Blog

Reproducibility can demonstrate that research results are not due to bias or chance, which is vital for accurate and reliable results. It refers to the degree to which other researchers can achieve the same results using the same dataset and analysis as the original research. Research is reproducible when other researchers can achieve the results again with high reliability. 

In this blog, we look at the differences between reproducibility and repeatability, why it is important and how you can make your work reproducible.

Reproducibility, repeatability and replicability

Reproducibility, repeatability, and replicability are related terms that are often used interchangeably. While various research communities have embraced these concepts, the terminology between disciplines differs.

At Gates Open Research, we define reproducible research as research that can be executed by different researchers using the same data to achieve the same results. This definition aligns with the Claerbout and Karrenbach definition first outlined in 1992. One study found the Claerbout and Karrebach definition is the most used across disciplines. The definitions are as follows:

• Repeatable: Research is repeatable when the original researchers perform the same analysis on the same dataset and consistently produce the same findings.
• Reproducible: Research is reproducible when other researchers perform the same analysis on the same dataset and consistently produce the same findings. 
• Replicable: Research is replicable when other researchers perform new analyses on a new dataset and consistently produce the same findings. 

Why is reproducibility important to Open Research?

Science is dependent on the sharing of information. The growth of open research and open data has led to increased calls for researchers to improve the reproducibility of their research. Sharing data, code, and detailed research methods accelerate scientific discovery by making more elements of research available to all researchers. Aside from faster progress, reproducibility offers numerous other benefits for the research community, including:

• Reproducible research strengthens scientific evidence and the reliability of results: When researchers can reproduce a study, they lend credibility to the findings of the original researchers. The results of the original research can be deemed reliable, and there is more evidence supporting the conclusions. This makes it more likely that research findings will be used to make a real-world impact by informing policy or practice.
• Reproducible research increases trust in science: When scientists cannot reproduce results, other researchers and the public lose trust in the scientific process. With so much public funding going towards research and so much research informing public policy, research must be reproducible so the public can trust science.   
• Reproducible research enables efficiency in research: How reproducibility increases efficiency is twofold. Firstly, the more reproducible research is, the greater the odds that the research, or at least parts of it, can be reused by other researchers in the future. Additionally, by publishing negative results, researchers can help other researchers avoid wasting time on analyses that will not return the expected results.  
• Reproducible research helps to minimize misinformation: Reviewers can spot mistakes more easily when researchers embrace transparency in the research process, provide detailed documentation of their methods and analyses, and share their research materials. This helps to avoid misinformation that can limit the replicability of research and, instead, leads to more accurate research papers. 

Benefits of reproducibility for authors

Making your work reproducible also offers additional benefits to you as an author. These include:

• Reproducible research has the potential for a greater impact: Researchers who produce reproducible research share their underlying data and methods. Sharing detailed research data is associated with higher citation rates as other researchers use and credit the data in their projects. 
• Reproducible research facilitates in-depth peer review: Reproducible research can result in higher-quality, faster peer review as reviewers have access to the data and analytical processes described in the manuscript. This increases the probability that errors are caught during the peer review process and reduces back-and-forth between authors and reviewers.  
• Reproducible research enables iterative science: Science is an iterative process, with many tasks repeated time and again. Reproducible research enables researchers to reuse previous research materials to execute similar research tasks more efficiently in new projects.  
• Reproducible research enables collaboration and reuse: Reproducible research opens the door to new partnerships to develop existing projects further. It also leads to greater reuse of research. Designing reproducible workflows and sharing all research outputs openly allows others to develop a deeper understanding of the work and build upon it. 

Learn more about how our Open Research guidelines help make your work reproducible and submit your research to Gates Open Research via VeriXiv