Framework and Infrastructure for Reproducible Research and Pedagogy in HEGS

Keywords: Reproducibility, Replicability, Research Infrastructure, Pedagogy
Abstract Type: Paper Abstract

Abstract

Scientific reproducibility is constrained by a lack of infrastructure, including a framework for conducting reproduction studies and pedagogy for teaching reproducibility (NASEM 2019 Consensus Report on Reproducibility and Replicability in Science). We are therefore conducting and publishing reproduction and replication studies and teaching reproducible research practices in advanced geography methods courses, and developing a framework and infrastructure for teaching and implementing geographic reproduction studies.

Our framework is organized into a practical and pedagogical four-stage iterative workflow.
First, we thoroughly review a published study's hypothesis, data, and methodology. Second, we write a pre-analysis plan for reproducing or replicating the study. Third, we implement the plan using open-source software, while comparing intermediary results, documenting errors or uncertainties, and adjusting the plan. Finally, we report and interpret results, and discuss any uncertainties in terms of scientific conceptualization, measurement, analysis, and communication.

Our infrastructure is a template research compendium-- a collection of all the data, metadata, code or procedures, and documentation required to communicate the full details of a research project. Our compendium template is a Git repository to facilitate provenance, collaboration, and archiving. Template pre-analysis plans and post-analysis reports facilitate transparent open science practices. The directory structure organizes data, metadata, code or other procedures, results, reports, and manuscripts, and facilitates management of confidential or proprietary data.

Our framework and template provide structure for teaching and practicing reproducible research, facilitating communication within and between research teams, and enhancing open science and reproducibility in geographic research.