Experimental science is undergoing a replication crisis. Estimates range from 75% to 90% of preclinical experiments fail to replicate. And researchers have an incentive to publish positive results (which appear to show that a treatment improves health) while downplaying negative ones (which indicate that a treatment doesn't help.) So when you see a study that appears to show that a drug is effective in animals, you can't necessarily trust that the effect is real.
This is a tool created by the Center for Open Science, a nonprofit that creates free and open source software to enable scientists to share and manage data and preregister their hypotheses. Preregistration allows us to commit our guiding hypotheses to public record before the experiments are completed, so that we can't cherry-pick significant effects after the fact, or refrain from acknowledging negative results. Data sharing allows researchers and the general public to learn from each other, accelerating scientific progress.
We believe in the movement for more open and reproducible science, and encourage more researchers to use preregistration and data sharing tools like the Open Science Framework.