The next speaker at the ECREA 2026 conference in Brno is Milin Zhang, whose focus is also on digital media use; it is critical to measure such use accurately, but self-reporting is often a dubious measure and correlates poorly with direct tracking data. This is true for any kind of media use, but political media use is of particular interest and relevance in our field. If both measurement sources have their problems, then, it is difficult to know which to trust, and the empirical consequences of such choices also remain unclear.
This project, then, explores how well these two sources correlate, and how data source choices affect empirical interpretation. It works with a sample of nearly 700 German residents, who were repeatedly surveyed for self-reports of their daily average digital media use, while their Web browsing data were also collected. The top 5,000 domains visited by those users were manually coded, with news and politics domains specifically highlighted; the specific content of these sites was then also coded by LLM.
These patterns were then correlated with the self-reports from the surveys. It turned out that for mainstream political information, survey and tracking data were strongly correlated; for alternative media, correlation is much weaker, and this is even worse for other media forms. Correlations with trust in media also varied across these categories and sources.
There remains a substantial measurement mismatch, then – and this is true for survey as well as tracking data.
