Nason/courtlistener-database

CourtListener Judges and Financial Disclosures

CourtListener's bulk judge database (16,190 judges, 51,289 positions, education, party affiliations, race), its registry of 3,353 courts, and 32,336 federal judicial financial disclosure reports with their gifts, debts, reimbursements, agreements, positions and non-investment and spousal income, typed straight from CourtListener's PostgreSQL schema. Judges carry the FJC id, so the tables join to the fjc-judges database. 1,238,528 rows across 20 tables.

Table	Description	Row Count	Column Count
opinion_cluster_panel	Judge-to-opinion-cluster panel links (person_id joins judges; cluster ids refer to CourtListener opinion clusters, not included here)	831,965	3
oral_arguments	Oral-argument audio metadata (case name, docket, court, date, duration); the 2024-05-06 export	89,468	21
positions	One row per position held: judgeships (court_id joins courts), clerkships, offices and other jobs, with nomination, confirmation, start and termination dates, appointer, predecessor and how the seat was obtained	51,289	38
originating_court_information	Lower-court information for appellate dockets (judges, dates, docket numbers)	38,926	15
disclosure_positions	Outside positions reported on a disclosure	37,050	7
disclosure_reimbursements	Travel reimbursements reported on a disclosure	33,472	10
financial_disclosures	One row per federal judicial financial disclosure report (person_id joins judges): year, report type, amendment flag, page count and the original PDF's storage path	32,336	16
disclosure_spousal_income	Spousal income reported on a disclosure	20,174	7
disclosure_debts	Liabilities reported on a disclosure, with value code	18,775	8
judges	One row per person in CourtListener's judge database (federal and state judges, plus some other public figures): names, birth/death, gender, religion, FJC judge id (fjc_id joins fjc-judges.judges.jid)	16,190	26
disclosure_non_investment_income	Non-investment income (teaching, book royalties ...) reported on a disclosure	15,302	8
educations	One row per degree (school_id joins schools)	12,777	8
disclosure_agreements	Agreements reported on a disclosure (financial_disclosure_id joins financial_disclosures)	10,007	7
political_affiliations	One row per recorded party affiliation with its source and dates	8,486	10
judge_races	Person-to-race links (race_id decoded in the race column)	6,542	4
schools	Law schools and universities	6,011	6
courts	Every court CourtListener tracks (federal, state, tribal, special), with jurisdiction, PACER and FJC ids, dates and parent court	3,353	20
courthouses	Courthouse addresses per court	3,352	11
disclosure_gifts	Gifts reported on a disclosure	2,025	8
opinion_joined_by	Judge-to-opinion 'joined by' links (person_id joins judges; opinion ids refer to CourtListener opinions, not included here)	1,028	3

Query it remotely

INSTALL httpfs; LOAD httpfs;
ATTACH 'https://huggingface.co/datasets/Nason/courtlistener-database/resolve/main/courtlistener.duckdb' AS cl (READ_ONLY);

SELECT j.name_last, j.name_first, d.year, COUNT(g.id) AS gifts
FROM cl.financial_disclosures d JOIN cl.judges j ON j.id = d.person_id
LEFT JOIN cl.disclosure_gifts g ON g.financial_disclosure_id = d.id
WHERE d.year >= 2018 GROUP BY 1, 2, 3 HAVING COUNT(g.id) > 0 ORDER BY 4 DESC LIMIT 20;

Or with the datapond packages: pip install datapond / pak::pak("datapond-db/datapond-r").

Build pipeline and caveats: https://github.com/ian-nason/courtlistener-database Source: Free Law Project, CourtListener bulk data (public records; please credit CourtListener).