The Great SCOTUS Data Wrangle, Part 4: Background and Chronological Variables

Who’s ready to delve deeper into the Supreme Court Database?

Anyone?

… Well don’t everyone speak up at once!

Last time I focused on fixing up the identification variables in the Supreme Court Database. Thanks to a couple trademark digressions on case names, this wound up being the longest post I’ve made in this series.¹ I should probably split that content out into several posts.

Today I’ll pick up where we left off looking over the SCDB variables of interest for our upcoming posts. I’m analyzing variables by category and in the order that they appear in the SCDB’s online documentation. The variables in this post will include a handful of the background and chronological variables that matter to us.

import os
import re
from pathlib import Path

import git
import joblib
import numpy as np
import pandas as pd
import requests
from IPython.display import HTML, Latex, display_html
from tqdm.notebook import tqdm

GLOBAL_DATA_CACHE = Path('~/.cache/python/data/').expanduser().absolute()
GLOBAL_DATA_CACHE.mkdir(parents=True, exist_ok=True)

GLOBAL_JOBLIB_MEMORY = joblib.Memory(GLOBAL_DATA_CACHE, verbose=0)

REPO_ROOT = Path(git.Repo('.', search_parent_directories=True).working_tree_dir).absolute()
DATA_PATH = REPO_ROOT / 'data'

os.environ['CAP_AUTH_TOKEN'] = (REPO_ROOT / 'secrets' / 'caselaw_access_project_auth_token.txt').read_text()

case_decisions = (
    pd.read_feather(
          DATA_PATH / 'processed' / 'scdb'
                    / 'case_decisions_from_2021-07-05_post.feather'))

Towards More Pythonic, Pandorable Data

Alright, before proceeding any further, a couple modifications to the case_decisions DataFrame are a long time coming. First and foremost, the camel case columns have got to go.

def camel_to_snek(camel_cased):
    camel_hump_not_after_underscore = '((?!^)(?<!_)[A-Z][a-z]+|(?<=[a-z0-9])[A-Z])'
    return re.sub(camel_hump_not_after_underscore, r'_\1', camel_cased).lower()


case_decisions.columns = case_decisions.columns.map(camel_to_snek)
case_decisions.columns[:10].tolist()

['case_id',
 'docket_id',
 'case_issues_id',
 'vote_id',
 'date_decision',
 'decision_type',
 'us_cite',
 'sct_cite',
 'led_cite',
 'lexis_cite']

Ah that’s much better. All credit for that lovely regular expression should go to SO user cco for his comment on this answer.

I may also make the column names more verbose in the future by getting rid of (or at least being consistent with) abbreviations, but this is a much better place to be. (Although I am upset every time I’m reminded that there are simultaneously date_argument and date_rearg columns.) Ridding myself of camel case has brought me enough joy for the time being.

Second, let’s assign a meaningful index. I confirmed in my last post that the column formerly known as caseId does what it says on the tin—provide a unique identifier for each case—so it’s the natural choice.

case_decisions = case_decisions.set_index('case_id', verify_integrity=True)

Processing Background Variables

Here we begin to get into the real metadata for each case, including the case names we addressed last post, a broad categorization of the types of petitioners and respondents in each case, jurisdictional information, and aspects of the outcomes of lower court decisions in each case. While there are some interesting questions that can be asked regarding case outcomes in relation to lower court decisions and actions, I’ll be ignoring lower court-related columns and in doing so restrict to about half of these columns.

background_variables_of_interest = [
    'petitioner', 'petitioner_state',
    'respondent', 'respondent_state',
    'jurisdiction', 'cert_reason'
]

IT’S PARTY TIME: petitioners and respondents

Out of the case_names fire and into the frying pan! Or out of the frying pan and into the room surrounding said frying pan … which is much cooler. The point is it isn’t as hot here. Maybe this isn’t the best choice of idioms.

The petitioner feature, astonishingly, describes the petitioner (a.k.a. the appellant) of a case, meaning the party that appealed a lower court’s decision to the Supreme Court. The other party to the case is called the “respondent” (a.k.a. the appellee) and is referenced in, you guessed it, the respondent feature. Rather than rather uselessly storing the name or specific identity of the appellant and appellee, these fields contain short descriptions indicating how SCOTUS, in its opinions, described the parties in relation to the substantive matters of their case. The SCDB’s methodology for labeling petitioners and respondents is described in more detail in the petitioner documentation alongside a complete list of values for this variable.

There are a couple interesting caveats in their labeling process that are worth noting. First, when a state is party to a case, it is described in these variables as 'State' instead of by name. To obtain which state is the party in question, we also need to use the petitioner_state and respondent_state variables. These two features are sparsely populated, with values provided only for litigants that are states or agents of states (including municipal governments, courts, etc.). Also, when a federal court, judicial district, or judge is party to a case, their state is entered as 'United States'.

The other interesting interaction between the party variables and the remainder of the SCDB has to do with specificity:

Where a choice of identifications exists that which provides information not provided by the legal provision or the issue is chosen. E.g., a federal taxpayer or an attorney accused of a crime as taxpayer or attorney rather than accused person, particularly if neither the lawType nor the Issue variable identifies the case as a tax matter or one involving an attorney.

— SCDB Petitioner Documentation

In other words, when new cases are entered into the SCDB with two equally-appropriate values for a party variable, the value that is chosen is the one that can’t be inferred from issue and law_type. (What happens when there are more than two possibilities isn’t clearly spelled out, probably because three or more-way ties never happen.) It would be interesting to get a sense of how large an effect this has. I imagine these ambiguous situations aren’t that common, but who’s to say? Either way, a thorough analysis involving either petitioner or respondent requires consideration of issue and law_type.

Cleanliness-wise, we’ve already found that both of the party features are in great shape.

display(case_decisions[['petitioner', 'respondent']].describe())

print('Number of Null Petitioners:', case_decisions.petitioner.isna().sum())
print('Number of Null Respondents:', case_decisions.respondent.isna().sum())

case_decisions[
    case_decisions[['petitioner', 'respondent']].isna().any(axis='columns')
].T

	petitioner	respondent
count	28873	28871
unique	273	259
top	United States	United States
freq	2509	3118

Number of Null Petitioners: 2
Number of Null Respondents: 4

case_id	1933-168	1939-143	1941-163	1949-112
docket_id	1933-168-01	1939-143-01	1941-163-01	1949-112-01
case_issues_id	1933-168-01-01	1939-143-01-01	1941-163-01-01	1949-112-01-01
vote_id	1933-168-01-01-01	1939-143-01-01-01	1941-163-01-01-01	1949-112-01-01-01
date_decision	1933-10-23 00:00:00	1939-12-11 00:00:00	1942-03-02 00:00:00	1950-04-24 00:00:00
decision_type	equally divided vote	equally divided vote	equally divided vote	equally divided vote
us_cite	290 U.S. 591	308 U.S. 523	315 U.S. 784	339 U.S. 940
sct_cite	54 S. Ct. 94	60 S. Ct. 293	62 S. Ct. 793	70 S. Ct. 793
led_cite	78 L. Ed. 521	84 L. Ed. 443	86 L. Ed. 1190	94 L. Ed. 2d 1356
lexis_cite	1933 U.S. LEXIS 1045	1939 U.S. LEXIS 1131	1942 U.S. LEXIS 873	1950 U.S. LEXIS 2598
term	1933	1939	1941	1949
natural_court	Hughes 3	Hughes 7	Stone 2	Vinson 3
chief	Hughes	Hughes	Stone	Vinson
docket	18	317	529	490
case_name	HELVERING, COMMISSIONER OF INTERNAL REVENUE, v...	HELVERING, COMMISSIONER OF INTERNAL REVENUE, v...	HOLLAND v. LOWELL SUN CO.	UNITED STATES v. COTTON VALLEY OPERATORS COMMI...
date_argument	1933-10-13 00:00:00	1939-12-05 00:00:00	1942-02-04 00:00:00	1950-04-18 00:00:00
date_rearg	NaT	NaT	NaT	NaT
petitioner	Internal Revenue Service	<NA>	<NA>	United States
petitioner_state	<NA>	<NA>	<NA>	<NA>
respondent	<NA>	<NA>	<NA>	<NA>
respondent_state	<NA>	<NA>	<NA>	<NA>
jurisdiction	cert	<NA>	cert	appeal
admin_action	<NA>	<NA>	<NA>	<NA>
admin_action_state	<NA>	<NA>	<NA>	<NA>
three_judge_fdc	no mentionof 3-judge ct	NaN	NaN	NaN
case_origin	<NA>	<NA>	<NA>	La. West. U.S. Dist. Ct.
case_origin_state	<NA>	<NA>	<NA>	<NA>
case_source	U.S. Ct. App., First Cir.	<NA>	U.S. Ct. App., First Cir.	La. West. U.S. Dist. Ct.
case_source_state	<NA>	<NA>	<NA>	<NA>
lc_disagreement	no mention of dissent	NaN	NaN	no mention of dissent
cert_reason	no reason given	<NA>	<NA>	no cert
lc_disposition	NaN	NaN	NaN	NaN
lc_disposition_direction	NaN	NaN	NaN	NaN
declaration_uncon	no unconstitutional	no unconstitutional	no unconstitutional	no unconstitutional
case_disposition	affirmed	affirmed	affirmed	affirmed
case_disposition_unusual	no unusual disposition	no unusual disposition	no unusual disposition	no unusual disposition
party_winning	petitioner lost	petitioner lost	petitioner lost	NaN
precedent_alteration	precedent unaltered	precedent unaltered	precedent unaltered	precedent unaltered
vote_unclear	vote clearly specified	vote clearly specified	vote clearly specified	vote clearly specified
issue	<NA>	<NA>	<NA>	<NA>
issue_area	NaN	NaN	NaN	NaN
decision_direction	NaN	NaN	NaN	NaN
decision_direction_dissent	NaN	NaN	NaN	NaN
authority_decision1	NaN	NaN	NaN	NaN
authority_decision2	NaN	NaN	NaN	NaN
law_type	NaN	NaN	NaN	NaN
law_supp	<NA>	<NA>	<NA>	<NA>
law_minor	<NA>	<NA>	<NA>	<NA>
maj_opin_writer	<NA>	<NA>	<NA>	<NA>
maj_opin_assigner	<NA>	<NA>	<NA>	<NA>
split_vote	1st vote on issue/provision	1st vote on issue/provision	1st vote on issue/provision	1st vote on issue/provision
maj_votes	4	4	4	4
min_votes	4	4	4	4

Ah, of course these are “equally divided vote” cases! We’ll talk about these more when looking at cert_reasons below, but the opinions for this category of cases tend to follow an absolutely minimal template, from which it’s next to impossible to obtain information about the substance of a case. These cases are no exception, with the opinions of the Court providing practically zero information. As you probably expect, summary opinions like these are noticeably less complete than other SCDB records, presumably due to the greater difficulty encountered by the researcher attempting to dig up case information.

Now, when I first noticed there were so few cases, I jumped head first into identifying the correct values of these and related fields, and this was a great excuse to use the Caselaw Access Project’s API.

@GLOBAL_JOBLIB_MEMORY.cache(ignore=['auth_token'])
def fetch_cap_case_data(scdb_id, with_opinions=True,
                        auth_token=os.getenv('CAP_AUTH_TOKEN')):
    request_kwargs = {}
    if with_opinions and auth_token is None:
        raise ValueError(
            'An auth_token is required when fetching case opinions')
    elif auth_token is not None:
        request_kwargs['headers'] = {'Authorization': f'Token {auth_token}'}

    case_query_configuration = {'cite': f'SCDB{scdb_id}'}
    if with_opinions:
        case_query_configuration |= {'full_case': 'true', 'body_format': 'text'}

    case_query = to_url_query(**case_query_configuration)
    results = requests.get(
        f'https://api.case.law/v1/cases/{case_query}',
        **request_kwargs
    ).json()['results']

    if results:
        return results[0] if len(results) == 1 else results


def to_url_query(**kwargs):
    if not kwargs:
        return ''
    query = '&'.join(f'{key}={value}' for key, value in kwargs.items())
    return f'?{query}'

This fetch_cap_case_data function is a more fleshed out version of one by the same name used in the previous post to corroborate SCDB case names with those provided by the CAP. It requests all of the data associated to a SCOTUS case by the Caselaw Access Project, including each published opinion as plain text. If you’re new to Joblib, the decorator ensures that responses for a given case ID are cached locally. This is a win-win for the Caselaw Access Project and the user, reducing strain on CAP servers, keeping subsequent notebook runs snappy, and helping to avoid CAP API quotas.

Let’s grab the CAP data for SCDB records missing parties.

null_party_scdb_ids = case_decisions.loc[
    case_decisions[['petitioner', 'respondent']].isna().any(axis='columns')
].index

null_party_case_cap_data = [fetch_cap_case_data(null_party_scdb_id)
                            for null_party_scdb_id in null_party_scdb_ids]

for case_data in null_party_case_cap_data:
    print('Case:', case_data['name'])
    print(case_data['casebody']['data']['opinions'][0]['text'], end='\n\n')

Case: Helvering, Commissioner of Internal Revenue, v. U.S. Refractories Corp.
Per Curiam:
Decree affirmed by an equally divided Court. Mr. Justice Roberts took no part in the consideration or decision of this case.

Case: Helvering, Commissioner of Internal Revenue, v. Johnson
Per Curiam:
The judgment is affirmed by an equally divided Court.

Case: Holland, Administrator, Wage and Hour Division, U. S. Department of Labor, v. Lowell Sun Co.
Per Curiam:
The judgment is affirmed by an equally divided Court. Mr. Justice Murphy took no part in the consideration or decision of this case.

Case: United States v. Cotton Valley Operators Committee et al.
Appeal from the United States District Court for the Western District of Louisiana.
Per Curiam:
The judgment is affirmed by an equally divided Court.
Mr. Justice Clark took no part in the consideration or decision of this case.

As I said, there’s not much to be seen here, but between the CAP API, Google, and a journal article, we can quickly arrive at the correct values for the missing party fields, among others.

• Helvering v. U.S. Refractories Corp.
  • Petitioner: 'Internal Revenue Service'
  • Respondent: 'taxpayer'
• Helvering v. Johnson
  • Petitioner: 'Internal Revenue Service'
  • Respondent: 'taxpayer'
• Holland v. Lowell Sun Co.
  • Petitioner: 'Department of Labor'
  • Respondent: 'employer'
• United States v. Cotton Valley Operators Committee
  • Petitioner: 'United States'
  • Respondent: 'employer'
    • Something more descriptive like 'farmer' or some kind of manufacturer might be more suitable here.
  • Related Cases:
    • 75 F. Supp. 1 (W.D. La. 1948)
    • 77 F. Supp. 409 (W.D. La. 1948)

The trickiest of these to track down was Holland v. Lowell Sun Company. Based on this critical review from the time (JSTOR paywall warning), Holland v. Lowell Sun Company was summarily affirmed per curiam by the Court after their decision in Cudahy Packing Company v. Holland from the previous day.² The relationship between the parties and the issues and legal provisions at play are shared by these two cases, albeit with the roles of petitioner and respondent reversed, so we could crib the content in Cudahy in order to fill in Lowell.

Unfortunately a more careful reading of the “decision rules” in the petitioner documentation has given me pause.

Parties are identified by the labels given them in the opinion or judgment of the Court except where the Reports title a party as the “United States” or as a named state.

That sounds to me like no additional references, including lower court decisions (let alone law review articles), are referenced when populating the party variables. You might say I’m taking this too literally, but come on. We’re talking about a data source curated by legal scholars and lawyers. They’re about as precise with their definitions as us mathematicians. It’s also bolstered by the disproportionate rate of 'Unidentifiable' parties in 'equally divided vote' cases.

equally_divided_votes_count = (case_decisions.decision_type == 'equally divided vote').sum()

print('Equally divided vote prevalence:')
print(
    equally_divided_votes_count / case_decisions.shape[0],
    'or', equally_divided_votes_count, 'of', case_decisions.shape[0],
    'cases',
    end='\n\n'
)

cases_with_unidentifiable_parties = case_decisions[
    (case_decisions[['petitioner', 'respondent']] == 'Unidentifiable').any(axis='columns')
]
equally_divided_votes_count_given_unidentifiable_parties = (
    cases_with_unidentifiable_parties.decision_type == 'equally divided vote'
).sum()

print('Equally divided vote prevalence among cases with unidentifiable parties:')
print(
    equally_divided_votes_count_given_unidentifiable_parties
        / cases_with_unidentifiable_parties.shape[0],
    'or', equally_divided_votes_count_given_unidentifiable_parties,
    'of', cases_with_unidentifiable_parties.shape[0],
    'cases'
)

Equally divided vote prevalence:
0.006337662337662337 or 183 of 28875 cases

Equally divided vote prevalence among cases with unidentifiable parties:
0.09403669724770643 or 82 of 872 cases

So, nearly 45% of all equally divided vote cases have unidentifiable parties, and equally divided votes occur among cases with unidentifiable parties at nearly 15 fold their overall rate. I think it’s safe to say we should either avoid pulling in supplementary material or systematically overhaul the parties in these cases.

But hey, at least our work wasn’t entirely for not; we can at very least add the IRS as the petitioner in Helvering, Commissioner of Internal Revenue v. Johnson. Ok, maybe we could have seen that from the title of the case. Maybe. But now we feel really confident in that correction, right?

The optimist in me is struggling. Let’s call this a wash and move on.

possible_helvering_v_johnson_case_ids = case_decisions.loc[
    case_decisions.case_name == 'HELVERING, COMMISSIONER OF INTERNAL REVENUE, v. JOHNSON'
].index

assert len(possible_helvering_v_johnson_case_ids) == 1, (
    'Multiple Helvering v. Johnson cases exist in the SCDB.'
)

helvering_v_johnson_case_id = possible_helvering_v_johnson_case_ids[0]

case_decisions.loc[
    helvering_v_johnson_case_id,
    'petitioner'
] = 'Internal Revenue Service'

assert len(null_party_scdb_ids) == 4, 'Unexpected number of cases with missing party values'
case_decisions.loc[case_decisions.petitioner.isna(), 'petitioner'] = 'Unidentifiable'
case_decisions.loc[case_decisions.respondent.isna(), 'respondent'] = 'Unidentifiable'

petitioner_states and respondent_states

When a state, state agency, state employee, or other state-related entity is party to a case, the state in question is captured in a feature. I’m using “state” here rather generally, allowing for territories, interstate compacts, federal entities, and even the occasional free association³ in addition to the Big 50⁴. As always, you can refer to the documentation for a complete rundown.

These are the first sparse features we’ve encountered in the dataset, with non-state parties being encoded as NaNs. We are still able to check for unintentional missing values by looking for NaNs in records with state actors serving as petitioners or respondents.

state_actor = (
    pd.concat([case_decisions.loc[lambda df: df[f'{actor}_state'].notna(), actor]
               for actor in ['petitioner', 'respondent', 'admin_action']])
      .pipe(lambda series: set(series.unique()))
)

state_actor

{'State',
 'State Agency',
 'U.S. Shipping Board Emer. Fleet Corp.',
 'bank of the united states',
 'county government',
 'court or judicial district',
 'female govt. employee',
 'former govt. employee',
 'governmental employee',
 'governmental official',
 'judge',
 'local government',
 'local governmental unit',
 'local school district',
 'minority fem. govt. employ.',
 'minority govt. employee',
 'state college or university',
 'state commission or auth.',
 'state department or agency',
 'state education board',
 'state legislature',
 'state or U.S. supreme court',
 'state, local govt. taxpayer',
 'timber company'}

All of these seem about right: states, state agencies, judges, state legislatures, timber companies … wait what?

case_decisions.loc[
    np.logical_or.reduce([
        (case_decisions[actor] == 'timber company') & case_decisions[f'{actor}_state'].notna()
        for actor in ['petitioner', 'respondent', 'admin_action']
    ]),
    ['case_name', 'petitioner', 'petitioner_state', 'respondent', 'respondent_state',
     'admin_action', 'admin_action_state']
]

case_id	case_name	petitioner	petitioner_state	respondent	respondent_state	admin_action	admin_action_state
1899-111	LINDSAY AND PHELPS COMPANY v. MULLEN	timber company	Minnesota	state commission or auth.	Minnesota	State Agency	Minnesota
1930-165	ISAACS, TRUSTEE IN BANKRUPTCY OF THE ESTATE OF...	bankrupt person or business	<NA>	timber company	Arkansas	<NA>	<NA>

I’m guessing these are entry errors, not sightings of the lumber arms of Minnesota and Arkansas. Although, now that I write that, I’m kind of hoping to be wrong about that. I’d love to learn some really obscure history here, where for whatever reason Minnesota, Arkansas, and other states nationalized their lumber industries. (Is “nationalized” still the right term when it’s a state or local government? “State-ified” is clumsy, and “municipalitized” is bordering on criminal.)

lindsay_and_phelps_co_v_mullen = fetch_cap_case_data('1899-111', with_opinions=False)
isaacs_v_hobbs_tie_and_timber_co = fetch_cap_case_data('1930-165')

The Lindsay and Phelps Co. case is a real page turner. You can read the full text here:

lindsay_and_phelps_co_v_mullen['frontend_url']

'https://cite.case.law/us/176/126/'

Alternatively, you can head to the U.S. Report PDF to clarify any OCR ambiguities here:

lindsay_and_phelps_co_v_mullen['frontend_pdf_url']

'https://cite.case.law/pdf/5704883/Lindsay%20&%20Phelps%20Co.%20v.%20Mullen,%20176%20U.S.%20126,%2044%20L.%20Ed.%20400,%2020%20S.%20Ct.%20325%20(1900).pdf'

If you actually followed either of those links, you’ll know that by “page turner”, I meant “a slog rich with late 1800s logging terminology”. (You know you’re in for a wild ride when “replevin” appears in the first fifteen words of the case summary.) Fortunately, you’ll also see that the case gods have been merciful, clarifying the relationship between Lindsay and Phelps Co. and the state almost immediately.

On August 1, 1893, the plaintiff in error commenced its action of replevin against one of the defendants in error, John H. Mullen, to recover possession of a quantity of logs said to be of the value of \$15,000. Mullen answered, alleging that he was the surveyor general of logs and lumber for the fourth district of Minnesota; that as such surveyor general he had scaled and surveyed a large number of logs in a boom belonging to the Minnesota Boom Company, for which service he was entitled to fees amounting to the sum of \$11,088.92, and had seized these logs, under the statute giving him a lien, to enforce payment thereof, and praying for a return of the property, or, if that could not be had, for judgment for the sum of \$11,088.92, together with ten per cent, \$1108.89, costs of collection as provided by law, and interest. […] On these pleadings the case went to trial before the court without a jury. No special findings of fact were made, but only a general finding for defendants. A bill of exceptions was preserved, reciting the testimony, showing that at the close the plaintiff requested of the court the following declarations:

[…]

Second. That the defendants have not shown themselves entitled to any lien upon the plaintiff’s logs:

“a. Because the scale bills, Defendants’ Exhibits 3 and 4, are not evidence of the scaling of the logs therein described.

“b. Because it appears affirmatively that the said scale bills were not, nor were either of them, recorded in any book in the office of the surveyor general of that district.

“g. Because it appears that a very great proportion of the logs mentioned in these scale bills, defendants’ exhibits 3 and 4, were not the plaintiff’s logs, and that the work done was not done at the request of the plaintiff or anybody else.

“d. Because the pretended records of said scale bills were not in fact any record whatever.

“e. Because it does not appear that any of the log marks shown on defendants’ scale bills, Exhibits 3 and 4, were ever recorded in the office of the surveyor general of logs and lumber of the fourth lumber district of the State of Minnesota, in accordance with the provisions of title 3, of chapter 32, General Statutes of the State of Minnesota. […]”

If I’m reading this correctly, Lindsay and Phelps Company logs were in the Minnesota Boom Company’s boom when Mullen, a surveyor general for Minnesota, put a lien on all of the logs in said boom. That lien would be in effect until Mullen was paid for his surveying and scaling services. The Lindsay and Phelps Company requested its logs, seemingly as a third party caught in the cross-fire between the Minnesota Boom Company and the state. At least, that’s what this and the dissenting opinion read like to me, although I admit I was struggling to read through the majority opinion without my eyes glazing over. In any event Lindsay and Phelps Company is a private timber company, not a state actor. The state had also already taken action against the petitioner prior to this case reaching SCOTUS, so both the respondent* and the admin_action* values are accurate.

lindsay_and_phelps_v_mullen_case_id = '1899-111'

case_decisions.loc[
    lindsay_and_phelps_v_mullen_case_id,
    'petitioner_state'
] = pd.NA

Now let’s see what Isaacs v. Hobbs Tie & Timber Co. has in store for us. The full description of the petitioner in this case is “Isaacs, trustee in bankruptcy of the estate of Henrietta E. Cunningham, bankrupt”. A trip to Wikipedia reveals that a “trustee in bankruptcy” is an individual employed (or certified by) the Department of Justice and who has been tasked with managing the estate of a bankrupt individual or corporation during the bankruptcy process. While, yes, this person is assigned by the Department of Justice, they act on behalf of a bankrupt person or corporation, serving in some capacities as an intermediary between and in others as a replacement for the debtor with their creditors. While I suppose this means Isaacs could be considered a state or federal actor here if the state or federal laws that were in effect during 1930 were similar to the federal bankruptcy laws of the present, Isaacs’s party in the context of the case is classified as a “bankrupt person or business”. This sounds to me like a class that probably shouldn’t be considered related to the state.

Bankruptcy is, however, a common enough occurrence that we can probably look to other examples to see how this type of party is usually categorized.

with pd.option_context('display.max_rows', 200, 'display.max_colwidth', 100):
    display(
        case_decisions.loc[
            (
                case_decisions.case_name.str.contains('trustee in bankruptcy', case=False) &
                (
                    case_decisions.petitioner_state.notna() |
                    case_decisions.respondent_state.notna()
                )
            ),
            ['case_name', 'petitioner', 'petitioner_state', 'respondent', 'respondent_state']
        ].style.applymap(lambda value: 'color: purple' if value in {'bankrupt person or business',
                                                                    'agent, fiduciary, trustee'}
                                        else 'color: black')
    )

	case_name	petitioner	petitioner_state	respondent	respondent_state
case_id
1923-054	BOARD OF TRADE OF THE CITY OF CHICAGO et al. v. JOHNSON, TRUSTEE IN BANKRUPTCY OF HENDERSON	local government	Illinois	bankrupt person or business	<NA>
1923-195	PEOPLE OF THE STATE OF NEW YORK v. JERSAWIT, TRUSTEE IN BANKRUPTCY OF AJAX DRESS COMPANY, INC.	State	New York	bankrupt person or business	<NA>
1924-231	DAVIS, FEDERAL AGENT FOR CLAIMS DUE IN OPERATION OF ATLANTIC COAST LINE RAILROAD, v. PRINGLE, TRUSTEE IN BANKRUPTCY OF ESTATE OF BOYD CO., INC.	governmental official	United States	agent, fiduciary, trustee	<NA>
1929-080	U.S. SHIPPING BOARD MERCHANT FLEET CORPORATION v. HARWOOD, TRUSTEE IN BANKRUPTCY, et al.	U.S. Shipping Board Emer. Fleet Corp.	United States	bankrupt person or business	<NA>
1930-165	ISAACS, TRUSTEE IN BANKRUPTCY OF THE ESTATE OF HENRIETTA E. CUNNINGHAM, BANKRUPT, v. HOBBS TIE & TIMBER COMPANY	bankrupt person or business	<NA>	timber company	Arkansas
1932-166	NEW YORK v. IRVING TRUST CO., TRUSTEE IN BANKRUPTCY	State	New York	bankrupt person or business	<NA>
1934-010	SCHUMACHER, SHERIFF OF BUTLER COUNTY, OHIO, v. BEELER, TRUSTEE IN BANKRUPTCY	county government	Ohio	bankrupt person or business	<NA>
1935-056	CITY OF LINCOLN et al. v. RICKETTS, TRUSTEE IN BANKRUPTCY	local government	Nebraska	bankrupt person or business	<NA>
1940-075	CITY OF NEW YORK v. FEIRING, TRUSTEE IN BANKRUPTCY	local government	New York	bankrupt person or business	<NA>
1941-088	MEILINK, TRUSTEE IN BANKRUPTCY, v. UNEMPLOYMENT RESERVES COMMISSION OF CALIFORNIA	bankrupt person or business	<NA>	state commission or auth.	California
1948-032	GOGGIN, TRUSTEE IN BANKRUPTCY, v. DIVISION OF LABOR LAW ENFORCEMENT OF CALIFORNIA	bankrupt person or business	<NA>	state department or agency	California
1948-047	CITY OF NEW YORK v. SAPER, TRUSTEE IN BANKRUPTCY	local government	New York	bankrupt person or business	<NA>

None of the other cases involving a trustee in bankruptcy classify that party as a state actor. In fact, none of the cases in the entire database involving someone (or something) in financial trouble or an agent thereof classify those parties as agents of a state.

possibly_bankrupt = r'bankrupt person or business|agent, fiduciary, trustee'

print('Cases involving possibly-bankrupt entities:',
      case_decisions[['petitioner', 'respondent']]
          .apply(lambda party: party.str.contains(possibly_bankrupt))
          .any(axis='columns')
          .sum())

print('Possibly-bankrupt entities that are state actors:',
      ((case_decisions.petitioner.str.contains(possibly_bankrupt) &
        case_decisions.petitioner_state.notna()) |
       (case_decisions.respondent.str.contains(possibly_bankrupt)
        & case_decisions.respondent_state.notna())).sum())

Cases involving possibly-bankrupt entities: 2087
Possibly-bankrupt entities that are state actors: 0

I think it’s safe to say that this was a typo. For the sake of due diligence, I also skimmed the majority opinion and gained a little added confidence that neither the trustee nor the timber company are associated to Arkansas.

isaacs_v_hobbs_case_id = '1930-165'
case_decisions.loc[
    isaacs_v_hobbs_case_id,
    ['petitioner_state', 'respondent_state']
] = pd.NA

A Diversity of jurisdictions

While the modern Court is mostly occupied with writs of certiorari, there are a couple more handfuls of ways a case can reach it. The jurisdiction field captures how each case falls into the Court’s domain, be that a writ of cert., appeal, or error (in older cases) or, occasionally, original jurisdiction. Ties go to the writ, in the sense that the type of writ is used as the jurisdiction when there are multiple ways that the Court takes jurisdiction over a case.

As always, see the documentation for more details about the data entry decision process. The various ways the Court gains jurisdiction over a case are laid out in Article III of the U.S. Constitution. The annotated version of the Constitution maintained by the National Constitution Center presents Article III in a pretty format with links to lengthy discussions of its meaning, interpretation, and modification by the 11th Amendment. Article III also has received a lot of love on Wikipedia and the related concept or jurisdiction stripping, which I really hope to see put to good use in the future.

For even more coverage of the Court’s original jurisdiction, check out the Federal Judicial Center’s history of its original jurisdiction. You can once again turn to the Wikipedia coverage of the topic, as well, which is more concise but with a couple interesting references to read through.

case_decisions.jurisdiction.value_counts(dropna=False)

cert                       10341
appeal                      8619
writ of error               8317
certification                671
original                     399
mandamus                     234
writ of habeas corpus         96
rehearing or reargument       73
unspecified, other            54
prohibition                   47
injunction                     8
stay                           7
docketing fee                  6
NaN                            2
bail                           1
Name: jurisdiction, dtype: Int64

… Ok I’m sorry, but I must know what that one bail case is about.

case_decisions.loc[case_decisions.jurisdiction == 'bail', 'case_name'].iloc[0]

'DAVIDSON AND ANOTHER, PLAINTIFFS IN ERROR, v. TAYLOR, DEFENDANT IN ERROR'

I don’t regret sating my curiosity. I do regret taking the time to parse the rather impenetrable Latin⁵ in this single paragraph opinion. The best portion of the opinion is its concluding sentence:

The plea is, therefore, bad; and the judgment is affirmed, with six per centum damages, and costs.

I’d love to see “this plea is bad” make a comeback.

And with that out of the way, time for more null value analysis.

with pd.option_context('display.max_colwidth', 100):
    display(
        case_decisions.loc[
            case_decisions.jurisdiction.isna(),
            ['case_name']
        ]
    )

case_id	case_name
1939-143	HELVERING, COMMISSIONER OF INTERNAL REVENUE, v. JOHNSON
2014-076	RAUL LOPEZ, WARDEN v. MARVIN VERNIS SMITH

Well, well, well. Looks like Helvering v. Johnson wants to dance another round. In fact, taking a closer look at its large number of missing values …

case_decisions.loc[helvering_v_johnson_case_id].isna().sum()

26

… we’ll be seeing Helvering repeatedly as we finish addressing missing SCDB values. (It might be worth addressing separately like we did with Granite State v. Tandy Corp. in the previous post.)

If you recall from our first look at Helvering, this is an early FLSA case that resulted in a split decision once it made its way to SCOTUS. To determine its jurisdiction, we can either look for the corresponding cases as they appeared in lower courts or impute the jurisdiction from other cases. We know the latter is feasible here, since Helvering was, as the commissioner of the IRS at the time, a common target of cases that made their way through tax court to the Supreme Court, and, indeed, what we see is telling.

case_decisions.loc[
    case_decisions.case_name.str.contains(r'Helvering.*\bv\.?\b.*', regex=True, case=False)
    & (case_decisions.petitioner == 'Internal Revenue Service'),
    'jurisdiction'
].value_counts(dropna=False)

cert    92
NaN      1
Name: jurisdiction, dtype: Int64

All of these cases arise to the Supreme Court through cert. petitions.

case_decisions.loc[helvering_v_johnson_case_id, 'jurisdiction'] = 'cert'

Lopez v. Smith, the other case missing a jurisdiction, states in its headmatter that it is a cert. case from the 9th Circuit, which means we’re close to putting a bow on jurisdiction.

lopez_v_smith_case_id = '2014-076'

case_decisions.loc[lopez_v_smith_case_id, 'jurisdiction'] = 'cert'

Cert. Cases on the Original Jurisdiction Docket?

But you’ll need to hold onto that bow for a few more minutes. We’ll put a bow on jurisdiction when we’re good and ready to put a bow on jurisdiction And by that I mean after we validate our jurisdictions with one more source of truth: the docket! As discussed in our last post and the docket-umentation⁶, the docket numbers assigned to Supreme Court cases since 1971 themselves distinguish between cases reaching the Court via its appellate and original jurisdictions, with cases being heard by the Supreme Court in the first instance being assigned docket numbers ending in "Orig.".

We can take advantage of this to correct the jurisdictions of cases since 1971.

case_decisions.loc[
    (case_decisions.term >= 1971)
    & (case_decisions.jurisdiction != 'original')
    & case_decisions.docket.str.contains('Orig', case=False),
    ['case_name', 'us_cite', 'jurisdiction', 'cert_reason']
]

case_id	case_name	us_cite	jurisdiction	cert_reason
1995-026	UNITED STATES OF AMERICA v. STATE OF MAINE et al.	516 U.S. 365	cert	<NA>
2005-043	STATE OF ARIZONA v. STATE OF CALIFORNIA et al.	547 U.S. 150	cert	<NA>

Given their parties, both of these cases are clearly misclassified.

case_decisions.loc[
    (case_decisions.term >= 1971)
    & (case_decisions.jurisdiction != 'original')
    & case_decisions.docket.str.contains('Orig', case=False),
    ['jurisdiction', 'cert_reason']
] = ['original', 'no cert']

Likewise, we could also look for cases identified as original jurisdiction cases that are lacking proper docket numbers, but we’ve already corrected these cases in the previous post.

Identifying Original Jurisdiction Cases by Party Information

We could take this farther using the current definition of the Supreme Court’s original jurisdiction as declared in 28 U.S.C. §1251:

1. The Supreme Court shall have original and exclusive jurisdiction of all controversies between two or more States.
2. The Supreme Court shall have original but not exclusive jurisdiction of:
   1. All actions or proceedings to which ambassadors, other public ministers, consuls, or vice consuls of foreign states are parties;
   2. All controversies between the United States and a State;
   3. All actions or proceedings by a State against the citizens of another State or against aliens.

Subsection (a) of 28 U.S.C. §1251 is straightforward to examine.

case_decisions.loc[
    # 28 U.S.C. §1251(a): The Supreme Court shall have original and
    # exclusive jurisdiction of all controversies between two or
    # more States.
    (case_decisions[['petitioner', 'respondent']] == 'State').all(axis='columns')
    & (case_decisions.jurisdiction != 'original'),
    ['us_cite', 'docket', 'case_name', 'petitioner_state', 'respondent_state',
     'jurisdiction']
]

case_id	us_cite	docket	case_name	petitioner_state	respondent_state	jurisdiction
1917-116	246 U.S. 565	2, Orig.	COMMONWEALTH OF VIRGINIA v. STATE OF WEST VIRG...	West Virginia	Virginia	mandamus
1927-178	276 U.S. 557	2, Orig.	NEW MEXICO v. TEXAS	New Mexico	Texas	rehearing or reargument
1981-112	457 U.S. 85	80-1556	CORY, CONTROLLER OF CALIFORNIA, et al. v. WHIT...	California	Texas	cert
1986-145	483 U.S. 219	85-2116	PUERTO RICO v. BRANSTAD, GOVERNOR OF IOWA, et al.	Puerto Rico	Iowa	cert
1991-037	503 U.S. 91	90-1262	ARKANSAS, et al. v. OKLAHOMA, et al.	Arkansas	Oklahoma	cert
1992-008	506 U.S. 73	91-1158	MISSISSIPPI, et al. v. LOUISIANA et al.	Mississippi	Louisiana	cert

The first six of these seven cases are accurately labeled. I verified by hand that the first four alleged cert. cases in this list—all but Arizona v. California—actually arose on cert. by skimming their opinions.

That leaves the three questionable looking cases that appeared on the original jurisdiction docket: Virginia v. West Virginia, New Mexico v. Texas, and Arizona v. California. According to the jurisdiction documentation, the rare cases that fall into multiple jurisdiction categories are classified “on the basis of the writ”. Virginia v. West Virginia reached SCOTUS via a writ of mandamus, and New Mexico v. Texas occurred in response to a petition for rehearing. Both of these cases also fall under the Court’s original jurisdiction, but when categorizing these two cases the SCDB authors have been consistent with applying the above rule.

That leaves Arizona v. California, which should have its jurisdiction set to original. This is made clear either by reading the head matter of the decision or by comparing it to its long line of predecessors in the SCDB, all of which were original jurisdiction cases in a long-running SCOTUS dispute between states.

arizona_v_california_case_id = '2005-043'

case_decisions.loc[
    arizona_v_california_case_id,
    ['jurisdiction', 'cert_reason']
] = 'original', 'no cert'

We can now have some confidence that cases falling under the Court’s original jurisdiction via 28 U.S.C. §1251(a) are accurately classified in the SCDB. Subsection (b) of 28 U.S.C. §1251 is an entirely different can of worms, however, since the Court’s jurisdiction is discretionary. So long as the cases satisfying the conditions in subsection (b) have non-null jurisdictions—a fact we ensured earlier in this section—there will be no way for us to determine if their jurisdiction values are correct, using any of the features in the dataset.

That leaves us with options like text mining, manual validation, and more complex NLP to distinguish which cases falling under 28 U.S.C. §1251(b) need corrections. Given that (1) jurisdiction appears to be very well maintained as-is and (2) I don’t have that much use for jurisdiction to begin with, none of these options sound worth the trouble.

In other words, dear reader, jurisdiction is ready for you. Ready for you and that bow you’ve been holding on to since the end of the last section.

Reason for Granting cert. (cert_reason), or the One Where Dan Reads $87+$ SCOTUS Opinions

When a case reaches SCOTUS through a petition for a writ of certiorari, the cert_reason variable documents the reason, if any, the Court states in its opinions for granting the petition. (More often than not, no reason is explicitly provided by SCOTUS.)

Although the docs say there should be no null values in this field, that isn’t entirely the case.

case_decisions.cert_reason.value_counts(dropna=False).to_frame()

	cert_reason
no cert	18577
no reason given	4941
question presented	1598
Fed. court conflict	1577
significant question	1187
Fed. ct. conflict, sig. quest.	244
other reason	233
putative conflict	195
Fed. ct., state ct. conflict	124
NaN	74
Fed. court confusion	57
state court conflict	50
Fed. ct., state ct. confusion	13
state court confusion	5

Seventy-four NaNs! Spanning which jurisdictions?

case_decisions.loc[case_decisions.cert_reason.isna(), 'jurisdiction'].value_counts()

cert        73
original     1
Name: jurisdiction, dtype: Int64

Unfortunately for us, only one non-_cert_ case is present, but that’s better than nothing.

case_decisions.loc[
    (
        case_decisions.cert_reason.isna() &
        (case_decisions.jurisdiction != 'cert')
    ),
    'cert_reason'
] = 'no cert'

Now for the hard part. Why are there $73$ cert. cases missing their cert_reasons? Before committing to a research project involving reading through $73$ SCOTUS cases, we should at least consider being a bit more clever with the cert_reason imputation process. If nothing else, we should look to see if there are any strong correlations between cert_reason and other features in the dataset that we can use to impute these missing values with confidence.

And before we impute, let’s put on our explorer hats for a minute.

Relationship with decision_type

If you start poking around in the data, you’ll quickly notice an odd relationship between the decision_types of cases with missing cert_reasons.

is_modern_decision = case_decisions.term >= 1946

case_decisions.loc[
    case_decisions.cert_reason.isna(),
    ['case_name', 'date_decision', 'decision_type']
].head(10)

case_id	case_name	date_decision	decision_type
1939-143	HELVERING, COMMISSIONER OF INTERNAL REVENUE, v...	1939-12-11	equally divided vote
1941-163	HOLLAND v. LOWELL SUN CO.	1942-03-02	equally divided vote
1955-001	LUCY et al. v. ADAMS, DEAN OF ADMISSIONS, UNIV...	1955-10-10	per curiam (no oral)
1955-080	CAHILL v. NEW YORK, NEW HAVEN & HARTFORD RAILR...	1956-05-14	per curiam (no oral)
1957-030	BARTKUS v. ILLINOIS.	1958-01-06	equally divided vote
1957-031	LADNER v. UNITED STATES.	1958-01-06	equally divided vote
1957-069	COLUMBIA BROADCASTING SYSTEM, INC., et al. v....	1958-03-17	equally divided vote
1960-004	DAYTON RUBBER CO. v. CORDOVAN ASSOCIATES, INC.	1960-10-24	per curiam (no oral)
1960-020	BUSH et al. v. ORLEANS PARISH SCHOOL BOARD et al.	1960-12-12	per curiam (no oral)
1963-122	MARKS v. ESPERDY, DISTRICT DIRECTOR, IMMIGRATI...	1964-05-18	equally divided vote

Notice that these cases all were per curiam or “equally divided vote” decisions, two of the less common decision types in the SCDB.

case_decisions[is_modern_decision].decision_type.value_counts(dropna=False)

opinion of the court    7052
per curiam (no oral)     971
per curiam (oral)        562
judg. of the Ct.         248
equally divided vote     115
decrees                   66
Name: decision_type, dtype: Int64

The overwhelming majority of SCOTUS decisions fall into the run-of-the-mill “opinion” category, wherein a justice writes an opinion signed onto by the majority of the Court, with other justices joining the majority in concurrences and opposing it in dissents. Of the remaining modern decisions, all but a few briefcase loads are opinions made per curiam or are equally divided votes. By contrast, we can see that all of the cases with missing cert_reasons are classified as one of the rarer decision_types.

(case_decisions[case_decisions.cert_reason.isna()]
               .decision_type.value_counts(dropna=False))

per curiam (no oral)    42
equally divided vote    25
per curiam (oral)        6
Name: decision_type, dtype: Int64

So why am I bringing this up here? How is this high-level categorization of case outcomes remotely related to missing cert_reason values? While this information is unlikely to aid in our data wrangling, it helps us guess how these missing values came not to be (ha).

In all likelihood, the relationship we’re seeing here has more to do with the contents of opinions in these rarer decision_type categories than some inherent connection between the two fields. To review:

• Per curiam opinions are decisions made by the Court as a collective and are not attributed to or signed by the Court’s justices at the time or in future case law. If you’re interested in learning more, Cornell’s Legal Information Institute is a great place to start. What matters for our discussion is that these are typically short opinions consisting mostly of conclusory statements.

• Equally divided votes are exactly that: tie votes where there is no majority (or plurality) among the justices. In fact, distinguishing tie votes here is mostly unnecessary; in the event of an equally divided vote, SCOTUS almost always affirms the lower court’s decision in a per curiam opinion. These opinions are some of the court’s shortest, usually consisting of a single sentence:

  “The judgment is affirmed by an equally divided court.”

  For a recent example of this, see the riveting SCOTUS opinion in Washington v. United States.

Little to no context for why a case is appearing before SCOTUS can be gleaned from the published records for either of these types of decisions. As far as I can tell, one is left reviewing transcripts of oral arguments (when they exist), attempting to dig up underlying cert. petitions (and other docket entries), and reviewing lower court opinions to scrape together information about what might have been presented to SCOTUS. If I had to guess, this is the reason why there are missing cert_reasons in these records; it’s simply harder to find any information about them!

Approach 1: Imputing the Problem Away

As usual, filling in missing cert_reasons with a central tendency (mode) can serve as a reasonable baseline approach here. Of course, the cases under consideration are supposed to be cert. cases, so doing this blindly would be a mistake:

case_decisions.cert_reason.value_counts()

no cert                           18578
no reason given                    4941
question presented                 1598
Fed. court conflict                1577
significant question               1187
Fed. ct. conflict, sig. quest.      244
other reason                        233
putative conflict                   195
Fed. ct., state ct. conflict        124
Fed. court confusion                 57
state court conflict                 50
Fed. ct., state ct. confusion        13
state court confusion                 5
Name: cert_reason, dtype: Int64

(Practice Point: Assigning 'no cert' to what you know are cert. cases is not a great idea.)

We know that the remaining cases that lack cert_reasons are all cert. cases based on their jurisdictions. We’ve also just discussed that these cases all fall into one of the uncommon per curiam or “equally divided vote” decision types, and that these decision types are almost always accompanied by laconic Court opinions. Presumably then, 'no reason given' is the mode of the cert_reasons for cert. cases with each of these three decision_types. Just how prevalent is 'no reason given' in each group?

cert_reasons_by_decision_type = (
    case_decisions[case_decisions.cert_reason != 'no cert']
                  .groupby('decision_type')
                  .cert_reason
                  .value_counts()
)

display(
    (cert_reasons_by_decision_type / cert_reasons_by_decision_type.sum(level=0))[
        ['equally divided vote', 'per curiam (no oral)', 'per curiam (oral)']
    ].to_frame('P_cert(reason | decision type)')
)

decision_type	cert_reason	P_cert(reason | decision type)
equally divided vote	no reason given	0.981818
	question presented	0.009091
	significant question	0.009091
per curiam (no oral)	no reason given	0.931915
	Fed. court conflict	0.026950
	question presented	0.022695
	other reason	0.008511
	Fed. ct., state ct. conflict	0.004255
	significant question	0.002837
	Fed. court confusion	0.001418
	putative conflict	0.001418
per curiam (oral)	no reason given	0.752137
	question presented	0.158120
	Fed. court conflict	0.029915
	significant question	0.025641
	other reason	0.023504
	Fed. ct., state ct. conflict	0.006410
	Fed. ct. conflict, sig. quest.	0.002137
	putative conflict	0.002137

So long as our missing cert_reason values are missing at random, this suggests 'no reason given' is a reasonable fill value, at least for cases with equally divided votes and per curiam opinions following no oral arguments. For per curiam opinions for cases with oral arguments, however, we’re on shakier ground. Still, the latter only account for six of the seventy-three remaining cases missing cert. reasons, and we can consider dropping them from calculations (when relevant) or correcting them manually.

That said, perhaps we’re splitting hairs when breaking down these three per curiam-like decision types into their own categories.

display(
    cert_reasons_by_decision_type[
        ['equally divided vote', 'per curiam (no oral)', 'per curiam (oral)']
    ]
    .unstack().sum()
    .pipe(lambda s: s / s.sum())
    .sort_values(ascending=False)
    .to_frame('P(reason | cert case w/ per curiam-like decision type)')
)

cert_reason	P(reason | cert case w/ per curiam-like decision type)
no reason given	0.870616
question presented	0.070928
Fed. court conflict	0.025721
other reason	0.013250
significant question	0.011691
Fed. ct., state ct. conflict	0.004677
putative conflict	0.001559
Fed. court confusion	0.000779
Fed. ct. conflict, sig. quest.	0.000779

The 87% is nothing to scoff at. For the work I’m planning to do with the SCDB, it’s more than fine as a replacement value. For the general purpose, Python-friendly versions of this dataset that I’m hosting on DagsHub, however, I need to be more precise.

Approach 2: Hitting the Books

Once again, we are in the rare (for my usual areas of research and work on the job, at any rate) situation where all of the information we need to fill in the missing values in these SCDB records is at our disposal, so why spend the time imputing values? Time and resource constraints are the obvious answers, but I was seeing these errors differently. I took them as an opportunity to correct the record and, hopefully, contribute a small amount to the SCDB cause, reading all $73$ of these cases (or at least the portions required to get the answers I was looking for) and compiling a list of corrections for the SCDB curators. In fact, I read around $90$ cases since, before reorganizing the content in this series, another briefcase load of cases contained missing cert_reason values. I did this in two passes, the first occurring while I wrote the first blog post in this series, the second while writing up this post. Along the second pass, I realized that I’ve accomplished a secondary goal; I’m now much more efficient (and proficient) at reading cases than I was at the start of this project!

And the truth was as hard to account for using the other variables in this dataset as we expected. Many of the cert_reasons were NaNs because they weren’t cert. cases to begin with. You’ll see in the results below that I wound up suggesting changes to both the cert_reason and jurisdiction fields to account for this.

cert_reason_corrections = pd.read_csv(
    '../data/interim/case_decisions_with_null_certReasons.csv',
    index_col=0, na_filter=False
)


def display_corrections(corrections_df):
    with pd.option_context('display.max_rows', 100, 'display.max_colwidth', 200):
        display(
            corrections_df
                .assign(**{
                    'Opinion Link': lambda df: df['Opinion Link']
                                                 .pipe(as_link, text='Opinion')
                })
                .style
                .set_properties(**{'text-align': 'left',
                                   'white-space': 'pre-wrap'})
                .apply(highlight_if_includes_value, axis='columns', value='Yes')
        )


def as_link(series, text='Link'):
    return series.str.replace(r'(.+)', fr'<a href="\1">{text}</a>', regex=True)


def highlight_if_includes_value(
        series, value, highlight_style='background-color: purple; color: white'
):
    return len(series) * [highlight_style if (series == value).any() else '']


display_corrections(cert_reason_corrections.head())

2020 v1 Index	caseId	usCite	term	caseName	Opinion Link	Suggested jurisdiction	Suggested certReason(s)	Comments re. certReason	Opinion Excerpts	Ambiguous certReason or jurisdiction
18899	1939-143	308 U.S. 523	1939	HELVERING, COMMISSIONER OF INTERNAL REVENUE, v. JOHNSON	Opinion	unchanged	no reason given	This is an equally divided vote case, with majority opinion following the usual minimal per curiam template.		No
19242	1941-163	315 U.S. 784	1941	HOLLAND v. LOWELL SUN CO.	Opinion	unchanged	no reason given	This is an equally divided vote case, with majority opinion following the usual minimal per curiam template.		No
20858	1955-001	350 U.S. 1	1955	LUCY et al. v. ADAMS, DEAN OF ADMISSIONS, UNIVERSITY OF ALABAMA	Opinion	stay	case did not arise on cert or cert not granted		“APPLICATION TO VACATE ORDER GRANTING STAY OF INJUNCTION PENDING APPEAL AND TO REINSTATE INJUNCTION PENDING APPEAL”	No
20937	1955-080	351 U.S. 183	1955	CAHILL v. NEW YORK, NEW HAVEN & HARTFORD RAILROAD CO.	Opinion	unchanged OR unspecified, other	no reason given OR case did not arise on cert or cert not granted	This is an amendment to the case 350 U.S. 898 (https://cite.case.law/us/350/898/12069920/), deeming the Court’s original order to remand the case to a District Court in error and to instead remand to the Second Circuit. Since the unamended case doesn’t appear in the SCDB and it arose on cert, I think that cert is the appropriate jurisdiction here, as well.	“ON A MOTION TO RECALL AND AMEND THE JUDGMENT” “We deem our original order erroneous and recall it in the interest of fairness.”	Yes
21120	1957-030	355 U.S. 281	1957	BARTKUS v. ILLINOIS.	Opinion	unchanged	no reason given		“CERTIORARI TO THE SUPREME COURT OF ILLINOIS”	No

Note: I’ve decided not to break up my post with a huge table here. Check out the appendix for the complete set of corrections.

I identified a correct value for jurisdiction and cert_reason in all but four cases. Each of the remaining four cases, those highlighted above, contains some kind of ambiguity in relation to the SCDB classification. For instance, in Chandler v. Judicial Council of the Tenth Circuit, the case somehow arises as two separate writs. This also occurs in In re Demos, but in the latter case the Court only noted one of the three writs in the case’s headmatter. In Chandler, on the other hand, the headmatter includes “ON MOTION FOR LEAVE TO FILE PETITION FOR WRIT OF PROHIBITION AND/OR MANDAMUS”.

The tie breaking decision rules laid out in the jurisdiction documentation all ultimately fall back on the writ for a case in question. They don’t seem to account for the possibility of there being two such writs, or for that matter a hypothetical future writ that has the potential to be either a writ of prohibition or a writ of mandamus. Who knows? Maybe this deserves its own “superposition of writs” class! That sort of decision is above my pay grade, and I think each of these cases could use review by the experienced eye of an SCDB researcher. I’ve sent these corrections off to the SCDB, so we’ll hopefully see them (or corrections to my corrections) in its next release.

Meanwhile, I have values I feel good about for the other three of these decisions and am ok with making an arbitrary choice in Chandler. In the latter, the headmatter suggests prohibition (or at least lists it first), but the majority opinion opens as follows:

Petitioner, a United States District Judge, filed a motion for leave to file a petition for a writ of mandamus or alternatively a writ of prohibition addressed to the Judicial Council of the Tenth Circuit.

This at least suggests that the petitioner’s preferred writ would be a writ of mandamus, so I’ll run with that for now.

cert_reason_corrections.loc[
    cert_reason_corrections.caseName == 'CAHILL v. NEW YORK, NEW HAVEN & HARTFORD RAILROAD CO.',
    ['Suggested jurisdiction', 'Suggested certReason(s)']
] = ['cert', 'no reason given']

cert_reason_corrections.loc[
    cert_reason_corrections.caseName == 'CHANDLER, U.S. DISTRICT JUDGE v. JUDICIAL COUNCIL OF THE TENTH CIRCUIT OF THE UNITED STATES.',
    ['Suggested jurisdiction', 'Suggested certReason(s)']
] = ['mandamus', 'case did not arise on cert or cert not granted']

cert_reason_corrections.loc[
    cert_reason_corrections.caseName == 'In re BERGER',
    ['Suggested jurisdiction', 'Suggested certReason(s)']
] = ['unspecified, other', 'case did not arise on cert or cert not granted']

cert_reason_corrections.loc[
    cert_reason_corrections.caseName == 'In re JOHN ROBERT DEMOS, JR.',
    ['Suggested jurisdiction', 'Suggested certReason(s)']
] = ['writ of habeas corpus', 'case did not arise on cert or cert not granted']

With those final corrections in place, we’re just about ready to put a bow on jurisdiction, cert_reason, and the SCDB Background Variables. We still need to incorporate the above corrections into jurisdiction and cert_reason. Moreover, some of the cases referenced in the suggestions CSV file no longer exist in our corrected dataset, after we found in our last post that they should never have been entered into the SCDB in the first place, so we’ll need to disregard those suggestions.

cert_reason_documentation_to_value = {
    'case did not arise on cert or cert not granted': 'no cert',
    'federal court conflict': 'Fed. court conflict',
    'federal court conflict and to resolve important or significant question': 'Fed. ct. conflict, sig. quest.',
    'putative conflict': 'putative conflict',
    'conflict between federal court and state court': 'Fed. ct., state ct. conflict',
    'state court conflict': 'state court conflict',
    'federal court confusion or uncertainty': 'Fed. court confusion',
    'state court confusion or uncertainty': 'state court confusion',
    'federal court and state court confusion or uncertainty': 'Fed. ct., state ct. confusion',
    'to resolve important or significant question': 'significant question',
    'to resolve question presented': 'question presented',
    'no reason given': 'no reason given',
    'other reason': 'other reason',
}

surviving_case_corrections = (
    cert_reason_corrections
        .set_index('caseId')
        .loc[
            set(cert_reason_corrections.caseId) & set(case_decisions.index),
            ['Suggested certReason(s)', 'Suggested jurisdiction']
        ]
        .astype('string')
)

corrected_jurisdictions = (
    surviving_case_corrections['Suggested jurisdiction']
        .loc[lambda jurisdiction: jurisdiction != 'unchanged'])

corrected_cert_reasons = (
    surviving_case_corrections['Suggested certReason(s)']
        .map(cert_reason_documentation_to_value)
)

valid_jurisdictions = case_decisions.jurisdiction.dropna().unique()
valid_cert_reasons = case_decisions.cert_reason.dropna().unique()

assert corrected_jurisdictions.isin(valid_jurisdictions).all(), \
        'Invalid jurisdiction correction found'
assert corrected_cert_reasons.isin(valid_cert_reasons).all(), \
        'Invalid cert_reason correction found'

case_decisions.loc[
    corrected_jurisdictions.index,
    'jurisdiction'
] = corrected_jurisdictions

case_decisions.loc[
    corrected_cert_reasons.index,
    'cert_reason'
] = corrected_cert_reasons

Blog Post Trivia

This is actually the first variable that I looked at during the data cleanup process that required me to read opinions in any amount of detail. (I decided to delay posting about it until now to keep things organized.) While perhaps not the most useful variable for my end goals with the dataset, cert_reason served as a useful starting point for getting my toes wet reading opinions, since the correct value of cert_reason for a case is (a) usually unambiguous and (b) almost always identifiable early in the majority opinion. This made cert_reason a good field for me to dive deeper into while also finding a workflow for accessing opinions online.

Processing Chronological Variables

These variables describe different aspects of the timing of a case in the broad sense. Given the numeric column and date processing I already performed in my previous post, we can limit ourselves to a couple variables here.

chronological_variables_of_interest = ['chief', 'natural_court']

As you’ve probably surmised, the chief variable tracks who presided as chief justice over each case. The natural_court variable is more interesting. Each record in the SCDB is assigned a natural_court value of the form “$\mathtt{chief}\,n$” where

1. $\mathtt{chief}$ is the presiding chief justice for said case (the value from the chief column) and
2. $n \in \mathbb{N}$ indicates that the set of justices taking part in the case is the $n$-th unique set since $\mathtt{chief}$ assumed the role of chief justice.

This makes natural_court a super simple and convenient way to spot temporary and permanent changes to the SCOTUS roster. The only exceptions occur in the exceptionally rare cases without chief justices.⁷

natural_court_chief, natural_court_ordinal = (
    case_decisions.natural_court.str.split(r' (?=\d+$)', n=1, expand=True)
        .pipe(lambda split_natural_court: (split_natural_court[0],
                                           split_natural_court[1])))

print('Anomalous Natural Court Chief Justices:')
display(
    natural_court_chief[case_decisions.chief != natural_court_chief]
        .value_counts()
)

print('\nAnomalous Natural Court Ordinals:')
display(
    natural_court_ordinal[case_decisions.chief != natural_court_chief]
        .value_counts(dropna=False)
)

Anomalous Natural Court Chief Justices:


No Chief (Post-Fuller)      42
No Chief (Post-Rutledge)     3
Name: 0, dtype: Int64



Anomalous Natural Court Ordinals:


NaN    45
Name: 1, dtype: Int64

Both the chief and natural_court variables are well-maintained, leaving us with nothing to do here for the time being.

assert case_decisions[['chief', 'natural_court']].notna().all().all()

That’s a Wrap!

Well, we’ve once again made it through examining a large chunk of the SCDB dataset. All told, we’ve now processed about half of the features in the SCDB! This time the techniques we wound up needing for imputation were arguably more basic than in earlier posts, although the work remained time consuming due to the need for manual data labelling.

There are three remaining categories of SCDB variables to look into in this series, which together comprise most of the meat of the dataset. These are the variables organized into the “Substantive”, “Outcome”, and “Voting & Opinion” categories within the database, and we’ll treat them in that order. If you’ve followed along so far, hopefully you’ll enjoy the thrilling second act in the series!

Appendix: All Manual cert_reason Corrections

For those of you really jonesing for cert_reason corrections, here is the complete table.

display_corrections(cert_reason_corrections)

2020 v1 Index	caseId	usCite	term	caseName	Opinion Link	Suggested jurisdiction	Suggested certReason(s)	Comments re. certReason	Opinion Excerpts	Ambiguous certReason or jurisdiction
18899	1939-143	308 U.S. 523	1939	HELVERING, COMMISSIONER OF INTERNAL REVENUE, v. JOHNSON	Opinion	unchanged	no reason given	This is an equally divided vote case, with majority opinion following the usual minimal per curiam template.		No
19242	1941-163	315 U.S. 784	1941	HOLLAND v. LOWELL SUN CO.	Opinion	unchanged	no reason given	This is an equally divided vote case, with majority opinion following the usual minimal per curiam template.		No
20858	1955-001	350 U.S. 1	1955	LUCY et al. v. ADAMS, DEAN OF ADMISSIONS, UNIVERSITY OF ALABAMA	Opinion	stay	case did not arise on cert or cert not granted		“APPLICATION TO VACATE ORDER GRANTING STAY OF INJUNCTION PENDING APPEAL AND TO REINSTATE INJUNCTION PENDING APPEAL”	No
20937	1955-080	351 U.S. 183	1955	CAHILL v. NEW YORK, NEW HAVEN & HARTFORD RAILROAD CO.	Opinion	cert	no reason given	This is an amendment to the case 350 U.S. 898 (https://cite.case.law/us/350/898/12069920/), deeming the Court’s original order to remand the case to a District Court in error and to instead remand to the Second Circuit. Since the unamended case doesn’t appear in the SCDB and it arose on cert, I think that cert is the appropriate jurisdiction here, as well.	“ON A MOTION TO RECALL AND AMEND THE JUDGMENT” “We deem our original order erroneous and recall it in the interest of fairness.”	Yes
21120	1957-030	355 U.S. 281	1957	BARTKUS v. ILLINOIS.	Opinion	unchanged	no reason given		“CERTIORARI TO THE SUPREME COURT OF ILLINOIS”	No
21121	1957-031	355 U.S. 282	1957	LADNER v. UNITED STATES.	Opinion	unchanged	no reason given		“CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE FIFTH CIRCUIT.”	No
21159	1957-069	356 U.S. 43	1957	COLUMBIA BROADCASTING SYSTEM, INC., et al. v. LOEW'S INC. et al.	Opinion	unchanged	no reason given		“CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT.”	No
21524	1960-004	364 U.S. 299	1960	DAYTON RUBBER CO. v. CORDOVAN ASSOCIATES, INC.	Opinion	unchanged	no reason given		“ON PETITION FOR WRIT OF CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE SIXTH CIRCUIT”	No
21540	1960-020	364 U.S. 500	1960	BUSH et al. v. ORLEANS PARISH SCHOOL BOARD et al.	Opinion	stay	case did not arise on cert or cert not granted		“ON MOTION FOR STAY”	No
22057	1963-122	377 U.S. 214	1963	MARKS v. ESPERDY, DISTRICT DIRECTOR, IMMIGRATION AND NATURALIZATION SERVICE.	Opinion	unchanged	no reason given		“CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE SECOND CIRCUIT”	No
22086	1963-151	378 U.S. 123	1963	VIKING THEATRE CORP. v. PARAMOUNT FILM DISTRIBUTING CORP. et al.	Opinion	unchanged	no reason given		“CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE THIRD CIRCUIT”	No
22140	1964-022	379 U.S. 130	1964	INDEPENDENT PETROLEUM WORKERS OF AMERICA, INC. v. AMERICAN OIL CO.	Opinion	unchanged	no reason given		“CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE SEVENTH CIRCUIT”	No
22296	1965-044	382 U.S. 1003	1965	CHANDLER, U.S. DISTRICT JUDGE v. JUDICIAL COUNCIL OF THE TENTH CIRCUIT OF THE UNITED STATES.	Opinion	mandamus	case did not arise on cert or cert not granted	This doesn’t look like a cert petition. It’s a request under the All Writs Act (28 U.S.C. §1651) for prohibition or mandamus. The classification seems like a toss up here and up to internal conventions at the SCDB.	“ON MOTION FOR LEAVE TO FILE PETITION FOR WRIT OF PROHIBITION AND/OR MANDAMUS” “Petitioner, a United States District Judge, filed a motion for leave to file a petition for a writ of mandamus or alternatively a writ of prohibition addressed to the Judicial Council of the Tenth Circuit.”	Yes
22363	1965-111	384 U.S. 269	1965	GREER v. BETO, CORRECTIONS DIRECTOR	Opinion	unchanged	no reason given		“ON PETITION FOR WRIT OF CERTIORARI TO THE COURT OF CRIMINAL APPEALS OF TEXAS”	No
22565	1967-025	389 U.S. 143	1967	HACKIN v. ARIZONA et al.	Opinion	appeal	case did not arise on cert or cert not granted	This case reached the Court on appeal, not on cert.	“APPEAL FROM THE SUPREME COURT OF ARIZONA”	No
22627	1967-087	390 U.S. 455	1967	ALITALIA-LINEE AEREE ITALIANE, S.P.A. v. LISI et al.	Opinion	unchanged	no reason given		“CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE SECOND CIRCUIT”	No
22628	1967-088	390 U.S. 456	1967	ANDERSON v. JOHNSON, WARDEN	Opinion	unchanged	no reason given		“CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE SIXTH CIRCUIT” “Four members of the Court would reverse. Four members of the Court would dismiss the writ as improvidently granted. Consequently, the judgment of the United States Court of Appeals for the Sixth Circuit remains in effect.”	No
23021	1970-003	400 U.S. 16	1970	COLOMBO v. NEW YORK	Opinion	unchanged	no reason given		“ON PETITION FOR WRIT OF CERTIORARI TO THE COURT OF APPEALS OF NEW YORK”	No
23135	1970-118	402 U.S. 690	1970	CONNOR et al. v. JOHNSON et al.	Opinion	stay	case did not arise on cert or cert not granted		“ON APPLICATION FOR STAY”	No
23531	1973-004	414 U.S. 12	1973	DENNETT v. HOGAN, WARDEN	Opinion	writ of habeas corpus	case did not arise on cert or cert not granted	Unusual decision. This actually wasn’t a cert case, but the Solicitor General recommended it be treated like one and remanded, and SCOTUS followed that advice.		No
23573	1973-046	414 U.S. 685	1973	SNIDER et al. v. ALL STATE ADMINISTRATORS, INC., et al.	Opinion	docketing fee	case did not arise on cert or cert not granted	Rule 39 petition to dispense with printing the cert petition		No
24042	1976-006	429 U.S. 17	1976	STANDARD OIL CO. OF CALIFORNIA v. UNITED STATES	Opinion	unspecified, other	case did not arise on cert or cert not granted	This is a “motion to recall mandate” from case 412 U.S. 924.		No
24246	1977-031	434 U.S. 425	1977	VENDO CO. v. LEKTRO-VEND CORP. et al.	Opinion	unchanged	to resolve important or significant question		The order of the United States District Court was affirmed by the Court of Appeals for the Seventh Circuit, 545 F. 2d 1050 (1976), and we granted certiorari to consider the important question of the relationship between state and federal courts which such an injunction raises. 429 U. S. 815 (1976). (p. 626 in linked volume)	No
24686	1979-156	448 U.S. 725	1979	HAMMETT v. TEXAS	Opinion	unchanged	no reason given		“ON MOTION TO WITHDRAW PETITION FOR WRIT OF CERTIORARI TO THE COURT OF CRIMINAL APPEALS OF TEXAS”	No
25085	1982-068	461 U.S. 230	1982	ALABAMA v. EVANS	Opinion	stay	case did not arise on cert or cert not granted	This appears to be an application to vacate a stay of execution (by a circuit judge to SCOTUS), not a cert case. It looks like similar cases that are applications for stays should be assigned “stay” as their jurisdictions. I’ve labeled applications to vacate stays with the stay jurisdiction as well, since it seems the most applicable. (Presumably the Court’s power to vacate stays falls into the same jurisdiction as its power to grant them.) I’ve also distinguished stays and requests to vacate stays in this column below.		No
25184	1983-001	464 U.S. 1	1983	AUTRY v. ESTELLE, DIRECTOR, TEXAS DEPARTMENT OF CORRECTIONS	Opinion	stay	case did not arise on cert or cert not granted	See ALABAMA v. EVANS discussion. This is an application for a stay.		No
25193	1983-010	464 U.S. 109	1983	SULLIVAN v. WAINWRIGHT, SECRETARY, FLORIDA DEPARTMENT OF CORRECTIONS, et al.	Opinion	stay	case did not arise on cert or cert not granted	See ALABAMA v. EVANS discussion. This is an application for a stay.		No
25202	1983-019	464 U.S. 377	1983	WOODARD, SECRETARY OF CORRECTIONS OF NORTH CAROLINA, et al. v. HUTCHINS	Opinion	stay	case did not arise on cert or cert not granted	See ALABAMA v. EVANS discussion. This is an application to vacate a stay.		No
25419	1984-063	470 U.S. 903	1984	BOARD OF EDUCATION OF OKLAHOMA CITY v. NATIONAL GAY TASK FORCE	Opinion	appeal	case did not arise on cert or cert not granted		“APPEAL FROM THE UNITED STATES COURT OF APPEALS FOR THE TENTH CIRCUIT”	No
25420	1984-064	470 U.S. 904	1984	FUGATE v. NEW MEXICO	Opinion	unchanged	no reason given		“CERTIORARI TO THE SUPREME COURT OF NEW MEXICO”	No
25424	1984-068	471 U.S. 81	1984	KENNETH CORY, LEO T. MCCARTHY AND JESSE R. HUFF v. WESTERN OIL AND GAS ASSOCIATION et al.	Opinion	appeal	case did not arise on cert or cert not granted		“APPEAL FROM THE UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT”	No
25425	1984-069	471 U.S. 82	1984	ROGER L. SPENCER, ET UX. v. SOUTH CAROLINA TAX COMMISSION et al.	Opinion	unchanged	no reason given		“CERTIORARI TO THE SUPREME COURT OF SOUTH CAROLINA”	No
25426	1984-070	471 U.S. 83	1984	BOARD OF TRUSTEES OF THE VILLAGE OF SCARSDALE, et al. v. KATHLEEN S. MCCREARY et al.	Opinion	unchanged	no reason given		“CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE SECOND CIRCUIT”	No
25429	1984-073	471 U.S. 146	1984	OKLAHOMA v. TIMOTHY R. CASTLEBERRY AND NICHOLAS RAINERI	Opinion	unchanged	no reason given		“CERTIORARI TO THE COURT OF CRIMINAL APPEALS OF OKLAHOMA”	No
25430	1984-074	471 U.S. 147	1984	PATRICK RAMIREZ v. INDIANA	Opinion	unchanged	no reason given		“CERTIORARI TO THE COURT OF APPEALS OF INDIANA”	No
25484	1984-129	472 U.S. 478	1984	JENSEN, DIRECTOR, DEPARTMENT OF MOTOR VEHICLES OF NEBRASKA, et al. v. QUARING	Opinion	unchanged	no reason given		“CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE EIGHTH CIRCUIT”	No
25538	1985-017	474 U.S. 213	1985	EASTERN AIR LINES, INC. v. MAHFOUD ON BEHALF OF MAHFOUD et al.	Opinion	unchanged	no reason given		“CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE FIFTH CIRCUIT”	No
26071	1988-071	490 U.S. 89	1988	WRENN v. BENSON et al.	Opinion	unchanged	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS on a cert petition.	The Clerk of the Court is directed not to accept further filings from petitioner in which he seeks leave to proceed in forma pauperis unless the affidavit submitted with the filing indicates that his financial condition has substantially changed from that reflected in the affidavits he submitted in Wrenn v. Benson and Wrenn v. Ohio Dept. of Mental Health, supra. Justice is not served if the Court continues to process his requests when his financial condition has not changed from that reflected in a previous filing for which he was denied leave to proceed in forma pauperis.	No
26157	1989-005	493 U.S. 38	1989	MICHIGAN CITIZENS FOR AN INDEPENDENT PRESS et al. v. THORNBURGH, ATTORNEY GENERAL OF THE UNITED STATES, et al.	Opinion	unchanged	no reason given		“CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE DISTRICT OF COLUMBIA CIRCUIT”	No
26227	1989-075	495 U.S. 320	1989	DELO, SUPERINTENDENT, POTOSI CORRECTIONAL CENTER v. STOKES	Opinion	stay	case did not arise on cert or cert not granted	See ALABAMA v. EVANS discussion. This is an application to vacate a stay.		No
26244	1989-092	495 U.S. 731	1989	DEMOSTHENES et al. v. BAAL et al.	Opinion	stay	case did not arise on cert or cert not granted	See ALABAMA v. EVANS discussion. This is an application for a stay.		No
26313	1990-021	498 U.S. 233	1990	In re BERGER	Opinion	unspecified, other	case did not arise on cert or cert not granted	This is a motion from a defense attorney appointed under Rule 39.7 to receive more than the $2500 cap of attorney’s fees provided by the Criminal Justice Act. I’ve suggested “unspecified, other” here but could see “docketing fee” as an alternative. The case in which Berger was assigned as counsel in Saffle v. Parks, 494 U.S. 484 (https://cite.case.law/u-s/494/484/), which was a cert case.		Yes
26319	1990-027	498 U.S. 335	1990	UNITED STATES v. DARLINA K. FRANCE	Opinion	unchanged	no reason given		“CERTIORARI TO THE UNITED STATES COURT OF APPEALS FORTHE NINTH CIRCUIT”	No
26358	1990-067	500 U.S. 16	1990	In re JOHN ROBERT DEMOS, JR.	Opinion	writ of habeas corpus	case did not arise on cert or cert not granted	This is an oddball of a case. Demos is a known frivolous litigator, and his case reaches the court on three separate, concurrent petitions for writs. These include a writ of certiorari, but I suggest this is classified as a writ of habeas corpus (viewing the “in re” case name and the petition in the headmatter as controlling).	“ON PETITION FOR WRIT OF HABEAS CORPUS” “Pro se petitioner Demos filed petitions for a writ of certiorari, a writ of habeas corpus, and a writ of mandamus all seeking relief from a single lower court order. He has made 32 in forma pauperis filings in this Court since the October 1988 Term began.”	Yes
26486	1991-065	503 U.S. 653	1991	JAMES GOMEZ AND DANIEL VASQUEZ v. UNITED STATES DISTRICT COURT FOR THE NORTHERN DISTRICT OF CALIFORNIA, et al.	Opinion	stay	case did not arise on cert or cert not granted	(p. 653 in reference volume) See ALABAMA v. EVANS discussion. This is an application to vacate a stay.		No
26494	1991-073	504 U.S. 188	1991	ROGER KEITH COLEMAN v. CHARLES E. THOMPSON, WARDEN, et al.	Opinion	stay	case did not arise on cert or cert not granted	See ALABAMA v. EVANS discussion. This is an application for a stay.		No
26541	1991-120	505 U.S. 1084	1991	LEONA BENTEN, et al. v. DAVID KESSLER, COMMISSIONER, FOOD AND DRUG ADMINISTRATION, et al.	Opinion	stay	case did not arise on cert or cert not granted	See ALABAMA v. EVANS discussion. This is an application to vacate a stay.		No
26568	1992-027	507 U.S. 1	1992	UNITED STATES v. JERRY J. NACHTIGAL	Opinion	unchanged	no reason given		“ON PETITION FOR WRIT OF CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT”	No
26660	1992-120	509 U.S. 823	1992	PAUL DELO, SUPERINTENDENT, POTOSI CORRECTIONAL CENTER v. WALTER J. BLAIR	Opinion	stay	case did not arise on cert or cert not granted	See ALABAMA v. EVANS discussion. This is an application to vacate a stay.		No
26661	1992-121	507 U.S. 1026	1992	GRANITE STATE INS. CO. v. TANDY CORP.	Opinion	unchanged	no reason given	Rule 46 dismissal		No
26662	1993-001	510 U.S. 1	1993	ROY A. DAY v. DONALD P. DAY	Opinion	docketing fee	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS.		No
26663	1993-002	510 U.S. 4	1993	In re GEORGE SASSOWER	Opinion	docketing fee	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS.		No
26706	1993-045	511 U.S. 364	1993	In re GRANT ANDERSON	Opinion	docketing fee	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS.		No
26759	1994-002	513 U.S. 5	1994	ANTHONY S. AUSTIN v. UNITED STATES	Opinion	unchanged	to resolve question presented		“CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE EIGHTH CIRCUIT” “In this case, we are asked to decide whether the Excessive Fines Clause of the Eighth Amendment applies to forfeitures of property under 21 U. S. C. §§881(a)(4) and (a)(7).”	No
26849	1994-093	515 U.S. 951	1994	J. D. NETHERLAND, WARDEN v. LEM DAVIS TUGGLE	Opinion	unchanged	no reason given		“ON PETITION FOR WRIT OF CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE FOURTH CIRCUIT”	No
26866	1995-017	516 U.S. 233	1995	LOTUS DEVELOPMENT CORPORATION v. BORLAND INTERNATIONAL, INC.	Opinion	unchanged	no reason given		“CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE FIRST CIRCUIT”	No
26870	1995-021	516 U.S. 297	1995	ROBERT ATTWOOD v. HARRY K. SINGLETARY, JR., SECRETARY, FLORIDA DEPARTMENT OF CORRECTIONS	Opinion	docketing fee	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS on a habeas petition.		No
26874	1995-025	516 U.S. 363	1995	SYLVESTER JONES v. ABC-TV et al.	Opinion	docketing fee	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS.		No
26875	1995-026	516 U.S. 365	1995	UNITED STATES OF AMERICA v. STATE OF MAINE et al.	Opinion	original	case did not arise on cert or cert not granted	This was a motion for entry of a supplemental decree	“ON MOTION FOR ENTRY OF SUPPLEMENTAL DECREE”	No
26891	1995-042	517 U.S. 343	1995	LIANG-HOUH SHIEH v. EDWARD KAKITA et al.	Opinion	docketing fee	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS.		No
26892	1995-043	517 U.S. 345	1995	MICHAEL BOWERSOX, SUPERINTENDENT, POTOSI CORRECTIONAL CENTER v. DOYLE J. WILLIAMS	Opinion	stay	case did not arise on cert or cert not granted	See ALABAMA v. EVANS discussion. This is an application to vacate a stay.		No
26982	1996-043	520 U.S. 303	1996	In re EILEEN VEY	Opinion	docketing fee	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS on a habeas petition.		No
27036	1996-097	521 U.S. 982	1996	WILLIAM R. POUNDERS, JUDGE, SUPERIOR COURT OF CALIFORNIA, LOS ANGELES COUNTY v. PENELOPE WATSON	Opinion	unchanged	no reason given			No
27137	1997-103	523 U.S. 1015	1997	CATERPILLAR INC. v. INTERNATIONAL UNION, UNITED AUTOMOBILE, AEROSPACE AND AGRICULTURAL IMPLEMENT WORKERS OF AMERICA	Opinion	unchanged	no reason given	Rule 46 dismissal	“Writ of certiorari dismissed under this Court’s Rule 46.1.”	No
27138	1997-104	523 U.S. 1113	1997	LEWIS v. BRUNSWICK CORP.	Opinion	unchanged	no reason given	Rule 46 dismissal	“Writ of certiorari dismissed under this Court’s Rule 46.1.”	No
27148	1998-010	525 U.S. 153	1998	In re MICHAEL KENNEDY	Opinion	docketing fee	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS on a habeas petition.		No
27156	1998-018	525 U.S. 315	1998	CALIFORNIA PUBLIC EMPLOYEES' RETIREMENT SYSTEM, et al. v. PAUL FELZEN et al.	Opinion	unchanged	no reason given		“CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE SEVENTH CIRCUIT”	No
27171	1998-033	526 U.S. 122	1998	BARBARA SCHWARZ v. NATIONAL SECURITY AGENCY et al.	Opinion	docketing fee	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS on a cert petition.		No
27173	1998-035	526 U.S. 135	1998	VINCENT F. RIVERA v. FLORIDA DEPARTMENT OF CORRECTIONS	Opinion	docketing fee	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS on a cert petition.		No
27202	1998-064	526 U.S. 811	1998	RICHARD M. CROSS v. PELICAN BAY STATE PRISON et al.	Opinion	docketing fee	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS on a cert petition.		No
27230	1998-093	527 U.S. 885	1998	RONALD DWAYNE WHITFIELD v. TEXAS	Opinion	docketing fee	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS on a cert petition.		No
27232	1999-002	528 U.S. 1	1999	DONALD H. BRANCATO v. PRISCILLA F. GUNN et al.	Opinion	docketing fee	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS on a cert petition.		No
27233	1999-003	528 U.S. 3	1999	MICHAEL C. ANTONELLI v. DALE CARIDINE et al.	Opinion	docketing fee	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS on a cert petition.		No
27234	1999-004	528 U.S. 5	1999	KEITH RUSSELL JUDD v. UNITED STATES DISTRICT COURT FOR THE WESTERN DISTRICT OF TEXAS et al.	Opinion	docketing fee	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS on a cert petition.		No
27235	1999-005	528 U.S. 9	1999	ROBERT E. PRUNTY v. W. BROOKS et al.	Opinion	docketing fee	case did not arise on cert or cert not granted	This is a motion for leave to proceed in forma pauperis that was denied by SCOTUS on a cert petition.		No
27487	2001-086	535 U.S. 682	2001	MATHIAS v. WORLDCOM TECHS.	Opinion	unchanged	no reason given		Because, after full briefing and oral argument, it is clear that petitioners were the prevailing parties below and seek review of uncongenial findings not essential to the judgment and not binding upon them in future litigation, certiorari is dismissed as improvidently granted.	No
27542	2002-055	538 U.S. 715	2002	CITY OF LOS ANGELES v. EDWIN F. DAVID	Opinion	unchanged	no reason given		“ON PETITION FOR WRIT OF CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT”	No
27731	2005-001	546 U.S. 1	2005	PAUL ALLEN DYE v. GERALD HOFBAUER, WARDEN	Opinion	unchanged	no reason given		“ON PETITION FOR WRIT OF CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE SIXTH CIRCUIT”	No
27732	2005-002	546 U.S. 6	2005	DORA B. SCHRIRO, DIRECTOR, ARIZONA DEPARTMENT OF CORRECTIONS, PETITIONER v. ROBERT DOUGLAS SMITH	Opinion	unchanged	no reason given		“ON PETITION FOR WRIT OF CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT”	No
27773	2005-043	547 U.S. 150	2005	STATE OF ARIZONA v. STATE OF CALIFORNIA et al.	Opinion	original	case did not arise on cert or cert not granted	(p. 150 in reference volume) This is a consolidation of fifty years of decrees originating from a bill of complaint from Arizona to California (and seven of its public agencies). These should fall under the Court’s original jurisdiction.	“on bill of complaint”	No
27775	2005-045	547 U.S. 188	2005	JEFFREY JEROME SALINAS v. UNITED STATES	Opinion	unchanged	no reason given	(p. 188 in reference volume)	“on petition for writ of certiorari to the United States Court of Appeals for the Fifth Circuit”	No
27818	2006-001	549 U.S. 1	2006	HELEN PURCELL, MARICOPA COUNTY RECORDER, et al. v. MARIA M. GONZALEZ et al.	Opinion	stay	case did not arise on cert or cert not granted	(p. 1 in reference volume)	“on application for stay”	No
27839	2006-022	549 U.S. 437	2006	KEITH LANCE, et al. v. MIKE COFFMAN, COLORADO SECRETARY OF STATE	Opinion	appeal	case did not arise on cert or cert not granted			No
27893	2007-001	552 U.S. 1	2007	BOARD OF EDUCATION OF THE CITY SCHOOL DISTRICT OF THE CITY OF NEW YORK v. TOM F., ON BEHALF OF GILBERT F., A MINOR CHILD	Opinion	unchanged	no reason given		"CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE SECOND CIRCUIT"	No
27915	2007-023	552 U.S. 440	2007	WARNER-LAMBERT CO., LLC, et al. v. KIMBERLY KENT et al.	Opinion	unchanged	no reason given		“CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE SECOND CIRCUIT”	No
27966	2007-074	552 U.S. 1085	2007	KLEIN & CO. FUTURES, INC. v. BD. OF TRADE	Opinion	unchanged	no reason given	(p. 1085 in reference volume) Rule 46 dismissal	“Writ of certiorari dismissed under this Court’s Rule 46.1.” (p. 1086)	No
28142	2009-094	561 U.S. 476	2009	WEYHRAUCH v. UNITED STATES	Opinion	unchanged	no reason given		“The judgment is vacated, and the case is remanded to the United States Court of Appeals for the Ninth Circuit for further consideration in light of Skilling v. United States, ante, p. ___.”	No

1. Thanks to Michael Rauh for pointing out that it is a quarter of the length of Harry Potter and the Sorcerer’s Stone. ↩

2. If you were hoping for interesting cases here, you’re in for some serious disappointment. Both Cudahy and Lowell center on who in the Department of Labor’s Wage and Hours Division has the power to issue subpoenas duces tecum (orders to produce documents and evidence) under the Fair Labor Standards Act (or, more pedantically, whether or not the FLSA confers the right to delegate that authority to regional administrators). Pretty dry stuff.

   Still, there are as always a few interesting pieces of trivia pertaining to the Cudahy case. It is the third Supreme Court case on the FLSA of 1938 (hurray for labor laws!), and I could definitely be convinced to read more about the legal battles that were presumably raging at this time about worker rights. Also, as the review discusses, the Cudahy decision is a pretty clear error in interpretation of the FLSA. Most interesting of all, however, is the history of the Cudahy Packing Company, a meat packing company founded in the 1880s. It is one of the “big five” meat packing companies of Chicago wrapped up into Durham’s company in The Jungle, companies that used appalling working conditions, deplorable quality control, brutal business tactics to drive up profits and maintain an oligopoly. Cudahy also owned one of the ugliest buildings I have ever seen on the National Register of Historic Places, although admittedly I don’t actively seek out industrial buildings. There’s a huge rabbit hole here that I don’t plan to follow at the moment, but I’ll leave my fellow Los Angeles natives with a great piece on Cudahy’s entrance, activities, and legacy in our city. ↩

3. Theoretically at least. It doesn’t look like Palau has been party to a case that’s reached SCOTUS yet. ↩

4. I hope this number looks outdated in the next year or two. ↩

5. For those playing along at home, ca. sa. is short for capias ad satisfaciendum. That and recognizing from context that the bail’s “principal” is a human being can get you most of the way through the aforementioned paragraph. I’m still not confident I understand how a writ of scire facias works, however. I’m so grateful that efforts have been made to make the law more readable in the intervening centuries between Davidson and the present. ↩

6. Again, I am so sorry. This is less wordplay and more a cry for help. ↩

7. “Fun” fact for anyone wondering why I use split over rsplit in the next cell: a bug has existed in Pandas for at least several years that prevents rsplit from recognizing regular expressions as separators. It also looks like work towards a fix has stalled. Maybe I’ll pick up that torch. ↩