Doob T-539-20 Federal Court affidavit replication

Part of Statistical Methods — MORIE’s statistical-methods reference.

morie.doob_trends replicates the analytical contribution of Prof. Anthony N. Doob’s expert-witness affidavit in Canadian Civil Liberties Association et al. v. The Attorney General of Canada, Federal Court file T-539-20 (Application Record Vol. 3 of 5, pp. 778-795).

The affidavit was filed in 2020 in support of the application for COVID-19-driven prisoner release; it presents 4 figures + 3 tables arguing that imprisonment rates are decoupled from crime rates and that conditional release is empirically safe.

Three CCRSO 2018 tables hardcoded

The module hardcodes the three numerical tables from Doob’s affidavit, sourced from Public Safety Canada’s Corrections and Conditional Release Statistical Overview 2018:

  • CCRSO_TABLE1_RELEASES — affidavit Table 1: 5-year average annual conditional releases by type.

  • CCRSO_TABLE2_FLOW — affidavit Table 2: annual prisoner counts, admissions, deaths, releases (2013/14-2017/18).

  • CCRSO_TABLE3_AGE — affidavit Table 3: age distribution (Canada adult population vs CSC custody).

from morie.doob_trends import (
    analyze_doob_table1_releases,
    analyze_doob_table2_flow,
    analyze_doob_table3_age_overrepresentation,
    analyze_doob_full_affidavit,
)

r = analyze_doob_table1_releases()
# Headline: violent revocation rate < 1% across all release types

r = analyze_doob_table3_age_overrepresentation()
# Headline: age 60+ is 29.5% of Canadian adult pop but only 9.3%
# of CSC custody (IRR_custody = 0.32)

Decoupling test

Doob’s central thesis (§§ 16-21 of the affidavit) is that imprisonment rates and crime rates are not consistently correlated. decoupling_test() tests this with Pearson r + Pettitt change-point:

from morie.doob_trends import decoupling_test

decoupling_test(crime_rate_series, imprisonment_rate_series,
                 years=range(1960, 2018))
# RichResult with r_pearson, two-sided p, Pettitt change-point per
# series.

pettitt_changepoint(series) exposes the Pettitt 1979 non- parametric change-point test as a standalone helper.

Note on Figures 1-4

Doob’s affidavit also contains four time-series figures (Canadian crime rate 1960-2018, Canadian imprisonment rate 1960-2018, Canada vs US imprisonment 1950-2017, Canada vs US homicide 1961-2017). These plots use StatsCan CANSIM data + Pastore-Maguire US Sourcebook of Criminal Justice Statistics. The data series themselves are not bundled with morie; users supply their own time-series for use with decoupling_test().

Position in the MRM stack

The Doob national-aggregate analyses complement the OTIS provincial evidence:

  • Federal national (this module): Tables 1-3 + decoupling.

  • Federal SIU regional (morie.sprott_doob): Mandela classifier + regional Pacific-vs-Ontario disparities.

  • Provincial individual + aggregate (morie.otis_*): Ruhela formulations ensemble on OTIS Ontario data.

analyze_ruhela_master() §5 surfaces key rows from this module.

Citation

from morie.siuiap import cite
cite("doob_t_539_20_2020")
# Anthony N. Doob (2020). Affidavit of Anthony N. Doob — Federal
# Court of Canada, T-539-20. Federal Court of Canada.