Description
Evaluates evidence against all plausible hypotheses in a matrix, focusing on disconfirmation, to counter confirmation bias in explaining what is happening.
Analysts, like researchers, tend to pick a favoured explanation and collect supporting evidence. ACH lists all reasonable hypotheses, lists significant evidence and arguments, and marks in a matrix whether each item is consistent, inconsistent or not applicable to each hypothesis. The key insight is that evidence consistent with all hypotheses has no diagnostic value; hypotheses are ranked by how much evidence is inconsistent with them, not consistent. Analysts then assess sensitivity to critical evidence, identify missing evidence that would discriminate, and report conclusions with milestones for re-evaluation. It is a rigorous structure for UX synthesis questions like 'why did activation drop?'.
Attributes
- Discipline
- Research Design & Rigor
- Loss type
- Unmigrated
- Stage
- Definition, Strategy
- Data
- Analytical / expert
- Effort
- Hours
Steps
- Brainstorm all plausible hypotheses with a diverse group, including unpopular ones.
- List evidence and assumptions: data, findings, logs, absence of expected evidence.
- Build the matrix and rate each item's consistency with each hypothesis; note diagnosticity.
- Refine the matrix: delete non-diagnostic evidence, add hypotheses if needed.
- Rank hypotheses by inconsistent evidence; test sensitivity to a few critical items and their reliability.
- Report conclusions with relative likelihoods and define indicators that would change the judgement.
Sources
- Heuer, Psychology of Intelligence Analysis (1999), CIA Center for the Study of Intelligence
- Heuer & Pherson, Structured Analytic Techniques for Intelligence Analysis (2nd ed., 2014)
Also known as
- ACH
- Heuer's ACH
- Hypothesis matrix
Categories
- Research Operations
- Research Design & Rigor