KM survival of SIU case-to-decision gap per police service

Computes the gap (in days) between the incident date and the Director's decision date for every SIU case, dropping rows where either date is missing. Reports per-stratum median + IQR + n. Cases without a decision date as of the snapshot are right-censored if censor_open_cases = TRUE (default).

Usage

mrm_siu_case_to_decision_km(
  data,
  incident_col = "date_of_incident_iso",
  decision_col = "date_of_director_decision_iso",
  service_col = "police_service",
  censor_open_cases = TRUE,
  min_n = 5L
)

Arguments

data

A data.frame with the SIU case schema.

incident_col

Column with ISO incident date (default "date_of_incident_iso").

decision_col

Column with ISO Director's decision date (default "date_of_director_decision_iso").

service_col

Stratifying jurisdiction column (default "police_service").

censor_open_cases

Logical (default TRUE). If TRUE, rows with missing decision_col contribute right-censored observations from incident to the most recent decision date in the data set. If FALSE, they are dropped.

min_n

Minimum cases per service to retain in the per-service summary (default 5L).

Value

A list with elements:

• pooled: a single-row data.frame with the pooled median, mean, IQR, n, n_censored.

• by_service: per-service data.frame with the same columns.

Details

This is the substantive "time-to-outcome" analysis the MA-thesis "210-day TTR" claim should have been; it operates on real per-case dates with a stable jurisdiction identifier.

Examples

if (FALSE) {
  siu <- read.csv("SIU.csv")
  res <- mrm_siu_case_to_decision_km(siu)
  head(res$by_service)
}