An adjusted random-effects model for binary-data meta-analysis

A new model with a variable size of random effect is introduced for the meta-analysis of 2 × 2 tables. The random-effects parameter has a simple interpretation in terms of sample size and offers a new measure of heterogeneity.