# PHD-08 Citation Audit

**Candidate:** PHD-08
**Source file:** `D:/Claude_Code/brain/collegium/candidates/dissertations/PHD-08/dissertation.md` (read-only; not modified)
**Audit date:** 2026-06-18
**Method:** Every reference DOI was resolved against the Crossref REST API (`https://api.crossref.org/works/<doi>`). Non-DOI identifiers (OpenAlex work IDs, NTRS citation IDs) were resolved against their native APIs. arXiv DOIs (`10.48550/...`) were cross-checked against DataCite, which is their registering agent (Crossref returns 404 for arXiv-registered DOIs by design, so a Crossref 404 there is expected, not a defect).

## Summary

| Metric | Count |
|---|---|
| Total references | 139 |
| Resolved (200) | 139 |
| Unverifiable (timeout / network / 403 publisher bot-block) | 0 |
| Flagged dead (confirmed 404/410 or near-miss) | 0 |
| — mistyped_recoverable | 0 |
| — misattributed | 0 |
| — fabricated_neighbor | 0 |
| — not_found | 0 |

**Result: clean.** Every one of the 139 references resolves to a live record at its registering agent. There are no confirmed-dead (404/410) DOIs and no near-miss/fabricated identifiers, so the ARGOS near-miss resolver was not invoked (it only runs on DOIs that fail to resolve; here none do). No repair worklist is required.

## Resolution breakdown by identifier type

| Identifier type | Count | All resolved? |
|---|---|---|
| Crossref DOI | 117 | yes (200) |
| NTRS citation (resolved via `ntrs.nasa.gov/api/citations/<id>`) | 13 | yes (200) |
| OpenAlex work ID | 8 | yes (200) |
| arXiv DOI (`10.48550/...`, resolved via DataCite) | 1 | yes (DataCite 200 + doi.org 200) |

NTRS references resolved via the hardened rule (hit the citations API, not the SPA route, since the SPA 200s every path). Refs resolved this way: 65, 80, 86, 94, 95, 96, 102, 106, 122, 123, 124. (Refs 49, 78, 79, 92, 101, 112, 118, 121, 138 are NTRS/grey-literature items carried under OpenAlex work IDs and resolved there.)

## Three references that required a second pass (all benign)

The first automated pass flagged three DOIs as Crossref-404. All three were confirmed false alarms on recheck and are recorded here for transparency, because each is exactly the kind of pattern that a naive audit would mis-flag as fabricated:

| Ref | Cited DOI | First-pass artifact | Truth | Evidence |
|---|---|---|---|---|
| 40 | `10.1061/(asce)co.1943-7862.0001593` | DOI contains a literal `(` that truncated the first regex parse to `10.1061/(asce`; truncated string 404s | **Resolves.** Full DOI returns Crossref **200**. The `doi.org` 403 seen separately is an ASCE publisher bot-block, which the guardrail classes as unverifiable, never dead. | Crossref 200 on full DOI; ASCE Journal of Construction Engineering and Management is the real venue named in the citation. |
| 100 | `10.48550/arxiv.2107.02637` | arXiv DOI; Crossref returns **404** because arXiv DOIs are registered with DataCite, not Crossref | **Resolves.** DataCite returns **200**; `doi.org` returns **200**. A Crossref 404 on a `10.48550/` DOI is the expected registry behavior, not a dead link. | DataCite 200 + doi.org 200; title "Difference-in-Differences with a Continuous Treatment" matches Callaway, Goodman-Bacon & Sant'Anna's arXiv:2107.02637. |
| 119 | `10.1002/(sici)1097-0266(199809)19:9<853::aid-smj977>3.0.co;2-b` | Legacy Wiley SICI DOI with literal parentheses and angle brackets; truncated to `10.1002/(sici` on first parse, which 404s | **Resolves.** Full DOI returns Crossref **200**. The `doi.org` 403 is a Wiley bot-block (unverifiable), never dead. | Crossref 200 on full DOI; Poppo & Zenger (1998), Strategic Management Journal, is a real and widely cited TCE article. |

## Flagged-reference repair worklist

None. No reference is dead, mistyped beyond recovery, misattributed, or fabricated. The dissertation's reference list passes the citation-integrity check in full.

## Guardrail compliance

- No reference was flagged dead on the basis of a 403 or a timeout. The two `doi.org` 403s observed (refs 40, 119) are publisher bot-blocks and were correctly held as unverifiable, then independently confirmed live via Crossref 200.
- The single Crossref-404 that is not a parse artifact (ref 100, arXiv) was confirmed live at its actual registering agent (DataCite) before being cleared, per the rule that a real publisher/registry-resolvable DOI is never to be called fabricated.