The reproducibility crisis was a measurement crisis all along
Authors
Abstract
Much of what we labelled irreproducible turns out to be under-specified: the method was never written down precisely enough to repeat. The analysis draws on a corpus assembled from institutional repositories, preprint servers and publisher metadata feeds, and the full dataset together with the code needed to reproduce every figure is deposited under a permissive licence. Limitations, including coverage gaps in non-English sources, are set out in the discussion.
Graphical Abstract
Keywords
Subjects