Investigating replicability in the social and behavioral sciences

Explainer of Tyner et al. (2026), Nature. Not the journal article.

Citation: Tyner, A., ..., Shakya, S., ... et al. (2026). Investigating the replicability of the social and behavioural sciences. Nature. DOI. PDF

Facts¶

Main finding¶

Systematic replication study examining replicability of published findings across social and behavioral sciences identifies patterns in which result types successfully replicate.

One caveat¶

Replication success depends on exact following the protocol and resource constraints may prevent perfect reproduction of original conditions.

Policy hook¶

How replicable is social and behavioral science research? Systematic assessment can identify which finding types replicate reliably and which warrant skepticism before policy application.

Data and setting¶

Large-scale replication study of social and behavioral science findings; Systematizing Confidence in Open Research and Evidence (SCORE) project in Nature publication.

Research design (plain language)¶

Coordinated replication of many published studies using standardized protocols to measure what percentage of original findings successfully replicate.

PDF or DOI¶

DOI · PDF

Why it matters¶

Science depends on cumulative knowledge, but cumulative knowledge is undermined if published findings don’t replicate. SCORE’s large-scale replication project is landmark work because it measures what percentage of social and behavioral science findings replicate with new data. If replication rates are high, the field is trustworthy; if low, policies built on social science are on shaky ground. For policy makers, the results inform how much confidence to place in social science evidence. For researchers, replication measures identify which finding types are strong versus fragile, directing future work toward understanding what makes results stable. The identification of patterns (which research areas replicate well, which don’t) is especially valuable - it guides researchers toward more reliable methods. For public trust in science, transparency about replication problems (and successes) builds credibility. The caveat that replication depends on following the protocol is important: negative results don’t always mean the original was wrong, but they do signal the need for caution before policy adoption. SCORE data provides a new gold standard for evidence quality assessment.