Only about half of social science findings held up when retested, major replication project finds
Psychology’s replication problems drew wide attention after 2015. A new study in Nature suggests the challenge is not unique to psychology: across economics, education, business, sociology, and political science, success rates were similar.
A recent international study published in Nature has revealed that only half of the tested claims in the social and behavioral sciences can be reliably replicated by independent researchers. This finding extends across various disciplines, including psychology, economics, sociology, education, business, and political science. The average effects observed in replication attempts were found to be less than half the size of those originally reported.
Replication, in scientific research, involves retesting the same question using independent data to determine if the original conclusion still holds true. Successful replication is crucial for verifying the reliability of initial findings before they inform public policy, education, or clinical practice. The study, led by Andrew H. Tyner of the Center for Open Science along with Timothy M. Errington and Brian A. Nosek, analyzed 296 replication attempts across 62 journals in six disciplines.
Researchers collected new data for 98 papers and analyzed existing datasets for 66 papers, totaling 274 unique scientific claims. Despite using large sample sizes and adhering to rigorous standards, the median sample size in original studies was 206, while replication studies utilized 545 samples. The researchers employed the original study materials and subjected each study's design to an internal peer review process to ensure fairness and methodological rigor.
Written by urgent.news from PsyPost's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
