Hosts slides and materials for a talk on "Collaborating on Reproducible Code Using R and GitHub" hold by Alexander Enge and Kirsten Stark at the research colloquium of the neurocognitive and biological psychology labs, Humboldt-Universität zu Berlin.
ABSTRACT: "Collaborating on Reproducible Code Using R and GitHub" (by Kirsten Stark & Alexander Enge)
Reproducing results in the mind and brain sciences remains challenging, which can at least partially be attributed to the complexity and heterogeneity of data analysis pipelines (Botvinik-Nezer et al., Nature, 2020). The shift away from point-and-click software to scripted languages such as R (R Core Team, 2020) has been an important first step towards better computational reproducibility. However, none of us is explicitly taught how to produce code in such a way that others can access, understand, and ideally reproduce our analysis. Luckily, a number of simple tools to facilitate this process can be borrowed from professional software development. These include version control (to keep one's code organised and its history accessible), package management (to prevent updates to software packages from breaking the code or changing its results), and containerization (to guarantee long-term computational reproducibility). These tools can be combined with other, more research-specific ones (such as dynamic report generation) within the RStudio environment. In this talk, we will provide an overview over some of these tools which we have found most helpful for making our R code more collaborative, accessible, and reproducible.