What comes after data collection? Processing, enriching, and analyzing different types of digital traces

Digital behavior is increasingly being measured with traces obtained through web tracking and data donation approaches. However, existing research has primarily focused on the collection process of traces, leaving open the question of how (raw) traces can be subsequently transformed into meaningful measurements of digital behavior. Going beyond the data collection stage, this talk is therefore guided by the following overarching research question: How can traces be processed, enriched, and analyzed? To address this research question, I present findings from three case studies that investigate different types of individual-level traces: (1) Website visits, search (engine) terms, and app usage obtained through tracking applications (measuring parental digital behavior), (2) viewing logs donated by Netflix users (measuring streaming behavior), and (3) chat logs donated by ChatGPT users (measuring AI-assisted knowledge production). Drawing on these case studies, I demonstrate how traces can be enriched with contextual information, classified using open-weight Large Language Models (LLMs), and aggregated into meaningful measurements of digital behavior. In doing so, I put special emphasis on critically discussing potential error sources, reflecting upon their epistemological implications, and deriving empirically informed strategies and best practices. Overall, the findings can build an important knowledge basis for future research employing digital trace data to measure digital behavior and social phenomena.

MaRCS is a seminar series jointly organized by the Mannheim Centre for European Social Research (MZES), the University of Mannheim School of Social Sciences, and GESIS – Leibniz Institute for the Social Sciences.