Translating Computational Methods into (and out of) Sociology: How Research Support Units Negotiate Values, Responsibilities, and Standards for Societal Analysis

Digital methods, Big Data, and AI increasingly shape societal analysis in two directions: sociologists adopt computational approaches developed in STEM fields, and STEM-trained researchers work with digital traces, large-scale data, and AI tools in projects addressing sociological questions. In both cases, methods travel between disciplines with implicit assumptions about what counts as evidence, acceptable risk, and “good” research practice. This talk asks: How can research support units negotiate values, responsibilities, and methodological standards to enable social-scientific use of computational methods across disciplinary boundaries? From a socio-technical perspective (Bostrom & Heinen, 1977), we propose a bidirectional translation framework from the viewpoint of the data competence center DataNord’s help-desk, hosted at the Data Science Center, University of Bremen (Steinmann et al., 2025). The framework maps five friction points that require careful specification: (1) governance and accountability: specifying GDPR-aligned data handling, third-party tool/API risks, and approval responsibilities (2) construct grounding: mapping labels, proxies, and features onto theoretical constructs (3) prediction vs. explanation: separating predictive performance from explanatory or causal claims (4) quality assessment: evaluating validity, reliability, representativeness, and bias (5) reproducibility: ensuring provenance and documentation for data pipelines, code, parameters, and model/tool drift We illustrate each friction point with practice-based examples and discuss how they are addressed through targeted training, individual consultations, and networking formats. The talk offers a practical roadmap of questions to ask and decisions to document as computational methods move between STEM and sociology, and clarifies where research support can intervene to strengthen accountability and methodological credibility. References Bostrom, R. P., & Heinen, J. S. (1977). MIS problems and failures: A socio-technical perspective. Part I: The causes. MIS Quarterly, 1(3), 17–33. https://doi.org/10.2307/248710 Steinmann, L., Hörner, T., Bohnebeck, U., Drechsler, R., Glöckner, F. O., & Pigeot, I. (2025). DataNord: Empowering Data Literacy - Strengthening Bremen's Research. E-Science-Tage 2025, Heidelberg, Germany. Zenodo. https://doi.org/10.5281/zenodo.15118351