Estimating the Reproducibility of Psychological Science
Open Science Collaboration (Brian A. Nosek et al.)
Science · 2015
Edited by Paweł Raja, PhD · Published · Updated
Abstract
The landmark large-scale audit of reproducibility in psychology, quantifying how often published experimental findings hold up under direct, high-powered replication.
Methodology
A collaborative effort replicating 100 studies from three top psychology journals using high-powered, pre-registered protocols, assessing replication by statistical significance, effect size, and subjective judgment.
Findings
Only about 36–39% of replications produced statistically significant results, and replication effect sizes were on average roughly half the originals. Findings that were surprising or weakly powered replicated worst, catalyzing reforms in pre-registration and reporting across the behavioral sciences.
Applied nudge
Discount surprising, single-study effects heavily; anchor design decisions on well-replicated, pre-registered findings rather than novel results that make good headlines.
Citation
Open Science Collaboration. (2015). Estimating the reproducibility of psychological science. Science, 349(6251), aac4716.
↗ Source page