Do social science studies replicate poorly overall?
Half of social-science studies fail replication
A large, years-long project reported that roughly half of social-science studies did not pass replication tests. The results come from a substantial effort testing whether published findings can be reproduced when researchers attempt repeatability checks or related robustness assessments.
The central significance is not just that some results failed, but what the failure rate implies for how knowledge is being built in social and behavioral sciences. If a large share of claims can’t be reproduced, then researchers, policymakers, and institutions risk over-weighting findings that may be sample-specific, dependent on analytic choices, or hard to recreate with new data.
What replication attempts are trying to detect
Replication in science is designed to test whether an effect is real and stable rather than an artifact of measurement, sampling, or analysis. In social sciences, researchers also face added challenges such as variations in how constructs are operationalized, how participants are recruited, and how data are analyzed.
Why it matters for decisions
Social-science research frequently informs interventions and public policy: education methods, clinical approaches, platform design, and institutional reforms. A high non-replication rate means some evidence base may need more verification before it’s used to justify costly or wide-ranging actions.
Where the field may go next
The provided coverage aligns with a broader push in recent years toward improved transparency and better confidence in results—through preregistration, stronger statistical practices, and more rigorous reanalysis.
The bottom line: a major multi-year effort finds that a sizable portion of social-science claims do not hold up under replication pressure, raising concerns about the credibility and reliability of parts of the evidence base.