When literatures are shaped by publication bias and small studies, meta‑analyses can exaggerate true effects more than a well‑designed single study. Funnel plots frequently show asymmetry, and simple corrections (e.g., trim‑and‑fill) substantially shrink pooled estimates. Trust should be weighted toward study quality and bias diagnostics, not the mere size of a literature.
— This warns policymakers and journalists against treating 'the literature says' as dispositive and pushes for bias‑aware evidence standards before adopting interventions.
Tyler Cowen
2025.12.01
85% relevant
The article documents that after correcting for publication bias and assessing study quality (ROBINS‑E, GRADE), the apparent negative effect of inequality on mental health vanishes—concretely illustrating the existing idea that meta‑analytic findings can be inflated and need bias‑aware diagnostics.
2025.10.07
68% relevant
Jussim argues that a large share of peer‑reviewed psychology claims are false, foregrounding widespread non‑replication and propagation of unreplicable findings—echoing the critique that pooled literatures and selective methods can inflate effects and mislead policy.
2025.10.07
100% relevant
The article’s funnel plots and trim‑and‑fill re‑estimates for air‑pollution and mindfulness literatures that markedly reduce pooled effects.
2025.10.07
78% relevant
The article cites Maier et al. reporting that, after correcting publication bias, average nudge effects vanish, and a mega‑dataset from UK/US nudge units showing weaker impacts than published studies—classic signs that pooled literatures can inflate effect sizes.