RESEARCH EVIDENCE
Keep the source, timing and assumptions with the result.
A number without its source, observation time and method is hard to interpret later. A useful research record links the proposition and contract to the evidence examined, then states uncertainty and limits in plain language.
What a research record should preserve
The product's documented response fields and an analyst's explanatory notes serve different purposes; check the response contract for each route.
| Evidence element | Why retain it | Scope |
|---|---|---|
| Source and source time | Identify where an observation came from and when the source recorded it. | Product response fields vary by endpoint; source delivery is rights-gated. |
| Event and contract references | Connect the proposition under review to the specific contract record examined. | Documented identifiers exist; meaning depends on the endpoint and mapping state. |
| Quality or freshness state | Show whether the observation is suitable for the intended comparison at that point in time. | Returned where defined by the response contract; never assume every route has the same fields. |
| Relationship and review state | Make the comparison rationale and unresolved terms visible to the next researcher. | Source-based review requires an eligible record, current rights and human assessment. |
| Method and version identifiers | Identify the calculation or mapping version relevant to the displayed result. | Identifier fields are endpoint-specific; execution and methodology versions are distinct. |
| Assumptions, limitations and conclusion | Explain what the record supports and what it does not establish. | Analyst-supplied in the educational template; not claimed as a built-in export schema. |
Illustrative record: one unresolved relationship
This compact example shows how to retain a conclusion without pretending it came from a live source.
- Record type
- EDUCATIONAL EXAMPLE · SYNTHETIC
- Proposition
- Will the Meridian Council ratify the Northwind agreement by 30 November?
- Contract references
- SYNTHETIC-A and SYNTHETIC-B; fictional identifiers
- Source and source time
- None supplied; this example contains no venue observation.
- Comparison result
- UNRESOLVED — “ratify” and “announce” do not state the same outcome.
- Limit
- No price, settlement, legal conclusion, executable offer or source entitlement is represented.
Separate captured evidence from interpretation
Preserve source material and its observation time where source terms and rights permit. Record the versioned reference definition and the exact terms reviewed. Keep the human interpretation and rationale attached to that evidence, and mark a prior assessment as needing review after a material source or definition change.
A digest can help detect whether retained content changed; a digest is not a digital signature, certification or proof that the source itself was correct. Provenance improves traceability but does not make every inference valid.
Access, availability and reuse remain separate questions
The public terminal and Developer Sandbox provide synthetic examples. Canonical organization data routes require a signed-in organization session and applicable source rights; a Developer Sandbox key does not authenticate those routes, and production API-key transport is not available for them. Bulk historical exports remain planned.
The downloadable contract-comparison checklist and evidence-record template is an educational worksheet. It is not a built-in Predict Here export, a source license, or permission to redistribute captured material. See the documented response envelope and risk-research workflow.
Discuss a source-linked research workflow
Scope the records, permissions, retention needs and research decisions an institutional evaluation would need to cover.