The issue of low data robustness and nonreplicable results in biomedical research is accompanied by an increasing interest in reforming research incentives and practices. How to optimally perform these reforms is a scientific problem, and economics has several scientific methods that can help evaluate research reforms. In this article, Thomas Gall, John P.A. Ioannidis and Zacharias Maniadis have reviewed these methods and analysed their potential. Prominent among them are mathematical modelling and laboratory experiments that constitute affordable ways to approximate the effects of policies with wide-ranging implications.