Immigration Judge Analytics Methodology

Scope and assignment

The directory universe is the current EOIR immigration-judge roster in the serving dataset. “Current EOIR assignment” describes that roster. A “TRAC report court” and report period describe a source report and may differ from a judge's current assignment. Every preserved report remains visible in report history.

Outcomes and denominators

These statistics cover completed asylum decisions in the cited TRAC reports, not every immigration-court matter. Decisions, or N, equal asylum granted plus other relief granted plus denied. “Any relief granted” equals asylum granted plus other relief granted. Rates divide an outcome count by N; displayed rates are rounded to one decimal place.

Direct and fallback coverage

TRAC publishes a judge report when its eligibility rules are met, including at least 100 asylum decisions over the report period. Direct judge data appears only for a verified judge report. Court and national fallback values are explicitly labeled benchmarks and are never presented as individual statistics. Rate sorts keep direct results distinct from fallback benchmarks.

Comparators, periods, and annual trends

Judge comparisons use the court and U.S. comparator stored with the same TRAC report. Aggregate court and U.S. annual trends are precomputed from exact structured TRAC outcome counts. Judge-level annual values appear only when exact fiscal-year observations are published; annual observations with fewer than 25 decisions are suppressed, and values are never inferred from chart pixels. Missing and suppressed years remain unavailable, and FY2026 is year-to-date through the published source cutoff.

Cohorts, ranks, and percentiles

Peer cohorts contain direct-data judges with the same TRAC report court and report period and require at least five judges. Percentile is the share of qualifying denial rates at or below the subject rate, including the subject. Small cohorts also show competition rank from highest denial rate; tied rates share a rank. P25, median, P75, and interquartile spread use the same cohort. These are descriptive positions, not quality scores.

Representation and nationality

When a report supplies representation or nationality context, KnowJudges displays those report-specific distributions with the report period and source. They describe the decided-case mix and do not measure judge quality or causation.

Sources and limitations

Primary analytics source: TRAC Immigration. Current assignments come from the DOJ EOIR immigration-judge directory. Each result retains its reporting period, report ID, source URL, publication date when available, and retrieval date. Multiple reports for a judge remain separate observations and are never merged by averaging percentages. Outcomes can reflect case mix, representation, nationality, detention status, policy, geography, time, and other factors. KnowJudges does not turn these figures into a composite strictness, leniency, quality score, recommendation, or causal claim.