Model Variants

The independent model fits tests without shared parameters and is equivalent to fitting each test independently. The model variants trade off flexibility, borrowing of strength, and data requirements. The ANOVA-type model assumes tests evaluated in the same study are more similar than tests evaluated in different studies. The ANOVA-plus model adds a hierarchical distribution over test-specific variances to include tests reported in only one study. The meta-regression model borrows strength by sharing the random-effect covariance matrix across tests.