startthat.work home

TC1-15 Bayesian updating (likelihood reasoning, likelihood ratio, Bayes factor)

Revises the probability of a hypothesis in proportion to how much more likely the evidence is under it than under alternatives and produces a posterior probability or odds.

Trace cause and test claimsEvidence: UnconfirmedPrimary technique, runs in step 5.3Strongest statement: predict, without claiming a cause

What it does

Revises the probability of a hypothesis in proportion to how much more likely the evidence is under it than under the alternatives. Artefact: a posterior probability or odds; read the likelihood ratio as the strength of the evidence.

When to use it

Use when you hold a prior view on a hypothesis and need to say how far a piece of evidence should move it.

When it does not fit

The prior or the likelihoods are guesses presented as precise numbers.

Result format

The result gives a prior probability for the hypothesis, a likelihood of each evidence item under the hypothesis and under the alternatives, and the likelihood ratio for each item. It ends with a posterior probability or odds, with the likelihood ratio read as the strength of the evidence.

Requirements on the material

PropertyRequirement typeRequirement
Content typeAlways appliesrequires=text; requires=tabular
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 materialA prior probability and the probability of each evidence item under each hypothesis.
Additional sources needednone
Not for (rules)none
Not for (text)Wrong choice when the prior or the likelihoods are guesses presented as precise numbers. Treating dependent evidence items as independent overstates the posterior.

Choosing and ordering

Kind of techniquePrimary technique, runs in step 5.3
Phase taganalyse
Other phase tagsconclude
Kind of findingcause; option ranking
Question answeredhow much; why
Strongest statement it supportspredict, without claiming a cause
Needs as input
Makes as output
Must exist firstNone
Loop targetproceed with limit stated

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-informingsensitivity analysisnoneno
decision-informingcode execution for arithmeticnoneno
decision-informingalternative explanations checknoneno
high-stakesassumption and limitation statementnoneno
high-stakessensitivity analysisnoneno
high-stakescode execution for arithmeticnoneno
high-stakesalternative explanations checknoneno
high-stakessecond technique with different assumptionsnoneconditional
high-stakesexpert reviewneededno
high-stakesindependent reviewneededno

Failure modes and limits

Model suitabilityhuman judgement required
Typical model failureStates posterior probabilities to false precision from invented priors and likelihoods, or multiplies likelihoods of dependent evidence.
Error sourcesinput data; expert opinion; statistical assumptions
Certainty limitsCertainty is limited by the quality of the prior and the likelihoods, which are usually judgements. It drops further when evidence items are dependent or when the hypothesis set is incomplete (imprecision).

Evidence and tag basis

Unconfirmed. No source was opened in this research.

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