PetitionHQ

Methodology

Verification Methodology

PetitionHQ does two things, in order: first it resolves one real researcher across 7+ public databases — dropping namesakes — then it scores how well that person's self-reported claims hold up against the public record. Trust is about authenticity, not prestige.

What We Verify

Each report queries 7+ independent sources. A source only counts when it agrees with the resolved identity — a database that returns a different person who happens to share the applicant's name is dropped as a namesake, never counted and never held against them.

SourceWhat It ConfirmsWhat It Corroborates
OpenAlexPublications, citations, co-authors, research topics, affiliationThe primary identity anchor for most researchers — and the body-of-work signal
ORCIDA self-claimed researcher ID, publication list, affiliationThe strongest deterministic identity anchor, when the applicant has one
CrossrefDOI-level authorship and publication metadataIndependent confirmation of authorship on indexed works
RORThat an institution is a real, registered organization, its type and countryThe institution component — the named affiliation actually exists
NSF Award SearchNSF grant awards by PI nameFederal funding under the resolved identity
NIH RePORTERNIH grant awards by PI nameFederal funding under the resolved identity
USPTO PatentsViewGranted patents by inventor nameInventorship under the resolved identity

What We Don't Verify

Some credentials lack comprehensive public databases. These items are marked “Self-Reported” in the verification report — meaning they come from the applicant's intake form and have not been independently confirmed.

  • Invited talks and conference presentations (no comprehensive public database exists)
  • Awards and honors (issuing bodies rarely maintain searchable databases)
  • Peer review and editorial board roles (reviewer data is confidential by design)
  • Specific degree dates or transcripts (not available via public APIs)
  • Media coverage or press mentions
  • Letters of recommendation content or authorship

How the Trust Score Is Computed

The trust score is a deterministic number from 0 to 100 — no machine learning, no subjective judgment. It is built as an evidence floor, not a pile of additive bonuses: the floor is set by how many independent sources agree it's the same person, and a few components adjust it from there. The whole thing runs after we've resolved who the applicant is.

ComponentRangeWhat it measures
Identity corroboration0 / 35 / 45 / 55The floor. Set by how many independent sources agree on one resolved person: none → 0, one source → 35, two → 45, three or more → 55. This is the heart of the score.
Publication substance0 to +15A real, indexed body of work — number of works, aggregated citations, and h-index, counted only over sources that agree on the same person.
Claim consistency−15 to +20Self-reported numbers checked against the public record. Honest, modest claims earn points. The only way to lose points is an egregious overclaim against a confidently-resolved record (e.g. claiming 5,000 citations against a record showing ~500).
Institution & funding0 to +10The named institution is registered in ROR, and/or grants are corroborated under the resolved identity.
No corroborating source (ghost)0If not a single public source corroborates the identity, the score is 0 — full stop. We never invent trust from a name alone.

The components sum and the result is clamped to 0–100. A researcher corroborated by three agreeing sources starts at a 55 floor, then adds publication substance, claim consistency, and institution corroboration on top — while someone no public source can place stays at 0.

What “Publicly Corroborated” Means — and Doesn't

PetitionHQ confirms that a researcher profile matching your credentials exists in independent public databases. This is different from confirming that you own that identity.

Concretely: if OpenAlex returns a profile matching your name, field, and institution with a consistent publication count, that profile is “publicly corroborated” — an independent database agrees it exists. We do not confirm you are the person named on that profile.

The one exception: ORCID OAuth

When you connect your ORCID account directly, you prove ownership of that identifier — the credential is issued by ORCID to you personally, not inferred by name-matching. We label this “identity confirmed” rather than “publicly corroborated.” This is a stronger and different claim.

Attorney-facing reviews display both the corroboration level and the methodology, so attorneys can weigh the evidence for themselves. The trust score makes the existence-vs-ownership distinction explicit in every lead.

What the Trust Score Does Not Measure

Trust answers one question: is this a real, findable researcher whose self-reported claims hold up? It is not a measure of how strong their NIW case is — case strength is a separate assessment. A modest but completely honest researcher can score high on trust and still have a borderline case, and vice-versa.

Two guarantees protect honest applicants. First, a database your field simply doesn't use never costs you points — a clinician is not penalized for being absent from a computer-science index. Second, under-indexing is never read as a lie: if your record shows fewer works than you claimed, we treat it as indexing lag, not misrepresentation — the only penalty is an egregious overclaim against a record we've confidently resolved as yours.

Refund Policy

Material misrepresentation

If the verification report reveals that an applicant materially misrepresented their credentials, the claiming attorney can request a full refund within 30 days. Refund requests are reviewed by our team and processed within 5–10 business days.

Applicant non-response

If an applicant does not complete intake within 14 days of a claim, the attorney is automatically refunded and the lead is returned to the available leads pool.

Last updated: June 9, 2026

How PetitionHQ works →
Verification Methodology — PetitionHQ Confirms Credentials | PetitionHQ