Working Data skill
Resolve conflicting sources
Diagnose why credible sources disagree and determine what can be concluded without forcing false consensus.
Disagreement is evidence about definitions, methods, populations, timing, or uncertainty. Make those differences comparable before judging the conclusions.
When to use
- Use when reputable sources report different counts, effects, histories, standards, or advice.
- Distinguish factual disagreement from different values, thresholds, or decision objectives.
Procedure
- Put the conflicting claims side by side in their original wording with dates and versions.
- Parse each claim’s population, unit, geography, period, definitions, comparison, and uncertainty.
- Trace each source to its evidence and identify shared data, citations, authors, or institutional dependencies.
- Compare collection methods, samples, exclusions, adjustments, models, incentives, and publication timing.
- Check whether later corrections, revisions, events, or definitions explain the difference.
- Recalculate or normalize only where units and constructs are genuinely comparable.
- Assess which source is more direct and fit for the exact question, not which is more prestigious.
- Classify the conflict as semantic, temporal, methodological, evidentiary, value-based, or unresolved.
- Write a reconciliation: what agrees, what differs, why, which conclusion is justified, and what evidence would resolve the rest.
Failure plan
- Do not split the difference between incompatible estimates.
- Do not treat multiple reports using one upstream dataset as independent support.
- If the conflict affects a high-stakes decision, seek domain review and current authoritative material.
- Preserve a genuine unresolved conflict instead of choosing a convenient source.
--- name: resolve-conflicting-sources category: data description: Diagnose why credible sources disagree and determine what can be concluded without forcing false consensus. Use when research returns incompatible facts, estimates, or recommendations. --- # resolve-conflicting-sources Disagreement is evidence about definitions, methods, populations, timing, or uncertainty. Make those differences comparable before judging the conclusions. ## When to use - Use when reputable sources report different counts, effects, histories, standards, or advice. - Distinguish factual disagreement from different values, thresholds, or decision objectives. ## Procedure 1. Put the conflicting claims side by side in their original wording with dates and versions. 2. Parse each claim’s population, unit, geography, period, definitions, comparison, and uncertainty. 3. Trace each source to its evidence and identify shared data, citations, authors, or institutional dependencies. 4. Compare collection methods, samples, exclusions, adjustments, models, incentives, and publication timing. 5. Check whether later corrections, revisions, events, or definitions explain the difference. 6. Recalculate or normalize only where units and constructs are genuinely comparable. 7. Assess which source is more direct and fit for the exact question, not which is more prestigious. 8. Classify the conflict as semantic, temporal, methodological, evidentiary, value-based, or unresolved. 9. Write a reconciliation: what agrees, what differs, why, which conclusion is justified, and what evidence would resolve the rest. ## Failure plan - Do not split the difference between incompatible estimates. - Do not treat multiple reports using one upstream dataset as independent support. - If the conflict affects a high-stakes decision, seek domain review and current authoritative material. - Preserve a genuine unresolved conflict instead of choosing a convenient source. ## Done - Claims are checked on aligned definitions, periods, populations, methods, and provenance - The cause of disagreement is evidenced, verified where possible, or explicitly unresolved - The final conclusion states its scope, confidence, and decision consequence