Datasets:

Nason
/

scdb-database

license: cc-by-nc-3.0
task_categories:
  - tabular-classification
  - tabular-regression
tags:
  - supreme-court
  - courts
  - judicial-politics
  - scdb
  - law
  - duckdb
  - datapond
pretty_name: >-
  The Supreme Court Database (SCDB), 1791-present, case- and justice-level with
  the code book
size_categories:
  - 100K<n<1M

Supreme Court Database (SCDB)

Every U.S. Supreme Court decision 1791-present from Washington University's Supreme Court Database: the modern (1946-, release 2026_01) and Legacy (1791-1945) databases at the case and justice level, all four units of analysis, plus the complete online code book as a codes table so every integer code decodes in SQL. v_cases is pre-decoded. 677,452 rows across 9 tables.

Table	Description	Row Count	Column Count
`justice_votes`	One row per justice per case, 1791-present: the case variables plus the justice's vote, opinion, direction, majority and agreement	256,469	62
`justice_votes_vote`	Justice-level rows of cases_vote (1946-present)	126,237	62
`justice_votes_legal_provision`	Justice-level rows of cases_legal_provision (1946-present)	125,571	62
`justice_votes_docket`	Justice-level rows of cases_docket (1946-present)	97,918	62
`cases`	One row per Supreme Court case by citation, 1791-present (modern 1946+ release 2026_01 plus the Legacy database): dates, citations, parties, origin, issue, decision direction, disposition, vote counts, opinion writer	29,270	54
`cases_vote`	Modern cases split one row per issue/vote split (1946-present)	14,090	54
`cases_legal_provision`	Modern cases split one row per legal provision considered (1946-present)	14,016	54
`cases_docket`	Modern cases split one row per docket number consolidated in the decision (1946-present)	10,932	54
`codes`	SCDB online code book: the label for every coded value (variable, code, label; justices carry their service dates in extra)	2,949	4

Query it remotely

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

SELECT term, decision_direction, COUNT(*) AS n
FROM scdb.v_cases WHERE term >= 2015 GROUP BY 1, 2 ORDER BY 1, 2;

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

Build pipeline and caveats: https://github.com/ian-nason/scdb-database Source: http://scdb.wustl.edu (Spaeth, Epstein et al.; CC BY-NC 3.0 - cite the SCDB).