Preregistration, Data Sharing, and Other Salutary Behaviors

Abstract

The most powerful and relatively painless strategies for improving the reproducibility of scientific research involve the preregistration of all proposed empirical research efforts in publicly accessible registries (accompanied by any procedural changes accruing prior to or during study conduct), the availability of research materials and data accruing from said research efforts (also publically registered), and the evaluation of these by an outside party comparing the preregistration document with the final research report prior to publication. The US preregistration movement was boosted approximately two decades ago by federal legislation requiring the preregistration of all government-funded clinical trials. Since then significant progress has been observed in other, nonmedical disciplines although the results are disappointing when preregistered protocols are compared directly to their published counterparts. Data registration and its codes have also increasingly become a publication requirement for many journals, although compliance remains imperfect. Some progress has been made in all of these arenas, bolstered by initiatives such as the use of “badges” accompanying published articles whose authors certify the availability of appropriately preregistered protocols and the willingness to share all accruing study data plus any pertinent materials that may be required for direct replications.