Fig. 2.

The magnitude of the effect size and the measurement error of the studies included in the (R)-[¹¹C]PK11195 meta-analysis by Marques et al. (left figure) show a high degree of association (r = 0.9). Potential reasons for such a shape are publication bias or inflated effect sizes in studies with unreliable outcomes and small sample sizes, leading to an inflated overall effect size. When Marques et al. corrected for this bias, the difference between healthy controls and patients with psychosis or schizophrenia was no longer statistically significant (right figure).