🧪 This platform is in early beta. Features may change and you might encounter bugs. We appreciate your patience!
source_selection
Source Selection Bias occurs when a speaker or media outlet systematically chooses sources that confirm a predetermined narrative while excluding equally credible sources that offer contradicting evidence or perspectives. Unlike cherry picking (which selects data), source selection bias operates at the level of who gets to speak — which experts are quoted, which studies are cited, which witnesses are interviewed. The result is an artificially one-sided picture that appears well-sourced.
A news report on climate policy interviews three industry-funded economists but no climate scientists or environmental economists.
A documentary about nutrition exclusively features advocates of one dietary philosophy while ignoring the broader scientific consensus.
A political talk show consistently invites guests from one end of the political spectrum, creating the impression that expert opinion uniformly supports one party's position.
Binary (yes/no) questions an LLM must answer to identify this aspect:
Are the cited sources predominantly one-sided, consistently supporting a single perspective?
Type: binaryAre credible sources with opposing viewpoints or contradicting evidence excluded or underrepresented?
Type: binaryDoes the selection of sources appear designed to confirm a pre-existing narrative rather than to present a balanced picture?
Type: binarySource Selection Bias occurs when a speaker or media outlet systematically chooses sources that confirm a predetermined narrative while excluding equally credible sources that offer contradicting evidence or perspectives. Unlike cherry picking (which selects data), source selection bias operates at the level of who gets to speak — which experts are quoted, which studies are cited, which witnesses are interviewed. The result is an artificially one-sided picture that appears well-sourced.
The presence of named sources and citations creates an appearance of journalistic rigor. Audiences assume that if experts were consulted, the reporting must be balanced. They rarely check whether equally qualified experts with opposing views were deliberately excluded.
Examine the source pool: 'Who was consulted and who was left out? Are all relevant perspectives represented? Do the sources have conflicts of interest?' Look for diversity in expertise, institutional affiliation, and viewpoint. Compare coverage across multiple outlets.
Common in partisan media where the same ideologically aligned commentators are repeatedly featured, in pharmaceutical reporting that only quotes company-funded researchers, in political coverage that only interviews one party's representatives, and in tech journalism that relies solely on industry insiders.
Use these tools to detect, analyze, or train this aspect.