Cue-Validity Audit¶
Method — instantiates Relevance-Substitution Detection and Correction
Tests whether a psychologically active cue has a genuine evidential relation to the target claim, and of what kind.
Cue-Validity Audit takes a single cue that is exerting psychological pull — a candidate's prestige, a report's fluent prose, a vivid anecdote, a confident tone — and asks the diagnostic question the pull tends to skip: what is this cue's actual evidential relation to the claim it is being used to support? The method identifies the cue, names the pathway by which it grips attention, and then runs it through a typed test: is the relation causal, logical, predictive, a matter of construct validity, source reliability, or mere membership — or is there no probative relation at all? The output is a relation verdict, a statement of whether and how the cue bears on the target. What makes this mechanism distinct is its focus on the nature of the relation for a single active cue, worked out on the merits; it does not pre-register a whole grid, dissect an analogy's feature mapping, or mask the cue from view — it interrogates one cue's claim to be evidence.
Example¶
A hiring manager notices that a candidate for a senior engineering role graduated from an elite university, and feels the pull of it — the résumé suddenly reads as stronger. The Cue-Validity Audit takes that cue seriously enough to test it. First, the cue is named (alma-mater prestige) and its pathway logged (status halo: prestige of the school bleeding into perceived competence). Then the typed test: does elite-school attendance have a predictive relation to performance in this role? The manager reasons through it — the school signals general ability screened years ago, but the role turns on recent systems-design work and collaboration, for which the work sample and references are far more predictive; the degree adds little once those are in hand, and much of its pull is halo. The verdict: weak predictive relation, largely redundant with better evidence, substantial non-evidential halo. The cue is not banned, but it is demoted to near-zero incremental weight, and the decision leans on the work sample the audit identified as actually predictive.
How it works¶
- Isolate the cue. Pull the single active signal out of the case so it can be examined on its own rather than as part of an overall impression.
- Name the pathway. Record how the cue is exerting force — halo, fluency, vividness, authority — since the pathway hints at whether the pull is evidential or merely psychological.
- Type the relation. Run the cue against a menu of possible evidential relations (causal, logical, predictive, construct validity, source reliability, issue membership) and determine which, if any, actually holds for this claim.
- Issue a verdict. Classify the cue as valid, conditionally valid, weak, redundant, confounded, or non-evidential, with the relation type named — the audit's deliverable.
Tuning parameters¶
- Relation menu breadth — how many relation types the cue is tested against. A broad menu catches subtle valid relations but is slow; a narrow one is fast but may miss the one way a cue is genuinely probative.
- Incremental-validity stance — whether the cue is judged in isolation or for what it adds beyond evidence already in hand. Judging incrementally exposes redundant cues but requires knowing the rest of the evidence; judging in isolation is simpler but over-credits redundant signals.
- Verdict granularity — binary valid/invalid versus a graded, typed relation. Graded verdicts feed reweighting precisely but take more judgment; binary is quick but blunt.
- Evidence demanded — how much justification the relation claim itself requires. Demanding empirical support for "predictive" resists armchair rationalization but slows the audit.
When it helps, and when it misleads¶
Its strength is precision about why a cue does or does not count: by typing the relation, it separates cues that feel evidential from cues that are, and it names the exact relation a valid cue supplies. This is the working sense of cue validity from Brunswik's lens model — the real correlation between a cue and the criterion it is taken to indicate[1] — made into a deliberate test rather than an implicit assumption. Naming the relation type is what lets a downstream reweighting be principled rather than a shrug.
Its failure mode is analysis that outruns the evidence about the relation itself: it is easy to assert that a cue is "predictive" or "confounded" without any real basis, so the audit can launder a prior into a relation verdict. It also handles one cue at a time and can miss substitution that lives in the interaction of several cues. The classic misuse is selective auditing — running the test only on cues pointing the wrong way and waving through the ones that flatter the preferred conclusion. The guarding discipline is to audit the cues on both sides of a judgment, to require that a claimed relation be backed by something more than plausibility, and to hand the graded verdict to a separate reweighting step rather than letting the auditor also decide the outcome.
How it implements the components¶
substitute_signal_register— isolating the cue registers the specific psychologically active signal under test.psychological_activation_pathway— naming halo, fluency, or authority records how the cue grips, distinguishing its pull from its proof.relevance_test— typing the relation (causal, predictive, construct, membership, or none) is the core probative test.
It does not set the fidelity bar a proxy must clear (relevance_standard) or cap the weight it may carry via a correction_and_reweighting_rule — that proxy-specific fidelity-and-weight assessment belongs to its close cousin Proxy-Relevance Audit; this method types the relation of any active cue rather than auditing a metric standing in for a construct.
Related¶
- Instantiates: Relevance-Substitution Detection and Correction — provides the typed test of whether and how a single active cue bears on the claim.
- Sibling mechanisms: Affect–Evidence Split Prompt · Red Herring Filter Checklist · Analogy Mapping Table · Question–Evidence Matrix · Decision-Basis Disclosure · Reweighted Update Log · Proxy-Relevance Audit · Blinded or Masked Review
Editorial Notes¶
Form Classification¶
Form family: Assessment, Review & Assurance
Rationale: Cue-Validity Audit operates as a bounded evaluation of existing evidence or work that produces a finding or disposition because it tests whether a psychologically active cue has a genuine evidential relation to the target claim, and of what kind.
Independent corroboration: The frozen evidence defines Cue-Validity Audit as 'Tests whether a psychologically active cue has a genuine evidential relation to the target claim, and of what kind', so its operative form is Assessment, Review & Assurance.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Psychology
Origin pattern: Single lineage
Present-day reach: Multi-domain
Rationale: Brunswikian judgment psychology established ecological cue validity; statistical evidence assessment operationalizes the typed relation verdict.
Related originating lineages:
- Statistics & Experimental Design — Correlation, prediction, reliability, and confounding analysis quantify whether a cue actually bears on the criterion.
Review resolution: Brunswikian judgment psychology established ecological cue validity; statistical evidence assessment operationalizes the typed relation verdict.
Review outcome: Reconciled after independent review; high confidence.
References¶
[1] Brunswik, E. Perception and the Representative Design of Psychological Experiments. 2nd ed., rev. and enl. University of California Press (1956). Defines cue validity in a Brunswik lens model as correspondence between a cue and the criterion it is taken to indicate. registry ↩