By Simon Hettrick, Deputy Director.
This is a story about reproducibility. It’s about the first study I conducted at the Institute, the difficulties I’ve faced in reproducing analysis that was originally conducted in Excel, and it’s testament to the power of a tweet that’s haunted me for three years.
The good news is that the results from my original analysis still stand, although I found a mistake in a subdivision of a question when conducting the new analysis. This miscalculation was caused by a missing “a” in a sheet containing 3763 cells. This is as good a reason as any for moving to a more reproducible platform.
2014: a survey odyssey
Back In 2014, I surveyed a group of UK universities to see how their researchers used software. We believed that an inexorable link existed between software and research, but we had yet to prove it. I designed the study, but I never intended to perform the analysis. This was a job better suited to someone who could write code, and I could not. Unfortunately, things didn’t go to plan and I found myself in the disquieting situation of having an imminent deadline and no one available to do the coding. Under these circumstances, few people have the fortitude to take some time out to learn how to…Continue Reading