Document Type
Paper
Start Date
6-10-2023 1:30 PM
End Date
6-10-2023 2:30 PM
Abstract
The I-GUIDE cyberinfrastructure project for convergence science is a leading example of the possibilities the geospatial data revolution holds for scientific discovery. However, rapidly expanding access to increasingly complex data sources and methods of computational analysis also presents a challenge to the research community. With more data and more potential analyses, researchers face the possibility of jeopardizing the inferential power of convergence research with selection bias. Well-designed infrastructure that can flexibly guide researchers as they record and track decisions in their research designs opens a path to mitigating this problem, while also expanding the reproducibility and replicability of research. Much of the infrastructure needed for convergence research can be borrowed and adapted from other disciplines, but geographic convergence research confronts at least five novel challenges. These are the need for geographically-explicit project metadata, managing diverse and complex data inputs, handling restricted data, specifying and reproducing computational environments, and disclosing researcher decisions and threats to validity that are unique to geographic research. We introduce a template research compendium and analysis plan for study preregistration to address these novel challenges.
DOI
10.5703/1288284317675
Decentralized Infrastructure for Reproducible and Replicable Geographical Science
The I-GUIDE cyberinfrastructure project for convergence science is a leading example of the possibilities the geospatial data revolution holds for scientific discovery. However, rapidly expanding access to increasingly complex data sources and methods of computational analysis also presents a challenge to the research community. With more data and more potential analyses, researchers face the possibility of jeopardizing the inferential power of convergence research with selection bias. Well-designed infrastructure that can flexibly guide researchers as they record and track decisions in their research designs opens a path to mitigating this problem, while also expanding the reproducibility and replicability of research. Much of the infrastructure needed for convergence research can be borrowed and adapted from other disciplines, but geographic convergence research confronts at least five novel challenges. These are the need for geographically-explicit project metadata, managing diverse and complex data inputs, handling restricted data, specifying and reproducing computational environments, and disclosing researcher decisions and threats to validity that are unique to geographic research. We introduce a template research compendium and analysis plan for study preregistration to address these novel challenges.
