Datenwahl

Datenwahl (literally data choice) is a concept in data governance and information ethics referring to the process by which individuals or organizations select data sources, data sets, or features to be used in analysis, modeling, policy-making, or reporting. It encompasses decisions about what data to collect, which variables to include, and which data sources to rely on. The term is used especially in discussions of bias, transparency, and reproducibility in data-driven work, where different selections can lead to different conclusions.

In practice, Datenwahl occurs at multiple levels: data sourcing and collection strategies, feature or variable selection in machine learning, and the curation of data used for dashboards or official statistics. It often involves trade-offs among data quality, coverage, timeliness, and cost. Proponents emphasize the efficiency and clarity that deliberate data choice can bring to complex analyses. Critics warn that selective data choice may introduce or mask biases, particularly if criteria are opaque or driven by desired outcomes.

Good governance of Datenwahl includes documenting selection criteria, maintaining data provenance, and conducting sensitivity analyses. Techniques include randomized sampling, stratified sampling, and cross-validation to assess how alternative data selections affect results. In journalism and policy, a transparent account of Datenwahl is considered essential to credibility.

See also data governance, data provenance, data quality, and selection bias.