MRM-framework analyses on Ontario OTIS data

Five callables for the OTIS (Offender Tracking Information System) public-release datasets, used in the MRM (Multilevel Reconciliation Methodology) empirical companion paper. Every analysis is computed directly from the OTIS CSV files; no precomputed artifacts are required.

Value

Each mrm_otis_*() callable returns a named list with the computed statistics (concentration indices, survival curves, or association measures) and a plain-language interpretation.

Details

Functions:

• mrm_otis_placement_concentration(): Hill-MLE Pareto exponent + Gini coefficient + top-k% concentration on the b09 per-individual placement-count distribution (within-fiscal-year).

• mrm_otis_seg_duration_km(): Kaplan-Meier survival on the b01 NumberConsecutiveDays_Segregation durations (per-placement; strata = alert profile).

• mrm_otis_mortification_cooccurrence(): pairwise Cramer's V across the three b01 alert flags (MentalHealth, SuicideRisk, SuicideWatch).

• mrm_otis_region_locality(): chi-square + Cramer's V on the Region_AtTimeOfPlacement x Region_MostRecentPlacement contingency table, with the diagonal/off-diagonal share.

• Plus the existing mrm_classify_mandela() (in mandela.R).

The OTIS UniqueIndividual_ID column has format YYYY-XXXXX-SG and is randomly reassigned every fiscal year. Cross-year tracking is therefore invalid; all analyses below operate within fiscal year.

Examples

if (FALSE) {
  b09 <- read.csv("b09_individuals_in_segregation.csv")
  mrm_otis_placement_concentration(b09)
}