Reproducibility

While reproducibility is a broad concept in academic research, we will define reproducibility in terms of one overarching goal: a researcher who isn't you should be able to derive your exact results independently. Specific instances of good reproducibility in statistical computing include:

• Sharing code used to produce results and figures as they appear in published work
• Providing clear instructions for obtaining or generating data used in the analysis
• Specifying software versions and external dependencies (e.g. Python and R packages) used in the analysis

A related concept is helping other researchers build upon your existing analysis, which goes above and beyond reproducibility. This is discussed in XXXX

Tools for Facilitating Reproducibility

The department recommends using these pre-existing tools below in your research projects for purposes of reproducibility. These tools have seen widespread adoption in academic research, and resources for learning how to use these tools are readily available.

Git and GitHub

Git is a command line tool for version tracking and collaboration in shared codebases. It is a tool that is installed locally on your machine or department servers.