Evidence — Documented Exclusions as Corpus Methodology

A self-reflective synthesis on the AntiAlienate corpus's 243 documented exclusions across the practitioner directories. Compiled to dignify the exclusion ledger as a methodological feature of the corpus rather than treating it as a register of failures. CC BY 4.0.

Why exclusions are a feature, not a defect

Most public PA practitioner directories operate by silent omission: names that did not pass internal vetting simply do not appear, with no documented reason and no audit trail. The reader cannot tell whether a missing name was never considered, was considered and rejected, or was considered and is pending verification.

The AntiAlienate corpus operates a different epistemic standard. Every entry that was considered but not included carries:

• a stable id resolvable across the corpus as practitioner:<id>,
• a documented exclusion_reason that explains why the decision was taken,
• an optional primary_source_url pointing to the documentary basis for the decision,
• optional references pointing into the broader corpus where the exclusion connects to other evidence.

This converts each exclusion from invisible non-inclusion into a documented decision with an audit trail. Future contributors can re-examine each decision, re-verify each candidate, and promote exclusions to entries when verification standards are met. The ledger is a working register, not a graveyard.

The 243 — by reason cluster

Cluster	Count	Character
Unverifiable	120	No public primary-source record located via systematic search across regulator registries, LinkedIn, ResearchGate, university bios, peer-reviewed PubMed/Crossref, and judgment text (court-appointed expert). Distinct from "does not exist" — the practitioner may exist but failed the directory-quality verification threshold.
Other	91	Mixed: includes structural-non-discrete entities (e.g. Argentine "mediación familiar" practiced via multiple discrete entities), placeholder records, and reasons not cleanly captured by the other clusters.
Out-of-scope	16	Verified practitioners whose primary professional identity falls outside the receiving file's scope (e.g. attorney appearing in a therapists file, post-traumatic-growth coach appearing in a forensic-evaluator file). These are typically re-routed to the correct file in subsequent passes.
Regulatory	6	License inactive, struck off, suspended, or otherwise regulatorily out-of-status with the receiving jurisdiction's professional regulator. Primary-source-required.
Duplicate	5	Practitioner already present under another canonical entry (often a cross-jurisdictional dual-identity case where the primary entry is in another country file).
Safeguarding	5	Court-documented safeguarding concerns or "ethical-naming safeguard" patterns (LATAM editorial convention) where naming would risk amplifying stigma without epistemic gain.

Cluster-by-cluster dignification

Unverifiable (120)

These are not an indictment of the practitioners named. The verification threshold for the directory is intentionally high: HCPC/HPCSA/APBs registration, public practice site, peer-reviewed publication record, or court-appointed expert role in a published judgment. Many practitioners legitimately work outside this footprint — private practice without a website, paper-publication in journals not indexed in CrossRef, registration in a regulator whose register is not publicly searchable. The exclusion records the verification gap; it does not record an evaluative judgment of the practitioner's work.

What an unverifiable exclusion means: at the moment of corpus compilation, primary-source verification could not be completed within the directory's editorial standard. The entry remains available for promotion when new sources surface.

Out-of-scope (16)

These are dignified relocations. The practitioner is verified, but the receiving file is wrong. Howard Watson (attorney, PAFSA co-founder) was excluded from practitioners/therapists/africa.json because he is primarily an attorney; the same entity would be valid in practitioners/lawyers/africa.json. The exclusion is a routing note, not a rejection.

Safeguarding (5)

The LATAM "ethical-naming safeguard" pattern is methodologically distinctive within the corpus. Where a Spanish-language search surfaces a practitioner whose only public artefact is a tabloid-style attack-piece or a name surfacing in a contested family-court proceeding, naming them in the directory would amplify the stigma without producing epistemic value. The exclusion records the editorial decision; the structural finding records the pattern.

Regulatory (6)

A regulator's published striking-off, suspension, or inactive-registration record is treated as primary-source-sufficient for exclusion with a regulator-citation primary_source_url. These are the highest-evidence exclusions in the corpus.

Duplicate (5)

A practitioner whose canonical entry is in another file appears in a country directory as an excluded reference rather than as a duplicate entry. Sonia Vaccaro is excluded from practitioners/therapists/ar.json as ar.vaccaro-sonia-cross because her canonical entry is practitioner:es.vaccaro-sonia (Spain). The exclusion is a cross-link, not a rejection.

Other (91)

The largest remaining cluster captures structurally heterogeneous reasoning that resists a single label. Subsequent re-categorisation passes will refine this cluster.

How to use exclusions

As referenceable nodes

Every exclusion has a stable id resolvable as practitioner:<id>. Case studies, evidence pages, jurisdiction sidecars, and other practitioner files can cite an exclusion in their references array. The cross-link resolver (bin/aa-build refs) treats exclusions and entries as a single namespace.

Example: a case study analysing a court-appointed evaluator whose practice is unverifiable in the public record may cite practitioner:ca.jane-doe (excluded) to record the audit-trail link, while explaining in editorial_notes why the practitioner was not promoted to an entry.

As a promotion pipeline

An exclusion is not permanent. When primary-source verification becomes available (regulator register comes online, peer-reviewed publication appears, practice site indexed), the exclusion can be promoted to an entry:

1. Verify the new primary source against the editorial standard.
2. Move the exclusion JSON from excluded[] to entries[] in the country file.
3. Add stance, stance_notes, verification_notes, and the new primary_source_url.
4. Run bin/aa-build validate + bin/aa-build refs to confirm integrity.

The audit trail of the exclusion is preserved in git history. The promotion is itself a documented decision.

As a critique-camp safeguard

Exclusions are not concentrated on one side of the recognition-vs-critique register. The cluster-by-jurisdiction distribution shows exclusions across both recognition-camp and critique-camp practitioners. This is itself a methodological signal: the corpus does not preferentially exclude critique-camp voices, nor does it preferentially exclude recognition-camp voices. Both camps face the same primary-source standard.

Comparative posture

Most PA-active practitioner directories in the public domain (PASG member registry, BAPAA practitioners list, various national professional-association rosters) operate by inclusion-only with no documented exclusion register. The AntiAlienate corpus's documented-exclusion approach is structurally adjacent to:

• Cochrane systematic-review methodology: studies excluded from a meta-analysis are documented with reasons, not silently dropped.
• Peer-review desk-rejection logs: journals that publish desk-rejection statistics with cluster-categorised reasons (e.g. "scope mismatch", "methodological concerns", "outside review remit").
• Open-source security advisory databases (CVE, GHSA): every excluded vulnerability report carries a documented "not-applicable" reason rather than disappearing.

In each case the documented exclusion is treated as a feature of epistemic rigour, not as a defect. The AntiAlienate corpus operates the same convention for the practitioner directories.

Future work

1. Re-categorisation pass: the 91-entry "other" cluster will be refined into structurally meaningful sub-clusters in subsequent passes.
2. Promotion pipeline tooling: bin/aa-promote-exclusion <id> to streamline the JSON-edit-and-validate cycle for promoting exclusions to entries when new sources surface.
3. Verification-status field: optional last_verified timestamp on exclusions to record when the verification attempt was last made, supporting periodic re-verification cycles.
4. Cross-link audit: surface exclusions referenced by case studies / evidence pages in analytics.html with their connecting context.

Cross-references

• See /practitioners/README.md for editorial standards including exclusion documentation rules.
• See evidence:methodology-and-corpus-construction for the broader four-entity-type methodology.
• See analytics.html for the full 243-row exclusion table with reasons and source URLs.

---

Canonical: https://raw.githubusercontent.com/AntiAlienate/antialienate-knowledge/main/evidence/documented-exclusions-as-corpus-methodology.md