judicial_common_space

Version
Status
Documentation
Tools
Compatibility
Stats

---

What is judicial_common_space?

judicial_common_space computes the Judicial Common Space (JCS) score of every appointment of a federal judge who has served since 1953 (the first year of the original study, or any year you choose), following Epstein, Martin, Segal, and Westerland, "The Judicial Common Space," 23 J.L. Econ. & Org. 303 (2007). It saves them in a spreadsheet with each judge's name, the Federal Judicial Center's two numbers for the judge (nid and jid), and the judge's score for each appointment, and checks them against the published scores.

Why use judicial_common_space?

This package:

• Scores every Article III appointment: district courts, courts of appeals, and the Supreme Court.
• Uses the current sources, downloaded when needed: the Federal Judicial Center's Biographical Directory of Article III Federal Judges, Voteview's DW-NOMINATE scores, the congress-legislators dates of senators' service, the Martin-Quinn scores, and Wikipedia's articles on the courts of appeals (for circuit judges' duty stations).
• Shows its work: each row says which state the judge was appointed from, how that was found, and which senators (or the president) the score came from.
• Checks itself against the published JCS scores of circuit judges and justices and against Christina Boyd's scores of district judges, which are made the same way.

How the scores are made

The paper scores the lower courts by the rule of Giles, Hettinger, and Peppers (2001), which follows the norm of senatorial courtesy. If a judge is appointed from a state where the president and at least one senator are of the same party, the judge gets that senator's NOMINATE score, or the mean of the senators' scores. Otherwise the judge gets the president's score. Justices get their Martin-Quinn score in each term, put on the same scale by the paper's arctangent transformation: 2/π · arctan(b0 + b1 · MQ).

The package applies the rule this way:

Step	How
Appointment	Each row of the center's file of federal judicial service. The senators and the president are those of the nomination (or of the recess appointment, if there was no nomination). A judge renominated by a new president after a recess appointment (as Roger Gregory was) is scored by the president who nominated the judge to the Senate. A judge moved to a court without an appointment (as the Fifth Circuit's judges were to the Eleventh in 1981) keeps the score of the appointment before.
Home state	A district judge's comes from the court's name. A circuit judge's is the state of a district court in the circuit that the judge served on before; or else of the job the judge held when appointed (or the most recent before), as the center's record of the judge's career names it; or else of the judge's duty station, as Wikipedia's article on the court gives it; or else the judge's birth state, if it is in the circuit. The duty station comes after the career because it is where the judge sat, which is sometimes not where the judge came from, but it agrees with the state from the career for 98.5% of the judges who have both. (The state of the judge whom the appointee replaced, from the center's seat numbers, made the matches worse: a seat changes states about one time in five.) Judges of the District of Columbia Circuit, the Federal Circuit, and the other national courts get the president's score, as they do in the published scores.
Senators	By default, every senator from the state who sat in the Congress of the appointment and belonged to the president's party, as the published scores count them (--senators congress). With --senators date, only the senators who held the seats on the day, by their party on that day, which is the rule as the paper states it and closer to Boyd's district scores.
Scores	Voteview's nominate_dim1 for senators and presidents (one score for each member's whole service).
Justices	The arctangent transformation with the coefficients of the 2024 release of the JCS scores (b0 = 0.1934, b1 = 0.6025), recovered from its 782 justice-terms (r = 0.9998), because both scales have changed since the paper estimated b0 = −0.1736 and b1 = 0.461 from the scores of 2005 (--transformation paper uses those). A justice's appointment gets the score of its first term, and the "Justices" sheet has every term.

How well they match the published scores

With the default settings, against the JCS 2024 release and Boyd's district court scores (through the 118^th Congress):

Check	Published	Found	Same (within 0.0005)	Same or within 0.02	Correlation
Circuit judges (JCS 2024)	705	703	88.3%	96.9%	0.998
District judges (Boyd)	2,615	2,615	83.6%	95.4%	0.998
Justice-terms (JCS 2024)	782	782	24.9%	99.4%	1.000

With --senators date, the district judges match Boyd's scores 99.2% of the time (within 0.02) and the circuit judges match the JCS scores 93.2% of the time, so the two published sources count senators differently.

Most differences under 0.02 are not errors. Voteview re-estimates the score of every member who is still serving whenever it adds a Congress, so judges whose senators are still in office (Fetterman, Rick Scott, Hickenlooper, and others) drift a little from scores published with an older file. The same is true of recent justices in the Martin-Quinn scores. The larger differences that remain are judges whom the published file places in a state that neither the center's records nor Wikipedia show (such as Richard Cudahy in Illinois and Stephen Trott in Idaho), a few D.C. Circuit judges and the Federal Circuit judges who came from the Court of Customs and Patent Appeals, which the published file scores differently, and some possible slips in the published files. The spreadsheet's check sheets list every comparison, and the next section explains every one that differs by more than 0.05.

Where they differ

These are every judge and justice whose computed score differs from the published one by more than 0.05, with the default settings. (The 689 other circuit judges, 2,529 district judges, and 780 justice-terms are within 0.05 of the published scores.)

Circuit judges and Epstein et al.

Judge	Circuit	Published	Computed	Why
Giles Rich	Fed.	0.692	0.281	Eisenhower appointed him to the Court of Customs and Patent Appeals, whose judges were moved to the new Federal Circuit in 1982. The published score is Reagan's, as if he were appointed then; the computed one is Eisenhower's.
John Nalbandian	6th	0.648	0.371	The computed score is Ohio's Rob Portman's (Nalbandian practiced in Cincinnati, and Wikipedia gives Ohio as his duty station). The published one matches no Ohio senator and is the same as the two Florida judges' of 2019.
Richard Cudahy	7th	−0.333	−0.583	The published score is Illinois's Adlai Stevenson III's. Cudahy's career and duty station were in Wisconsin.
Stephen Trott	9th	0.590	0.352	The published score is Idaho's two senators', where Trott later sat. He came from California.
Jay Bybee	9th	0.693	0.508	The published score is George W. Bush's. Bybee taught in Nevada, which is also his duty station, and Nevada's John Ensign was a Republican.
Morris Arnold	8th	0.381	0.557	Arkansas's two senators were Democrats, so the computed score is the first President Bush's. The published one matches only Arizona's John McCain.
Karen Henderson	D.C.	0.391	0.557	The published score is South Carolina's Strom Thurmond's, where she had been a district judge. The published file gives every other D.C. Circuit judge who came from a district court the president's score.
Luther Swygert	7th	−0.580	−0.420	He came from Indiana's district court, so the computed score is Indiana's Vance Hartke's. The published one is Illinois's Paul Douglas's.
James Browning	9th	−0.461	−0.311	The published score is Montana's two senators', where he was born. The computed one is Washington's, from his last job in the circuit before he worked in Washington, D.C.
Howard Markey	Fed.	0.692	0.552	As with Rich: Reagan's score in the published file, and Nixon's, who appointed him, here.
John Hartigan	1st	−0.299	−0.391	The published score is the mean of the two senators who held Rhode Island's seats when he was nominated. The computed one counts all four who sat in that Congress (--senators date gives the published score).
H. Nathan Swaim	7th	−0.447	−0.374	The published score is Illinois's two senators'. He came from Indiana, whose senators were Republicans, so the computed score is Truman's.
Cindy Chung	3rd	−0.313	−0.245	The published score is Bob Casey's alone; John Fetterman, who joined the Senate in 2023, was apparently not in the file it was made from.
Carl McGowan	D.C.	−0.405	−0.472	The published score matches no senators; the computed one is Kennedy's, as for other D.C. Circuit judges.

District judges and Boyd

75 of the 86 district judges who differ by more than 0.05 differ only because of how senators are counted. By default, the score counts every senator who sat in the Congress of the nomination; Boyd counts the senators who held the seats on the day, as --senators date does, which gives every one of these judges Boyd's score:

State	Nominated	Judges	Boyd	Computed	Senators counted who did not hold the seats that day
Alabama	1938	1	−0.275	−0.336	Hugo Black (left for the Supreme Court in 1937)
Alabama	1945	1	−0.275	−0.177	John Sparkman and George Swift (joined in 1946)
Alabama	2018	5	0.446	0.518	Jeff Sessions and Luther Strange (left in 2017 and January 2018)
Arkansas	1937	1	−0.302	−0.228	John E. Miller (joined after the nomination)
Arkansas	1977	1	−0.173	−0.246	Kaneaster Hodges (joined after the nomination)
Arkansas	1978	1	−0.378	−0.246	John McClellan (died in 1977)
California	1950	1	−0.374	−0.263	Sheridan Downey (resigned in 1950), so Boyd has Truman's score
California	1964	1	−0.462	−0.341	Clair Engle (died in 1964)
California	2021–2022	12	−0.345	−0.473	Kamala Harris (left in January 2021)
California	2024	9	−0.435	−0.378	Dianne Feinstein (died in 2023) and Adam Schiff (joined in December 2024)
Colorado	1995–1996	2	−0.438	−0.141	Ben Nighthorse Campbell (a Republican from March 1995), so Boyd has Clinton's score
Georgia	1999	1	−0.220	−0.037	Zell Miller (joined in 2000)
Indiana	1943	1	−0.115	−0.248	Samuel D. Jackson (joined in 1944)
Kentucky	1939	1	−0.384	−0.318	Happy Chandler (joined after the nomination)
Kentucky	1946	1	−0.375	−0.281	Happy Chandler (resigned in 1945)
Massachusetts	2014	4	−0.631	−0.519	John Kerry and Mo Cowan (left in 2013)
Missouri	1945	1	−0.258	−0.309	Harry Truman (left to be vice president)
New York	1937–1938	2	−0.243	−0.311	James Mead (joined in 1938)
New York	1949	2	−0.458	−0.558	Herbert Lehman (joined in November 1949)
New York	1950	5	−0.658	−0.558	Robert Wagner (resigned in 1949)
North Carolina	1949	1	−0.020	−0.167	Frank Graham and Willis Smith (joined after the nomination)
North Carolina	1985	1	0.617	0.501	James Broyhill (joined in 1986)
Oregon	1937	1	−0.368	−0.490	Alfred Reames (joined in 1938), so Boyd has Roosevelt's score
Rhode Island	1950	1	−0.299	−0.391	J. Howard McGrath and Edward Leahy (left the Senate before the nomination; Leahy is the judge)
South Carolina	1941	2	−0.212	−0.297	James Byrnes (left for the Supreme Court in 1941)
South Carolina	1966	1	−0.031	−0.099	Olin Johnston (died in 1965)
Texas	1942	2	0.018	−0.106	Morris Sheppard (died in 1941)
Texas	1962	4	−0.416	−0.155	William Blakley (left in 1961)
Texas	1993–1994	7	−0.438	−0.117	Bob Krueger (left in June 1993), so Boyd has Clinton's score
Washington	1939–1940	2	−0.149	−0.243	Monrad Wallgren (joined in December 1940)

The other 11 differ for reasons that seem to lie in Boyd's file:

Judges	Boyd	Computed	Why
James Connell (Ohio, 1954)	0.164	0.446	Boyd's score is the mean of John Bricker's and Thomas Burke's, but Burke was a Democrat and Eisenhower a Republican.
John Cannella, Charles Tenney, Bernard Moynahan (New York and Kentucky, 1963), and Orrin Judd, Morris Lasker, Anthony Travia (New York, 1967–1968)	−0.337	−0.472	Boyd gives all six Lyndon Johnson's score. The first three were nominated by Kennedy, before his death; for the others, the senator of Johnson's party was Robert Kennedy, whose score is −0.472.
Cyrus Tavares (Hawaii, 1961)	0.281	0.165	Boyd gives Eisenhower's score, leaving out Hiram Fong, a Republican.
Leonard Stark (Delaware, 2010)	−0.353	−0.281	Boyd's score is the mean of Ted Kaufman's and Joe Biden's, who had left the Senate to be vice president.
Edmond Chang (Illinois, 2010)	−0.345	−0.403	Boyd leaves out Roland Burris, who held a seat until November 2010.
James Allred (Texas, 1939)	−0.219	−0.275	Unexplained: Boyd's score matches no Texas senators.

Justices and Epstein et al.

Justice	Term	Published	Computed	Why
Ketanji Brown Jackson	2022	−0.444	−0.577	Her first term. The current Martin-Quinn scores, which include her later terms, estimate it again.
Clarence Thomas	2022	0.648	0.701	The same re-estimate of the 2022 term.

Using the published scores

--use-published (off by default) gives a judge or a justice's term the published score wherever it differs from the computed one by more than 0.05 (or --published-threshold): Epstein et al.'s for circuit judges and justices, and Boyd's for district judges. The Scores and Justices sheets then say where each score came from (jcs_source) and keep the computed score beside it (computed_jcs); the check sheets still compare the computed scores. With the default settings, it replaces the scores of 95 appointments since 1953 (80 district judges, 14 circuit judges, and Justice Jackson's appointment, from her first term) and of two justice-terms; with --senators date, it replaces 46 (13 district judges, 32 circuit judges, and Justice Jackson's).

Use it to reproduce the published scores, as when a study must line up with earlier work that used them. It is off by default because:

• The two sources count senators differently (Boyd by the day, Epstein et al. by the Congress), so taking both mixes their rules: with either setting of --senators, one of them is overruled.
• Some of the published scores that it would take look like slips, such as John Nalbandian's, Morris Arnold's, James Connell's, and the six New York and Kentucky judges'.
• Only judges in the published files can be replaced, so judges appointed in the same circumstances can end up with scores made by different rules.
• The published scores were made with older Voteview files, so they lag behind the current scores of senators who are still serving.

Getting started

Requirements

• Python 3.11 or later
• An internet connection, the first time, to download the source files (about 25 megabytes)

Installation

To install judicial_common_space, use pip:

pip install judicial-common-space

Usage

From the command line:

judicial-common-space --output judicial_common_space.xlsx

Or from Python:

import judicial_common_space

judicial_common_space.run('judicial_common_space.xlsx', start_year = 1953)

Option	Default	Description
--output	judicial_common_space.xlsx	Where to save the spreadsheet.
--start-year	1953	Include judges who served on a court in this year or later, and the justices' terms from the one sitting in January of that year. The default is the first year of the original study, which covered 1953 through 2000; the published scores now run through 2024, and these through the latest data.
--senators	congress	Which senators count: congress (as the published scores do) or date (those holding the seats on the day).
--transformation	jcs2024	How Martin-Quinn scores are put on the scale: jcs2024 or paper.
--martin-quinn	the data folder's	The Martin-Quinn justices.csv, if it cannot be downloaded. The Martin-Quinn site checks for automated browsers, so if the download fails, save the file from its page and pass its path (or put it in the data folder's politics folder).
--data-folder	data/courts beside the code folder	Where the downloads are kept: the same folders that courtpy uses, so the projects share them. Set JUDICIAL_COMMON_SPACE_DATA_DIR to change it for good.
--overwrite	off	Download the files again, as when the center has added judges or Voteview has added votes.
--no-download	off	Use only the files that are already there.
--use-published	off	Use the published score (Epstein et al.'s for circuit judges and justices, Boyd's for district judges) where it differs from the computed one by more than the threshold (see Using the published scores).
--published-threshold	0.05	The largest difference from a published score that keeps the computed score, with --use-published.
--no-check	off	Do not compare the scores with the published scores.

The spreadsheet has these sheets:

Sheet	Contents
Scores	A row for each appointment: judge, nid, jid, the court, the president and party, the dates, the Congress, home_state (a postal code) and state_source (how it was found), duty_station (Wikipedia's state for a circuit judge), method and senators (how the score was found), president_score, computed_jcs, jcs, and jcs_source ("computed", or with --use-published, "Epstein et al." or "Boyd").
Justices	Each justice's Martin-Quinn and JCS scores in each term, with computed_jcs and jcs_source.
Verification	The agreement with each set of published scores.
Circuit check, District check, Justice check	Every comparison, with both scores and how the computed one was found.
About	The method, the sources, and the settings.

Contributing

Contributors are always welcome. Feel free to grab an issue to work on or make a suggested improvement. If you wish to contribute, please read the Contribution Guide and Code of Conduct.

Similar Projects

• courtpy adds the published JCS scores (and the other measures here) to court cases, and shares this package's downloads.
• The JCS scores of Lee Epstein, Andrew D. Martin, Jeffrey A. Segal, and Chad Westerland, for the judges of the courts of appeals and the justices.
• Christina L. Boyd's Federal District Court Judge Ideology Data, which scores district judges by the same rule.

Acknowledgments

The scores rest on the work and data of Epstein, Martin, Segal, and Westerland; Micheal W. Giles, Virginia A. Hettinger, and Todd Peppers; Keith T. Poole, Howard Rosenthal, and Voteview; Andrew D. Martin and Kevin M. Quinn; the Federal Judicial Center; the contributors to congress-legislators and to Wikipedia (whose text is available under the Creative Commons Attribution-ShareAlike License); and Christina L. Boyd. Please cite them when you use the scores.

This project was generated from @WithPrecedent's template.

Citation

If you use judicial_common_space in your research, definitely cite these works:

@misc{boyd_district_2015,
  author = {Boyd, Christina L.},
  title  = {Federal District Court Judge Ideology Data},
  year   = {2015},
  url    = {http://cLboyd.net}
}

@article{epstein_jcs_2007,
  author  = {Epstein, Lee and Martin, Andrew D. and Segal, Jeffrey A. and Westerland, Chad},
  title   = {The Judicial Common Space},
  journal = {Journal of Law, Economics, and Organization},
  volume  = {23},
  number  = {2},
  pages   = {303--325},
  year    = {2007}
}

@article{giles_picking_2001,
  author  = {Giles, Micheal W. and Hettinger, Virginia A. and Peppers, Todd},
  title   = {Picking Federal Judges: A Note on Policy and Partisan Selection Agendas},
  journal = {Political Research Quarterly},
  volume  = {54},
  number  = {3},
  pages   = {623--641},
  year    = {2001}
}

And, although my contributions to the JCS project are relatively minor, feel free to cite the repository as well:

@software{yung_judicial_common_space,
  author  = {Yung, Corey Rayburn},
  title   = {judicial_common_space: Judicial Common Space scores for every Article III appointment},
  version = {0.1.0},
  year    = {2026},
  url     = {https://github.com/WithPrecedent/judicial_common_space},
  license = {Apache-2.0}
}

License

Use of this repository is authorized under the Apache Software License 2.0.