startthat.work home

TC1-13 Analysis of competing hypotheses (ACH)

Lists plausible hypotheses, rates each evidence item against each, and produces a hypothesis by evidence matrix aimed at disproving hypotheses.

Trace cause and test claimsEvidence: TestedPrimary technique, runs in step 5.3Strongest statement: show a link between things

What it does

Lists all plausible hypotheses and rates each evidence item against each, aiming to disprove hypotheses. Artefact: a hypothesis by evidence matrix; read the evidence that discriminates between hypotheses and the hypothesis with the fewest inconsistencies.

When to use it

Use when several explanations are plausible and you need to see which evidence actually discriminates between them.

When it does not fit

The hypothesis list is incomplete, analysts skip steps, or non-diagnostic evidence shifts the result.

Result format

The result lists the hypotheses and gives a matrix of evidence items against hypotheses with a consistency rating in each cell. It names the evidence items that discriminate between hypotheses and the hypothesis with the fewest inconsistencies.

Requirements on the material

PropertyRequirement typeRequirement
Content typeAlways appliesrequires=text
Unit of analysisAlways appliesrequires=claim
Appraises source credibilityApplies in some casesno

Not restricted on: Structure, Variable types, Time structure, How the sample was chosen, Level of detail available, Number of sources, Whether sources can be joined, Amount of material, Known data problems, Producer interest, Source dependence, Origin, Document genre, Language, Personal information, Security classification, Licence.

Minimum materialAt least two rival hypotheses and several evidence items rated against each.
Additional sources needednone
Not for (rules)none
Not for (text)Wrong choice when the hypothesis list is incomplete, analysts skip steps, or non-diagnostic evidence shifts the result. Experiments found no accuracy benefit and some harm to coherence.

Choosing and ordering

Kind of techniquePrimary technique, runs in step 5.3
Phase taganalyse
Other phase tagsvalidate
Kind of findingcontradiction; cause; difference
Question answeredwhy; what
Strongest statement it supportsshow a link between things
Needs as input
Makes as output
Must exist firstNone
Loop targetreturn to frame

Library checks by stakes

Each stakes level includes the checks of the level below it. Checks that need a person apply at high stakes only, so the process leaves them out at lower stakes (decision 38). Repeatability: judgement dependent.

StakesCheckPerson involvedCounts as a different method
exploratoryassumption and limitation statementnoneno
decision-informingassumption and limitation statementnoneno
decision-informingalternative explanations checknoneno
decision-informingsource-span citationnoneno
high-stakesassumption and limitation statementnoneno
high-stakesalternative explanations checknoneno
high-stakessource-span citationnoneno
high-stakessecond technique with different assumptionsnoneconditional
high-stakesexpert reviewneededno
high-stakesindependent reviewneededno

Failure modes and limits

Model suitabilityhuman judgement required
Typical model failureRates most evidence items as consistent with every hypothesis so that nothing discriminates, and omits the hypothesis the analyst finds unlikely.
Error sourcesinput data; source selection; coder or rater judgement
Certainty limitsCertainty drops when the hypothesis list is incomplete, when evidence is not diagnostic, or when all ratings come from one analyst (risk of bias and inconsistency).

Evidence and tag basis

Tested. Tested in experiments with no accuracy benefit and some signs of harm to coherence; see the evidence section. Heuer concedes it will not always yield the right answer. (Dhami, Belton and Mandel (2019), Strathprints record; Karvetski and Mandel (2020), Cambridge Core, does ACH help coherence; Cambridge Core paper on what works and what fails in improving analysts' accuracy; Heuer, Psychology of Intelligence Analysis)

Tag basis: input profile inference; selection inference; sequencing inference; robustness inference. All tag values are inference unless a source is named.