A manifesto for reproducible science (by Munafò et al., 2017)

Towards a future of valid and reliable research results.

When doing research, we want to minimise all kinds of biases as much as possible to ensure that findings are valid and reliable. To achieve this goal, it is really important to replicate past findings. In a nutshell, replications help us to understand when a research finding holds up and when it doesn’t. In order to promote transparency in research and the principles of open science, Munafò and colleagues suggest several measures that can be taken to improve the research process. 


The proposed measures concerning the method of a scientific study include blinding the researcher to the experimental conditions, to improve the methodological training for future researchers (especially in statistics), and ensure continuous education for present researchers. Furthermore, collaboration between researchers should be encouraged, for example in the form of distributed collaboration between several labs or departments, in order to ensure an interdisciplinary approach, high expertise and sufficient statistical power. 

Reporting and Dissemination

Regarding reporting and disseminating studies, the authors suggest to promote pre-registration, and to have reporting checklists specifying which statistics to report in a specific type of study (e.g., observational studies/meta-analyses, etc.). One important aspect of publishing a scientific paper is to disclose any conflicts of interest (e.g., between the researcher’s aim to conduct objective research and the interests of an organisation or institution funding the research project). 


One further crucial part in ensuring reproducible science is to align the academic incentive structure with the principles of open and reproducible scientific practices. For example, journals can award badges for open research practices, research associations can provide funding for replication studies, and open science practices can be included in performance evaluations and hiring decisions at universities and research institutes. 

The Empirical Cycle 

To finish, let’s briefly summarise what the empirical research process usually looks like. You start with observations that are imperfectly explained by existing theory. You develop a new theory (either from scratch or by adapting/integrating/combining existing theories) to explain these observations. From this new theory, you make a prediction (a hypothesis) about the phenomena you’re studying. You then test your hypothesis by collecting data that may confirm or disconfirm it. After evaluating whether your hypothesis was confirmed or not, you revisit and potentially revise and modify your theory, and the empirical cycle begins again. 

In an ideal world, the insights gained from your research help improve society for the better. Whether by informing social or organisational policies, leading to the development of new drugs and treatments, or by bringing about technological advancement. 

We hope you drew value out of our best practice guide and enjoyed reading it! Please don’t hesitate to ask us questions about any of the above, or to provide any feedback or comments you may have!



Was this article helpful?
0 out of 1 found this helpful