x

Reproducibility

Reproducibility

Reproducibility Resources

The results of a research study are considered reliable only when they can be reproduced by independent scholars. As such, the systematic, ongoing replication of important research is an essential component of scientific progress. While this is undisputed, the vast majority of scientific research has never been reproduced, making it impossible to verify the accuracy of accepted findings. Stanford School of Medicine Professor John Ioannidis performed an analysis which suggests that approximately 90 percent of medical research is likely false. The pharmaceutical industry raised similar concerns, estimating that 70 percent of published cancer studies cannot be reproduced.

The problem is not limited to medicine. It exists in virtually every field of scientific inquiry and is largely related to the fact that scientific journals tend to favor new and novel research over replication experiments. Since funding is often tied to publication, scientists are frequently motivated to conduct original research instead of replications. This absence of reproducibility in the scientific process is putting people at risk and is stifling the development of drugs, interventions, and other treatments that could have a positive impact.

LJAF is working with members of the scientific community to promote reproducibility by sponsoring major replication projects. One is focused on replicating dozens of landmark cancer studies. It is being conducted by the Reproducibility Initiative, an effort launched by several prominent scientific journals and organizations, including Science and others. Independent validation of the studies could be crucial to the development of future cancer drugs.

The second project is the largest reproducibility effort ever undertaken in the field of psychology. More than 250 researchers conducted replications of 100 published studies. Strikingly, only 39 percent of the studies could be reproduced.  The research is generating broad discussion about the need for more replications.