How Research is Made Reproducible
The aim of reproducible data analysis is to document and communicate the analysis in such a way that other researchers can easily understand the procedure and know how to replicate the results. Ultimately, the following questions must be answered: How can I access the data? Where can I find the code? How do I successfully execute the code?
Even small steps can help in achieving this. They also benefit you directly, because you will find it easier to understand and extend your own analysis later on. In addition, many journals apply a data and code policy. For example, the journals of the American Economic Association (AEA) and the Royal Economic Society (RES) and the Econometric Society require a so-called “replication package” before final acceptance of an article, and a data editor checks its reproducibility. A replication package typically includes all necessary materials – most notably data and code – as well as information on the software and computational environment used, in order to enable others to understand the study and reproduce its results. Arguably the most important component of a replication package is a well-written README file. It documents the package and guides users through the reproduction process.
You can make your research reproducible by following these steps:
- Prepare a study plan or protocol in advance. Carry out a pre-registration or submit a Registered Report.
- Choose reproducible tools and materials. The use of open source tools can promote reproducibility.
- Set up a reproducible project: Centralise and organise your project management by using an online platform, a central repository or a folder for all research files. For example, you can store all your project files on GitHub or GitLab. Follow best practices by separating your data from your code in different folders. Make your raw data “read-only”.
- Organise your data, files and folders: Use file naming conventions, construct folder trees with a consistent, scalable structure, separate raw data from analysed data and so on.
- Learn the basics of version control, even if your own research does not require programming knowledge: The possibility to restore a specific version of a document that has been written over a period of several years can be very valuable.
- Automate certain recurring tasks: You will raise the reliability of your results and make writing scientific articles easier, because you are able to vary parameters more easily.
- Automate your data processing and analysis workflows: Write scripts to process your data and manage your work steps. For example, avoid using spreadsheet programs such as Excel for large datasets.
- Create a README file for your replication package: Document your data sources, describe your code and provide step-by-step instructions for reproducing your results. Start writing it from day one — not at the point of submission. Even if something seems obvious while you are working on it, it can be unclear two months later, even to yourself. When creating the README file, use established templates such as the one from the Social Science Data Editors, the World Bank's README Generator, or templatereadme.org.
- Combine code and documentation: Consider using Jupyter Notebooks, Quarto or R Markdown to integrate your code with your comments and documentation.
- Share and license your research: Make your data available via a repository. For software, notebooks and containers: license your code to inform others how it may be (re)used.
- Make your research transparent: Describe and publish your methods and procedures clearly, transparently and fully, in order to enable replication.

You can find additional and more detailed information on how to make your research reproducible here:
- Guide for Reproducible Research by the Turing Way
- Step-by-step-guide to achieving computational reproducibility using R and Markdown by Daniel Lakens
- The paper “A Realistic Guide to Making Data Available Alongside Code to Improve Reproducibility” by Nicholas Tierney and Karthik Ram offers a practical, feasible guide on how to share data alongside research
- “Practical Steps Towards Open and Reproducible Research” by Heidi Seibold (online seminar and factsheet by the Helmholtz Open Science Office, 2021)