysis study needs: comparability
Comparability. Pooling studies that measured different things in different populations produces a number that describes nothing real, and heterogeneity statistics exist to tell you when that has happened. A high heterogeneity figure is not a problem to be reported and moved past; it is the analysis saying these studies should probably not be added together. The correct response is often to describe rather than pool.
Publication bias, and why the funnel plot matters
If studies finding nothing are less likely to be published, the literature you can pool is already skewed toward positive results, and a meta-analysis of it will be confidently wrong. That is what funnel plots and their tests are looking for, and it is why a systematic review that searched only published journals is weaker than one that also looked for trial registries and grey literature.
Other questions people ask
What is meta analysis in research, in simple terms?
Combining the numerical results of several studies that asked a comparable question into one estimate, so that studies too small to answer on their own can answer together. It normally sits inside a systematic review rather than standing alone.
Do I need a meta-analysis for my systematic review?
No. Pool only if the studies are comparable enough that a combined number means something. Where they are not, a narrative synthesis is the right answer and saying so is better practice than pooling anyway.
What does high heterogeneity mean?
That the studies disagree more than chance explains, which usually means they are not measuring the same thing in the same population. It is a reason to reconsider pooling rather than a caveat to report alongside the pooled figure.