Get the data · Anticipatory Scandal

Every file, the R script, and a way to run it in the browser.

What this is. This site reproduces what the OPCC publishes about concluded discipline cases. We added one flag: it marks the allegations involving gender-based violence. The methods page explains how the flag was built and checked. The numbers compare groups within one administrative record; they show how the gender-based violence cases were treated and cannot say why. That group is small, so a handful of cases can shift a percentage. We describe the allegations as substantiated because almost every one here, 1,235 of 1,243, ends in a disciplinary or corrective measure; the OPCC's own table does not use the word.

Every file

Everything behind this site can be downloaded: the OPCC's export as published, the analysis table, every hand-coding sheet with its supporting quotes, and the scripts behind every number. The archive is 4.2 MB. The 576 adjudication PDFs (569 MB) stay at opcc.bc.ca, and the index in the archive lists every URL.

Or take the files one at a time

bc_police_discipline_complete.zip 4.2 MB

The complete archive: the OPCC's export as published, the coded tables, every hand-coding sheet with its evidence quotes, the passage corpora, the figures and the scripts that produce every number.

What is inside the archive

The README maps its five folders: the source as published, the coded tables, the hand-coding sheets, the passage corpora, and the results with their scripts. Each hand-coded sheet opens with a provenance header naming who or what coded it, when, and with which script.

allegations.csv 1,243 rows · 272 KB

The analysis frame. One row per allegation, with the four flag columns kept separate; the dictionary below defines every column.

Column dictionary: what every column holds

Column	What it holds
allegation_id	Row number, 1 to 1,243.
file_number	The OPCC file number; one case per file.
agency	The municipal police agency.
year_concluded	The year the case concluded.
incident_year	The year of the incident, where the digest records it.
statutory_heading	The Police Act heading the allegation was filed under, as published.
opcc_sublabel	The OPCC's own keyword sub-label.
conduct_category	The machine-coded conduct category, one of fourteen; checks on the methods page.
is_gbv	The wide gender-based violence flag: any of the four rules.
is_gbv_narrow	Rules 1 and 2 only: conduct an officer committed.
is_gbv_reviewed	Rule 1: flagged on review and upheld by the refuting pass.
is_gbv_deferred	Rule 2: deference to the OPCC's own sexualized-conduct label.
is_gbv_hand_added	Rule 3: the single hand-added allegation.
is_gbv_neglect	Rule 4: neglect or mistreatment of a woman who reported violence.
dismissed	A dismissal is recorded on this allegation.
job_level_measure	A measure reaching the officer's job is recorded; the exact rule is stated in reproduce.R.
suspension_days	Suspension length in days, where recorded.
months_to_decision	Months from incident to conclusion, where both dates exist.
appealed	The digest records an appeal or review.
pathway	How the case entered: OI ordered investigation, PC public complaint, ID internal discipline, SH serious harm.
criminal_process	The digest records a parallel criminal process.
no_charge_approved	The digest records that Crown counsel approved no charge.
criminal_process_probe	The machine probe behind criminal_process; kept so the column can be audited.
no_charge_approved_probe	The machine probe behind no_charge_approved; kept so the column can be audited.
trust_language	The published text uses public-trust language.
allegation_members	Members named on this allegation, as published.
members_in_file	Members named anywhere in the file.
measure_members_named	Members the recorded measure names.
measure_is_member_split	The measure line assigns different measures to different members.
text_length	Character length of the published summary.
msr_*	One column per statutory measure (suspension, written reprimand, training, dismissal, advice as to future conduct, verbal reprimand, close supervision, counselling, program or activity, reduction in rank, transfer or reassignment): 1 where that measure is recorded on the allegation.

cases.json 1.8 MB

The published summary text and the supporting quote for each allegation, as shown in the browser on the decisions page.

dismissal_by_heading.csv 224 rows · 20 KB

Every bar in the four figures on the findings page, one row per bar: the outcome, whether rows count cases or allegations, the statutory heading, which of the two groups the bar shows, the count behind it, and its 95 per cent interval.

decision_pdf_index.csv 576 rows · 110 KB

The 576 documents on the OPCC's hearing and review pages (decisions, notices, submissions, rulings), one URL per row, with listing and file size.

reproduce.R 28 KB

An annotated base-R script that recomputes the main findings from allegations.csv and prints each one beside its rule. No packages.

$ data & code run real R in this tab no install · R, Python, curl, files, citation

static files · CORS open · no key · no rate limit

A real R session, running in your browser through webR. Nothing is sent anywhere. The table is loaded as allegations, one row per allegation. The first run downloads R itself, about 30 MB.

# Ron Levi <https://individual.utoronto.ca/ronlevi/>, Community Safety and
# Policing Research Program, Global Justice Lab, Munk School of Global
# Affairs and Public Policy, University of Toronto.
#
# The table is already loaded as `allegations`, one row per allegation.

# THE HEADLINE COMPARISON, COUNTED BY CASE.
# WHY max(): a case holds several allegations. Taking the max of is_gbv and
# dismissed within each file_number asks two yes/no questions per case:
# does it hold any gender-based violence allegation, and did any allegation
# in it end in dismissal.
per_case <- aggregate(cbind(flagged = is_gbv, fired = dismissed) ~ file_number,
                      allegations, max)

# WHAT TO LOOK FOR: the `1` column. Cases with a gender-based violence
# allegation (flagged = 1) end in dismissal about five times as often.
round(100 * prop.table(table(flagged = per_case$flagged,
                             fired = per_case$fired), 1))

# THE FOUR FLAG RULES, KEPT SEPARATE. Each column is one rule from the
# "What we counted" page, so any rule can be taken out and the numbers rerun.
colSums(allegations[, c("is_gbv_reviewed", "is_gbv_deferred",
                        "is_gbv_hand_added", "is_gbv_neglect")])

# WHAT THIS DOES NOT SHOW: why the gap exists. These are associations in an
# administrative record. reproduce.R (on this page) recomputes the main
# findings with every definition spelled out.

R has not started yet. Press Run.

Press Run. The first run takes a moment while R downloads.

# the annotated replication script recomputes the main findings:
#   Rscript reproduce.R
allegations <- read.csv("https://bc-police-discipline.pages.dev/data/allegations.csv")

# dismissal by case: max() within a file asks whether the case holds any
# gender-based violence allegation, and whether any allegation ended in dismissal
per_case <- aggregate(cbind(flagged = is_gbv, fired = dismissed) ~ file_number,
                      allegations, max)
round(100 * prop.table(table(per_case$flagged, per_case$fired), 1))

# pandas
import pandas as pd
a = pd.read_csv("https://bc-police-discipline.pages.dev/data/allegations.csv")

per_case = a.groupby("file_number")[["is_gbv", "dismissed"]].max()
print(per_case.groupby("is_gbv")["dismissed"].mean().round(2))

BASE=https://bc-police-discipline.pages.dev/data
curl -O $BASE/allegations.csv
curl -O $BASE/cases.json
curl -O $BASE/dismissal_by_heading.csv
curl -O $BASE/decision_pdf_index.csv
curl -O $BASE/reproduce.R
curl -O $BASE/bc_police_discipline_complete.zip

@misc{levi2026bcpd,
  author      = {Levi, Ron},
  title       = {Police Discipline in British Columbia: the published
                 record, 2010--2025},
  institution = {Community Safety and Policing Research Program,
                 Global Justice Lab, Munk School of Global Affairs and
                 Public Policy, University of Toronto},
  year        = {2026},
  url         = {https://bc-police-discipline.pages.dev}
}