Within ENIGMA‐OCD, the meta‐ and mega‐analyses revealed comparable findings for the subcortical regions (Boedhoe et al., 2017), but the mega‐analytic approach appeared more sensitive for detecting subtle cortical abnormalities (Boedhoe et al., 2018). We recently empirically evaluated whether a meta‐analysis provides results comparable to a mega‐analysis, and which of the two analytic frameworks (multiple linear regression mega‐analysis model versus linear mixed‐effects random‐intercept mega‐analysis model) performs better. Effect sizes and standard error estimates (and 95% confidence intervals), and (where possible) model fit, assessed using the Bayesian information criterion, were used to evaluate which of the methods performs best. Although effect sizes were similar for the meta‐analysis and linear regression mega‐analysis, we showed that in the case of cross‐sectional structural MRI data a mega‐analysis performs better than a meta‐analysis (lower standard errors and narrower confidence intervals), and in a multi‐center study with moderate variation between cohorts, a linear mixed‐effects random‐intercept mega‐analytical framework seems to lead to the best model fit (Boedhoe et al., 2019).
