Loading the Chicago analysis…
The findings
Halloween traffic safety in Chicago
Completed calendar years 2018–2025 · Snapshot October 5, 2026 · Police-reported crashes within CPD jurisdiction
01 · Calendar patterns
Does October 31 stand out?
Each bar sums one calendar date across the same eight complete years. The black marker identifies Halloween. This describes reported harm; it is not a risk per trip or per pedestrian.
02 · A fairer comparison
Halloween versus nearby days of the same weekday
The comparison for each year is the average of up to eight days: the same weekday one, two, three and four weeks before and after Halloween, excluding Thanksgiving. Both series show counts for one day.
Halloween Matched-day average
| Year | Weekday | Halloween | Matched average | Difference |
|---|
Earlier Halloweens: coverage matters
Halloween 2017 has citywide reporting and is shown as a supplemental comparison. The principal calendar analysis uses 2018–2025 so every calendar date has complete coverage in the same eight years. Records for 2015–2016 cover only part of Chicago, so their counts cannot establish a citywide spike.
| Year | Coverage | Crashes | Injuries | Incapacitating | Fatalities | Children walking injured |
|---|
When does the harm happen?
October 31 only, summed across 2018–2025. Ages 0–17 define children. The shaded hours are 4–10 p.m.; select the evening measure to compare that window directly.
03 · Place and protection
Halloween across Chicago’s wards
Explore reported harm and full-closure permits by ward. Historical years use the boundaries in effect on that Halloween. Ward numbers are not a constant geography across the 2023 remap.
Loading map…
Halloween weekend 2026 permits
Applications overlapping October 30–November 1 as of October 5. More may arrive before Halloween. This is a planning snapshot; no 2026 Halloween crash outcome exists yet.
| Date | Street start | Ward | Milestone | Closure |
|---|
04 · Evidence and limits
Methods, coverage and reproducibility
What we count
Crashes are distinct crash records. Total injuries include fatal, incapacitating, non-incapacitating, and reported but not evident injuries. “Serious” here means the police classification of incapacitating injury. Fatalities are reported deaths, not fatal crashes. Pedestrian and child outcomes use the People table; children have recorded ages 0–17. Missing ages are not assigned to an age group.
People are aggregated to crash ID before joining. Overall injuries use crash totals; pedestrian and child injuries use person classifications. Any disagreements are reported in the audit. Timestamps are Chicago local civil time.
Which records qualify
Crashes have citywide CPD coverage from September 2017. The principal comparisons use complete 2018–2025 years. Earlier partial coverage and the ongoing 2026 year are excluded. Interstate, ramp, and boundary-road crashes handled by other agencies may be absent. Reports can be amended, so this is a dated snapshot.
A car-free permit must be a BlockParty application, have an issue date and a Complete milestone, and specify Full street closure. Denied and cancelled applications are excluded. We exclude malformed dates and durations of three days or more. Permits issued after the event’s first calendar day are excluded. Distinct applications are counted once per ward and calendar day, even when they have several location records.
Statistical details and robustness
The calendar chart excludes February 29 so every plotted date has eight years of exposure. Rankings use competition ranks and report ties. The weekday comparison is a descriptive count ratio, not an exposure-adjusted risk ratio. Calendar-placebo tests reassign the event label among the Halloween date and its matched dates in each year, using 50,000 draws; those dates are not randomized treatment assignments. Placebo values are exploratory.
The ward maps change May 18, 2015 and May 15, 2023. We retrieve 2003, 2015 and 2023 polygons, discard the 2003 “OUT” polygons, and assign longitude/latitude points to numbered wards. A point exactly on a shared boundary is assigned the lowest matching ward number. Permit points describe the start of an address range, which can cross a ward line.
Source quality audit
Sources and downloads
- Traffic Crashes — Crashes · Entire CSV export
- Traffic Crashes — People · Entire CSV export
- CDOT permit archive · All BlockParty rows, including historical and future applications
- Current/future block party view · Retrieved and deduplicated by location key
- Legacy permit archive, 2011–2015 · Retrieved separately; overlaps modern converted records and is excluded from the post-2017 model
- Ward boundaries: 2003, 2015, 2023. Effective dates: CMAP ward history.
Daily aggregate CSV · Ward-day aggregate CSV · Chart data, estimates and source checksums · Reproduce with DuckDB
Maps use Protomaps and OpenStreetMap. The national chart supplied as a reference informed the calendar layout; its national estimates are not used in this Chicago analysis.