Pooled effect (meta-analysis)

The pooled effect is the weighted average of the effects of all studies. Precise studies (small standard error) count more. The random-effects model also takes into account how much the true effects vary between studies (τ²): the weights become more balanced and the interval wider than in a model with one common true effect. The 95% confidence interval gives the values for the average of the true effects that are compatible with the data.

Example: Correlation r = .15, 95% CI [.08, .22] over 16 studies: a small positive relationship between conscientiousness and treatment adherence (Molloy and colleagues, 2014).

Caution: The average only makes sense if the studies measure the same thing. With high heterogeneity, also look at the prediction interval.