Managing Subjectivity and University in Dataset Construction

Large-n observational datasets are essential tools in social science. They allow scholars to systematically study real-world phenomena, identify social patterns, and inform policy. When built attentively, datasets also have the potential to surface and challenge systems of oppression (D’Ignazio and Klein, 2020). However, datasets are not neutral representations of objective facts. They are abstractions—simplified interpretations of complex, often contested realities. In the process of creating datasets, researchers reduce lived human experiences to typologies that permit comparison and generalization. This process inevitably relies on human judgment, interpretation, and cultural framing (Orr and Crawford, 2024). Thus, creating observational datasets always involves “thick description”—extracting and interpretating meaning from observations of complex social behavior (Geertz 1973).