SourceBias

SourceBias refers to systematic distortions in information that arise from the properties of the sources used to generate that information. It describes how the selection, availability, presentation, or credibility of sources can influence conclusions, assessments, or narratives even when the underlying data are sound.

Causes include selection bias in data collection, publication bias that favors certain results, editorial or corporate influence, algorithmic amplification on platforms, language or access barriers, and funding or reputational incentives that favor particular outlets.

Types include data-source bias (overrepresentation of certain groups in datasets), publication bias (preferential reporting of positive or novel results), platform bias (algorithmic ranking and recommender effects shaping exposure), authority bias (overweighting statements from highly reputed sources), and geographic or linguistic bias (tendency to rely on sources from dominant regions or languages).

Impacts can include distorted risk assessments, skewed policy decisions, misperceptions in public discourse, reduced reproducibility in research, and reinforced stereotypes.

Examples include a news cycle that depends chiefly on official press releases; a scientific field where industry-funded studies dominate meta-analyses; and an AI model trained mainly on English-language content from a small subset of publishers.

Mitigation strategies involve diversifying sources, improving provenance documentation, applying transparent credibility criteria, triangulating information across independent sources, preregistration and replication for research, and promoting open data and audits.

See also: Bias, Selection bias, Publication bias, Information bias, Media bias.