# Claim-level dependency graph and text equivalent

Arrows run from the credited upstream source to the downstream account. They are attribution/dependency records, not evidence of the writer's actual retrieval process. All records concern the focal 2025 Singapore car-registration claim unless otherwise labelled.

```mermaid
flowchart LR
  LTA["LTA registration statistics · family F1"] -->|"Confirmed origin"| P["D01 · current annual PDF · primary record"]
  LTA -->|"Confirmed producer attribution + same pair · COM"| H["D03 · HardwareZone · selected"]
  LTA -->|"Confirmed producer attribution + same pair · COM"| B["D05 · Business Times · selected"]
  LTA -.->|"Inferred lineage; exact origin unknown · UNK"| C["D02 · CNA · selected"]
  ORG["LTA named producer · exact inputs unknown"] -.->|"Indexed attribution only · access-qualified · UNK"| F["D04 · F&L · selected"]
  ST["D06 · Straits Times · outside sample"] -->|"Explicit republication credit + exact link · DER"| H
  P -->|"Same-data recalculation · COM"| CALC["NE calculation 45.0% · no new observations"]
  T["Walter Theseira · reported view"] -->|"Attributed context · CTX"| C
  LO["Benjamin Loo · reported view"] -->|"Attributed context · CTX"| C
  ORG -->|"Confirmed named producer; inputs unknown · UNK"| M["D07 · Motorist · supplementary 45.1%"]
```

## Accessible text equivalent

1. LTA publishes the 2025 annual registration table D01. The current downloaded version supports 23,684 Electric registrations and 52,678 cars after extraction and reconciliation.
2. HardwareZone D03 and The Business Times D05 explicitly attribute that 2025 pair to LTA. They are two claim appearances within the selected four-publication sample grouped into observation family F1. Specific historical PDF editions remain unknown.
3. HardwareZone D03 separately credits The Straits Times D06, identifies the original author/date and links the exact article. This confirms DER reporting dependence. D06 is outside the selected downstream count; its credit alone is not the within-sample family-collapse evidence.
4. CNA D02 reports the same pair without visible LTA attribution. An LTA lineage is INFERRED; its origin assignment remains UNK. No account-to-account copying route is established.
5. F&L D04 explicitly names LTA in the inspected indexed excerpt but prints 45.1%. The numerical inputs and exact observation-family assignment remain UNRESOLVED. The attribution is ACCESS-QUALIFIED. It is not silently merged with 45.0%.
6. New Eruditions calculates 45.0% from D01. That checks extraction and arithmetic without becoming another observation family. An independent table-cell cross-check verifies extraction only.
7. CNA separately attributes contextual views to Theseira and Loo. Those two CTX claims are excluded from the national-statistic family count; independent interview collection is not established.
8. Supplementary Motorist D07 names LTA and prints 45.1% in an inspected body. Its underlying inputs are unresolved, and no route from Motorist to F&L is proved.

## Count reconciliation

Selected downstream publications: four (CNA, HardwareZone, F&L access-qualified, Business Times). Three inspected full claim-bearing bodies plus one indexed excerpt. Primary-origin records: one LTA PDF. Intermediate credited article outside sample: one ST record. Supplementary discrepancy account: one Motorist record. Methodological source: one Cochrane chapter. Do not add those quantities into a confidence score.

Identifiable observation families for the focal pair: one, F1. It has two confirmed selected downstream members plus its separately counted primary representative. CNA and F&L assignments remain unresolved at exact-observation level; neither is assumed independent or automatically merged. No IND or PAR national-registration family is established.