Implications of constructing a pre-analysis plan in Open-Source GIS Research
Pre-planning an analysis in the manner of our Gerrymandering Case Study (which utilizes a detailed research compendium) can be beneficial to researchers, as it forces them to more closely understand their data sources before analysis, and to be more mindful in how they approach working with the data in order to avoid potential sources of error. It makes public the thought processes that researchers may entertain on their way to constructing their code and analysis. A pre-analysis plan can also show the perspectives and information that researchers had prior to conducting research, as it could inform the outcomes of the research. Finally, this approach also helps to front-load work documenting metadata and organization/explanation of code that the researchers would need to complete later before publishing their research. Pre-planning an analysis ensures that subsequent code can be more legible to other researchers and more intentional in general. It is for all of these reasons that this aspect of a research compendium in reproducible, open-source GIS research is an important premilinary step for researchers.
Though I found the process of making the pre-analysis plan for this case study to be a good introductory exercise in using both an analysis plan and an open science research compendium in general, after using pre-analysis plans in later projects in this course, I feel that my plan for this case study was not very detailed. This would not be too big of an issue if I had done more version releases of the pre-analysis plan before implimenting, but ideally I would have included more information than just the metadata and some basic transformations. As this was my first pre-analysis plan, I still found it more difficult to fully think through my workflow before actually digging into data and working on procedure code. Another one of the challenges of this case study was that it was an original study, and as a result, there weren’t transformation and analysis steps from an original paper to interpret. In this way, it didn’t prepare me as much for the more involved process of interpreting the workflow of a study into a pre-analysis for a reproduction (as with the next two projects). Overall, however, it was a useful exercise for at least introducing these tools and providing the foundation for my work in this course.