Computational Reproducibility for Research: Why It Matters and How To Do It

Scientific research often relies on complex data wrangling, simulation, visualisation and statistical analyses. Our findings are summarised in tables and figures that are then (we hope) published in academic journals, nestled among paragraphs of explanatory text. These publications might be the most valued items on the academic market, but arguably, the most integral part of the whole process is the code that produced those tables and figures. And yet, successfully executing someone else’s code, and reproducing the findings from a scientific publication, is fraught with difficulty. Let’s be honest, even re-running your own code can be fraught with difficulty. While common and understandable, this lack of reproducibility raises serious questions about scientific rigour and what it means to make a scientific contribution. In this talk, we will discuss why reproducibility matters, and what we can do to make our own workflows more reproducible. Live demonstrations will be carried out in R, and we will introduce the CODECHECK framework (https://codecheck.org.uk) as one way of tackling these challenges.

Sam Langton is a Lecturer in Quantitative Criminology at the University of Manchester, with a background in open science, R, and reproducible research. He has previously worked as a postdoc at the NSCR and as a research software consultant at Amsterdam University Medical Center, where he co-authored a report on computational reproducibility at the organization.