Methodology & limitations
What this assessment can—and cannot—tell you.
NIWHQ is an independent informational research prototype. It organizes evidence questions and explanations; it does not provide legal advice or make an adjudication.
No approval probability
NIWHQ does not estimate a percentage chance of approval and does not use a 0–100 USCIS score. Immigration decisions are case-specific and the available datasets do not provide a representative denominator.
Approved-case survivor bias
Public approval summaries contain selected successful cases, often from law-firm clients. They do not represent every approved or filed petition.
AAO appeal-selection bias
AAO decisions describe a selected population of appeals and motions, not all denied petitions. They are useful for understanding reasoning—not denial rates.
Missing is not zero
A source may omit publications, citations, reviews, degrees, or roles. NIWHQ preserves unknown values instead of silently converting them to zero.
Country does not change substantive merit
Country of birth may later support chargeability information or transparent comparison filters. It does not increase or reduce evidence readiness.
Bounded, source-traceable research
Phase 1B uses 100 public approved-case summaries and 48 machine-extracted AAO adverse-decision records. Every displayed record retains a source link or exact dataset locator. These bounded samples are research fixtures—not representative approval or denial populations.
Transparent local matching
Comparable cases are ranked deterministically by mutually observed features such as pathway, field, career context, evidence issue, and recency. Missing source values do not become zero. Similarity labels are explanatory groupings, never approval probabilities.
Legal framework boundary
NIW educational labels are organized around Matter of Dhanasar and EB-1A labels around the regulatory criteria and Kazarian final-merits framework. The prototype does not replace current authority review or individualized legal analysis.
No external analysis provider
Assessment, matching, and optional endeavor generation run locally with deterministic fixtures. No production database, document upload, paid model, external AI provider, or account system is connected.