Content
In this seminar, students work with digital behavioral data (DBD). DBD refers to digital traces of human behavior that are knowingly or unknowingly left in online environments such as social media, messengers, entertainment media, digital collaboration tools, or conversational AI systems. These rich data are increasingly available to social scientific research in the public interest but can also be used to derive strategic insights for business decisions. Students learn to work with DBD alongside the empirical research process: assessing and accessing available data, preprocessing it, developing and validating measures, analyzing the data, and reporting the results reproducibly. All practical work is carried out in R and learning resources are provided. Students already proficient in Python may work in that language instead.
Each semester, the course zooms in on one or a few specific forms of DBD, such as data from API access/webscraping, mobile logging, webtracking, platform data donations, mock-up digital environments, or interaction logs from conversational systems. This will be complemented with input on the requirements for data protection, research ethics, and data quality that each form of DBD entails. Correspondingly, the module covers key computational methods for processing and analyzing large behavioral datasets across text, image, and behavioral log data. Foci might be the construction and validation of measures from unstructured data, manual and model-assisted annotation, classification, or the analysis of behavior over time.
By completing this course, participants gain an up-to-date overview and practical insight into how the potential of observational data traces can be harnessed to better understand media users’ behavior in digital environments.
Program in Winter term 2026/2027
This semester, the focus is on conversations between people and generative AI systems as a form of digital behavioral data. Students examine how such interactions can be segmented into communication episodes and analyzed from a Communication Science perspective. We will work with a shared corpus of human–AI conversations, possibly linked to participant survey data. In small groups, students develop their own research question on this corpus, present intermediate results in the seminar, and communicate their findings in a written assignment (project report).
Recommended prerequisites
- Interest in social scientific perspectives on media, communication, and digital technologies.
- Basic knowledge of working with statistical software, specifically R or Python, is required.
- Students are recommended, but not required, to first visit the introductory module 57125 Data Science: Foundations, Tools, Applications in Socio-Economics and Marketing, which is offered in the summer term.
Learning Objectives
Students will…
- overview and understand central opportunities of DBD and the accompanying challenges for data collection and preprocessing
- evaluate the strengths and weaknesses of different ways of collecting DBD
- understand central requirements for data protection, research ethics, and data quality
- know and overview key computational social science methods for analyzing DBD
- develop and validate measures from unstructured behavioral data
- practice and apply knowledge on DBD, statistics, and data analysis in a group project of their own, and communicate the results in reproducible form
- evaluate complex subject-specific problems and solutions presented by other course participants and provide appropriate feedback