Likelihood-Ratio Attribution Report¶
Evidence report — instantiates Intrinsic Signature Provenance
Turns a signature comparison into a calibrated likelihood ratio — how much more the evidence favors one origin than a stated alternative — with scope and limits attached.
A match is not an answer until someone states what it is evidence of, and how strongly. A Likelihood-Ratio Attribution Report first frames the specific provenance question, then expresses the strength of the signature evidence as a likelihood ratio: how much more probable the observations are if the sample came from the proposed source than if it came from a stated alternative. Its defining idea is that it reports evidential strength between two hypotheses — not a bare "match / no match," and not a probability that the source is guilty or genuine — and it binds that strength to an explicit scope and to the limits of what any signature can prove without a custody chain. It is the reporting layer that keeps a measurement from being over-read.
Example¶
An anonymous recording carries a threat, and a suspect's voice is available for comparison. Rather than declaring "it's him," the report defines two hypotheses — the recording's speaker is the suspect versus is an unrelated speaker from the relevant population — and states how much more probable the measured acoustic features are under the first than the second, say on the order of hundreds of times, within a defined population of comparable speakers and recording conditions. It states the assumptions that number depends on, and it stops short of asserting the probability that the suspect is the speaker, leaving that to the court once other evidence is weighed. The report communicates evidential weight, not a verdict.
How it works¶
- Fix the two hypotheses. State the proposed source and a specific, realistic alternative; the likelihood ratio has no meaning without both.
- Define the relevant population. Specify the reference class the alternative is drawn from and the conditions of comparison.
- Compute the ratio. Combine the comparison score with reference distributions under each hypothesis to get how much more expected the data are under one than the other.
- State scope and limits. Report population, conditions, and assumptions as first-class content, and bound the claim to what the signature supports — not to custody or possession.
The distinctive discipline is refusing to state a posterior probability of source: the report gives evidential strength; the decision-maker supplies the priors.
Tuning parameters¶
- Alternative hypothesis — which competing origin the ratio is computed against; a vaguer or wronger alternative silently inflates the number.
- Relevant population — how broadly the reference class is drawn, which sets what "an unrelated source" means.
- Verbal vs numeric scale — whether the strength is reported as a number or a calibrated verbal band; numbers are precise but easily misread.
- Assumption conservatism — how cautiously the reference data are used before the ratio is trusted.
When it helps, and when it misleads¶
Its strength is separating the strength of the evidence (the analyst's job) from the probability of the proposition (the decision-maker's job), which is exactly what keeps signature evidence from being overclaimed in a courtroom. This two-hypothesis, likelihood-ratio framing is the backbone of modern evaluative reporting guidance in forensic science.
Its failure mode and classic misuse are the same error read two ways: the prosecutor's fallacy — treating "the evidence is a thousand times more probable if same-source" as "it is a thousand times more probable they are the source" — which conflates the likelihood with the posterior and ignores the base rate.[n1] A poorly chosen or absent alternative hypothesis produces the same inflation more quietly. The guarding discipline is to state both hypotheses and the population explicitly, report a likelihood ratio rather than a probability of source, and flag the custody-independence limit so no one reads signature strength as proof of possession.
How it implements the components¶
provenance_question_boundary— it opens by pinning the exact question and its competing alternative, bounding precisely what is being attributed.attribution_comparison_rule— the likelihood ratio is the calibrated rule that maps a comparison score to evidential strength.uncertainty_and_scope_statement— it states population, conditions, and assumptions as primary content rather than a caveat.custody_independence_claim_limit— it explicitly bounds the claim to what the signature supports, leaving possession and custody to other evidence.
It does not implement signature_extraction_protocol or provenance_bearing_property — it neither measures nor reads the signature; that work is done by the laboratory and measurement mechanisms such as Isotopic Fingerprint Analysis and Spectral Signature Matching. Its nearest sibling is the Blind Proficiency Test: this report uses known error rates to state a calibrated ratio for one case, whereas that test measures those error rates with blind knowns.
Related¶
- Instantiates: Intrinsic Signature Provenance — it converts a signature comparison into disciplined, scoped evidential weight.
- Consumes: the measurement mechanisms (e.g., Isotopic Fingerprint Analysis, Manufacturing Toolmark Analysis) for the comparison score, and the Blind Proficiency Test for the error rates it calibrates against.
- Sibling mechanisms: Blind Proficiency Test · Chemical Taggant Program · Digital Watermark or Content Fingerprint · DNA or Biological Barcode · Isotopic Fingerprint Analysis · Manufacturing Toolmark Analysis · Reference Library Match · Spectral Signature Matching · Trace-Element Profile Matching
Editorial Notes¶
Form Classification¶
Form family: Analysis, Modeling & Optimization
Rationale: Likelihood-Ratio Attribution Report operates as a computation, comparison, model, or analytic representation used to infer, estimate, or choose because it turns a signature comparison into a calibrated likelihood ratio — how much more the evidence favors one origin than a stated alternative — with scope and limits attached.
Independent corroboration: The frozen evidence defines Likelihood-Ratio Attribution Report as 'Turns a signature comparison into a calibrated likelihood ratio — how much more the evidence favors one origin than a stated alternative — with scope and limits attached', so its operative form is Analysis, Modeling & Optimization.
Nearest alternative: Representation, Specification & Plan — A report carries the output, but the operative work is the likelihood calculation and attribution rather than the document itself.
Review outcome: Independent reviewer agreement; medium confidence.
Origin Attribution¶
Primary origin: Criminology & Forensic Studies
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Specialized
Rationale: The report form is characteristic of forensic evidence evaluation, where competing-source hypotheses are expressed through calibrated likelihood ratios.
Related originating lineages:
- Medicine & Healthcare — Diagnostic evidence reporting independently uses likelihood ratios with scope and calibration limits.
- Statistics & Experimental Design — Statistical likelihood and Bayesian evidence weighting supply the report's formal inferential basis.
Review resolution: Both independent reviews assign primary provenance to criminology_forensic. The queued secondary differences (alternate_origin_disagreement) are reconciled by retaining medicine_healthcare, statistics_experimental_design only as formative or independently established lineage(s), not merely as application domains. origin_mode=cross_disciplinary_synthesis records the provenance relationship, while domain_reach=specialized separately records applicability breadth. confidence=high preserves the more cautious assessment, and encyclopedia_synthesis=false records whether either reviewer identified a corpus-specific synthesis.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] The prosecutor's fallacy is the error of equating the probability of the evidence given a hypothesis with the probability of the hypothesis given the evidence — reading a likelihood ratio as if it were the posterior odds of same-source, while ignoring how many other sources in the relevant population could also produce the observation. ↩