ESRA 2027 sessions by theme
Back to Overview of Sessions
Not Just Who Donates, but What We Can Learn: Inference from Donated Digital Trace Data |
| Coordinator 1 | Dr Barbara Binder (GESIS Leibniz Institute for the Social Sciences) |
| Coordinator 2 | Dr Sina Chen (GESIS Leibniz Institute for the Social Sciences) |
| Coordinator 3 | Ms Fabienne Krämer (GESIS Leibniz Institute for the Social Sciences) |
| Coordinator 4 | Dr Jessica Daikeler (GESIS Leibniz Institute for the Social Sciences) |
Session Details
Data donation, where participants retrieve personal data exports from digital platforms and contribute them to research, offers access to rich behavioral records that are becoming increasingly difficult to obtain through platform APIs and other researcher-access programs or methods. Bias in donated data is commonly assessed by comparing donors and non-donors regarding their sociodemographic characteristics, attitudes, and behaviors. While such compositional differences may indicate potential representation bias, they are not sufficient to assess how much particular estimates are biased in a given empirical setting. The consequences depend on how closely the characteristics driving donation participation relate to the constructs and relationships under study as captured by the donated data itself.
Moreover, conclusions based on donated data may be affected by validity concerns beyond donor self-selection. Personal data exports may capture platform activity incompletely, while the transformation of platform records into research variables may introduce problems of construct measurement. Donation procedures, data processing, and record linkage may create further sources of error.
Against this background, this session invites contributions examining sources of error throughout the data donation process and their consequences for the scientific claims that can be made from these data. Relevant topics include who participates in donor samples and who completes the donation, which parts of platform activity are included, omitted, misrepresented or even missing in personal data exports, and how platform export systems, donation procedures, data processing, record linkage, and construct measurement shape the resulting data. Contributions may consider how these processes affect (descriptive) estimates, relationships between variables, and the generalizability of findings to a target population. Contributions on methodological approaches for detecting, quantifying, and mitigating theses errors or contributions showcasing specific cases are also welcome.
The session asks: What can researchers validly infer from donated digital trace data, for which populations, and under what conditions?
