When Meta‑Analyses Mislead

Meta‑analysis can amplify systematic distortions when the underlying literature suffers from publication bias, p‑hacking, or selective reporting; in such cases a well‑conducted single study (or an explicitly bias‑corrected analysis) may provide a more reliable guide. The post explains funnel‑plot asymmetry, 'trim‑and‑fill' correction, and gives concrete topical examples where pooled estimates exceed realistic effects. — This reframes how media, courts, and policymakers should treat 'the literature says' claims—demanding provenance, bias diagnostics, and robustness maps rather than relying on pooled estimates alone.