From Hedging to High Confidence: The Evolution of Uncertainty Language in U.S. Intelligence Assessments, 1983-2021

This dataset underpins the article “From Hedging to High Confidence: The Evolution of Uncertainty Language in U.S. Intelligence Assessments, 1983–2021.” It contains systematically coded confidence phrases extracted from seven major U.S. intelligence documents spanning four decades. The dataset records each instance where analysts expressed uncertainty, confidence, or probability. Variables include: Temporal Period – Historical phase (e.g., Cold War, Post-Cold War, Reform Era). Document Title – Source document (e.g., Iraq’s Continuing Programs for WMD, 2002). Date – Year or release date of the document. Document Type – NIE, Commission Report, Senate Investigation, etc. Confidence Phrase – Extracted phrase expressing certainty/uncertainty. Full Sentence – Source sentence in which the phrase appears. Administration – U.S. presidential administration at the time. Confidence Level – Coded level (e.g., High, Medium, Low, None). Topic – Thematic domain (e.g., terrorism, proliferation, climate, military). Notes – Additional context about coding decisions or interpretation. File Format: Excel (.xlsx) Use: The dataset allows replication of the content analysis, comparison across documents, and further research on how uncertainty communication in intelligence evolved after critical events such as the Iraq WMD failure and post-9/11 reforms.