Trends in sociology research methods in the 21st century: Automated journal article coding using large language models
Evidence on the methodological evolution of sociology in the 21st century is limited. Useful signals exist in the literature, but we lack a temporally extended description of how the discipline’s practice has changed across the period in which the data and method environment was most transformed. The constraint is the cost of hand-coding journal articles, capping feasible corpus sizes well below what disciplinary description requires. We use locally hosted large language models (LLMs) to break this constraint, validating their performance before applying the pipeline to nearly 20,000 full-text articles from 28 sociology journals over 2004–2024. We document a steady empirical turn, near-parity between quantitative and qualitative work, and a wholesale shift of survey administration online. These patterns bear on Savage and Burrows’ diagnosis of a ‘coming crisis of empirical sociology’: contrary to their expectations, the survey has held its share and the interview has grown, while big data has remained marginal.
Authors
- Ian Brunton‐Smith (ORCID: https://orcid.org/0000-0002-0077-5602)
- Patrick Sturgis (ORCID: https://orcid.org/0000-0003-1180-3493)
Institutions
- University of Surrey (GB)
- London School of Economics and Political Science (GB)
Publication Details
- Journal
- Sociology
- Published
- 2026-11-20
- DOI
- https://doi.org/10.1177/00380385261484085
- Primary Topic
- Computational and Text Analysis Methods
- Type
- article
- Field-Weighted Citation Impact
- 0.00