A Relational Ontological Framework for Identifying Unstated Assumptions in Scientific Datasets

Every field of modern science relies on data as something foundational, the basic material that provides researchers with evidence of all recordable phenomena. Data carries an epistemic weight, linking various sensory experience as proof to our propositions, supporting scientific pursuit to get to the truth of something and justify scientific belief. However, to measure or count amounts to categorization, and the categories we create emerge from the models and cognitive schemes constructed in the minds of human beings. Our data is collected, measured, divided, and otherwise organized according to how we think about the world, and retains inherent unstated assumptions that are perpetuated in our scientific endeavors. Here, I propose a novel theoretical framework, Warrant-Domain Episteme, for explicating unstated assumptions in scientific data sets. This provides a basis for any future work on formalizing and operationalizing these unstated assumptions, which is especially significant as science progresses toward increased automation and use of machine learning for analysis and decision-based outcomes. Warrant and domain analysis are used as a means to frame what is meant by unstated assumptions, and case-in-point examples from published scientific literature are also provided as evidence for this claim.