The Gold Standard · Anthropology
Make the decision path visible
A difficult proxy-record appeal may need an identity analyst, records specialist, privacy adviser, language expert and domain practitioner. Each can be careful and still answer a different question. When their conclusions conflict, a meeting alone does not create a sound decision. Majority vote can bury the one discipline that noticed the decisive limit.
The gold standard makes each specialist’s scope and evidence visible, tests whether the conflict is factual, methodological or value-based, and sends the remaining decision to an authorised resolver who explains what was accepted, rejected or left uncertain.
Choose the route closest to your decision
Chapter contents
Chapters 1–3
Chapters 4–6
Chapters 7–9
Chapters 10–12
CHAPTER 1 OF 12 · MAP THE CONFLICT
1. Define the decision before comparing opinions
State the record, appellant, disputed link or authority, consequence and exact outcome required. A specialist asked whether two records probably refer to one person is not answering whether the organisation may lawfully merge them.
Break the escalation into questions that each discipline can actually decide. Record which question is advisory, which is evidential and which requires formal authority.
CHAPTER 2 OF 12 · MAP THE CONFLICT
2. Name every specialist scope and limit
For each opinion, record qualifications, assigned task, evidence seen, method, assumptions, conclusion and stated limit. A conclusion outside the commissioned scope should be marked rather than silently adopted.
Seniority does not enlarge expertise. A technical analyst can describe linkage uncertainty without deciding cultural naming practice, and a legal adviser can define a lawful route without validating an algorithm.
Practical check
Write the decision, evidence and limit in three separate sentences. If one sentence depends on an unstated assumption, name it before continuing.
CHAPTER 3 OF 12 · MAP THE CONFLICT
3. Classify the kind of disagreement
Conflicts may concern an observed fact, source reliability, method, threshold, terminology, jurisdiction, risk tolerance or the value placed on different harms. These need different remedies.
A factual conflict may yield to a primary document. A method conflict may need blind reanalysis. A legitimate value conflict may remain for the authorised decision-maker rather than being disguised as science.
| Element | Record |
|---|---|
| Question | Decision to be made |
| Scope | What each discipline answers |
| Evidence | Version each reviewer saw |
| Limit | What remains unknown |
CHAPTER 4 OF 12 · BUILD FAIR EVIDENCE
4. Give reviewers enough context, but not a preferred answer
Prepare one evidence index and a neutral statement of questions. Where contextual bias is plausible, stage information so the first analysis can occur without knowing the desired organisational outcome or another reviewer’s conclusion.
Context management is not evidence suppression. Essential facts must be available before the decision, with a record of what was revealed at each stage and why.
CHAPTER 5 OF 12 · BUILD FAIR EVIDENCE
5. Compare reasoning, not only final labels
Place the opinions side by side: observations, exclusions, assumptions, decision rule and uncertainty. Two labels that look inconsistent may follow from different questions or different evidence cut-off dates.
Ask each specialist to identify the smallest factual or methodological change that would alter their conclusion. This exposes the real hinge without pressuring anyone to soften a defensible view.
Practical check
Write the decision, evidence and limit in three separate sentences. If one sentence depends on an unstated assumption, name it before continuing.
CHAPTER 6 OF 12 · BUILD FAIR EVIDENCE
6. Worked example: a name match, a kinship record and a legal limit
Suppose a linkage analyst finds a strong match between two proxy records. A language specialist warns that the name pattern is common within the relevant community, while an anthropologist finds the stated kinship role plausible but not unique. The privacy adviser says a merge would disclose protected information if wrong.
The team does not average four opinions. It confirms the shared observations, lowers confidence in identity uniqueness and distinguishes kinship plausibility from identity proof. The final authority declines a permanent merge, allows a reversible review flag and requests one independent source. Changed-context check: if a verified household identifier later appears, the identity evidence changes; the original cultural and privacy limits should still remain in the decision record.
| Type | Useful response |
|---|---|
| Fact | Verify primary source |
| Method | Independent reanalysis |
| Terminology | Define and translate |
| Values or risk | Authorised judgement |
CHAPTER 7 OF 12 · RECONCILE HONESTLY
7. Resolve shared facts first
Create a short schedule of agreed facts, disputed facts and unavailable facts. Verify source versions, timestamps, translations and whether every specialist used the same record state.
Correcting one shared factual error may dissolve the conflict. If it does not, preserve the remaining disagreement rather than reopening settled observations.
CHAPTER 8 OF 12 · RECONCILE HONESTLY
8. Use independent reanalysis when method is the hinge
A qualified reviewer can reproduce the competing methods from the same frozen evidence, preferably without being told which result is favoured. The task is to test support, not force unanimity.
If methods answer different questions, require a translation between their outputs. If neither has adequate validation for the case, the proper result can be unresolved or bounded action.
Practical check
Write the decision, evidence and limit in three separate sentences. If one sentence depends on an unstated assumption, name it before continuing.
CHAPTER 9 OF 12 · RECONCILE HONESTLY
9. Assign final decision authority explicitly
Name the role authorised to balance the evidence, legal limits, reversibility and consequences. The resolver should not claim specialist certainty that the specialists themselves did not provide.
Document conflicts of interest and a substitute authority. A committee may advise, but accountability should not disappear into the committee name.
| Stage | Output |
|---|---|
| Freeze record | Common evidence set |
| Compare reasons | True hinge identified |
| Resolve facts | Agreement schedule |
| Decide | Named authority and reasons |
CHAPTER 10 OF 12 · CLOSE AND IMPROVE
10. Write a decision that keeps the disagreement visible
The decision should identify each material opinion, the decisive facts, the rule applied, the remaining uncertainty and why one action was chosen. Explain whether any opinion was rejected, limited or simply not controlling.
Avoid generic statements such as “the experts agreed” when they did not. Readers can accept a bounded decision more readily when its unresolved edge is stated plainly.
CHAPTER 11 OF 12 · CLOSE AND IMPROVE
11. Propagate the outcome without spreading the whole case
Update linked records, access controls and downstream decisions that rely on the disputed link. Send each recipient only the correction, status and reason they need.
Preserve the original versions and the reconciliation trail. A final decision is incomplete if an obsolete automated link continues to operate elsewhere.
Practical check
Write the decision, evidence and limit in three separate sentences. If one sentence depends on an unstated assumption, name it before continuing.
CHAPTER 12 OF 12 · CLOSE AND IMPROVE
12. Calibrate disciplines without erasing independence
Review recurring conflict types, overturned conclusions, missing evidence and unclear referral questions. Joint case review can improve shared terminology and referral design.
Calibration should help specialists understand one another’s boundaries, not train them to converge on a preferred result. The gold standard protects reasoned dissent while making final authority answerable.
| Check | Proof |
|---|---|
| Opinions | All material views retained |
| Authority | Decision role recorded |
| Propagation | Dependent systems updated |
| Learning | Recurring cause reviewed |
Frequently asked questions
Should the majority specialist view win?
Not automatically. Disciplines may answer different questions, and one minority view may identify a decisive legal, cultural or technical limit.
Can reviewers see one another’s opinions?
Yes when reconciliation requires it, but an initial blind or staged review can reduce anchoring where contextual bias is a concern.
What if no method resolves the conflict?
Use a bounded or unresolved outcome, reversible safeguards and a clear plan for the evidence that could change the decision.
Who owns the final decision?
The role with formal authority to integrate the evidence and consequences, named in advance or transparently assigned for the case.
Continue through verified routes
Core routes
Anthropology subject root · Gold Standard of Education · Gold Standard directory · How X Works hub
Specialist routes
- Escalation queue overload, specialist capacity and resolution ownership
- Appeal reviewer drift, case-mix change and recalibration
- Appeal outcome consistency and escalation thresholds
- Match-threshold appeals and reversing automated links