Abstract

This thesis discusses the practical and metascientific merits of reproducibility of scientific workflows and results. I present repro, an R package which simplifies the adherence to the best practices of reproducibility proposed by Peikert & Brandmaier (2019). The resulting workflow complies with Open Science principles, is transferable across machines and preserved over time. Research projects are virtually guaranteed to reproduce and consequently are more comfortable to replicate. I explained how the technical solutions, namely Git, RMarkdown, Make, and Docker, overcome common problems of reproducibility and how the interaction with these tools is simplified through automation. The high degree of automation allows the workflow to scale well across researchers, letting them collaborate more transparently and effectively, as well as across machines, including high performance and cloud computing technology. To adapt to the ever-shifting landscape of software requirements repro has a modular design allowing other packages implementing research workflows to build upon its infrastructure.

References

Peikert, A., & Brandmaier, A. M. (2019). A Reproducible Data Analysis Workflow with R Markdown, Git, Make, and Docker [Preprint]. PsyArXiv. https://doi.org/10.31234/osf.io/8xzqy