Fabricated statistics invent or grossly distort numbers to make a claim feel proven. They exploit a well-documented bias toward numeracy-as-credibility: a specific figure reads as objective and researched, and the mere presence of a number tends to disarm the scrutiny a bare assertion would attract. “90% of experts agree,” “3x more effective,” “millions affected” — attached to nothing, these still shift belief because they wear the authority of measurement.
The mechanism blends this authority-of-numbers effect with anchoring: a figure introduced early sets a reference point that colors everything after it, even once doubt creeps in. Fabricated numbers are trivial to produce and, when emotionally charged, spread faster than any correction — a pattern Stephan Lewandowsky’s research on misinformation ties to how sticky first impressions are and how effortful it is to dislodge a false “fact” once it lands.
Because a statistic is only as trustworthy as its provenance, the defense is to route every consequential number back to its source. Demand the primary study and dataset, check the methodology behind the figure, sanity-test its magnitude against known base rates, and decline to repeat striking numbers that don’t check out. Genuine data withstands “according to whom, measuring what?” — a fabricated statistic collapses at the first request for its source.