Note: A Sociomaterial Perspective on Trace Data Collection: Strategies for Democratizing and Limiting Bias

Abstract

Researchers heavily use online platforms for collecting trace data, i.e., data capturing user interaction on and with sociotechnical systems. Human-computer interaction scholars have highlighted the role of reflexivity while analyzing such data in the case of marginalized communities. Drawing on sociomaterial perspectives, we highlight how data collection approaches involving lists of search phrases and APIs can embed researchers’ positionality, perspectives, and biases within the datasets. In this note, we reflect on the data collection approaches of two papers that studied the sociohistorically marginalized Bengali communities on the question-and-answer site Bengali Quora. We illustrate how recommendation systems and data labeling workers can be included in the data collection process to democratize and limit bias while broadening and contextualizing the trace datasets for research.