Research · 4 min read
After Dark? Midnight Timestamps Flip the Answer
Timestamp precision changes the direction of the result, so hour, season, offense mix, and property exposure matter more than an after-dark headline.
It sounds like the simplest question in security: does property crime go up after dark? In Seattle’s public data, the answer hinges on one odd detail — 10,156 records stamped at exactly midnight. Count them literally and night looks 14.0% worse than day. Handle them honestly and night looks 6.9% better. Here’s what that means for when coverage should actually start.
Our analysis of 116,814 accepted property-offense records, July 1, 2023–June 30, 2026.
Key takeaways
- Read literally, the night rate was 14.0% above daylight across 116,814 accepted Seattle property-offense records from July 2023 through June 2026.
- 10,156 records — 8.7% of the sample — are stamped at exactly midnight; excluding them, or marking them unknown, puts the night rate 6.9% below daylight.
- Even redistributing the midnight rows lands at 4,269 records per 1,000 night-hours, roughly level with the daylight rate of 4,187.
- This dataset does not support a simple sunset rule; hour, season, offense mix, and the property’s own hours carry more information.
Does sunset provide a defensible cutoff for security coverage?
Across 116,814 accepted Seattle property-offense records from July 2023 through June 2026, the recorded night rate is 14.0% above daylight when timestamps are used literally, but it becomes 6.9% lower after 10,156 exact-midnight records are excluded. This dataset does not support a simple sunset rule.
The numbers behind the answer
Selected measures only. Denominators and interpretation stay attached so the headline cannot stand alone.
1.140
Night/day rate ratio as recorded
Night records per 1,000 night-hours were 14.0% higher when exact-midnight timestamps were accepted literally.
0.931
Night/day ratio without midnight
The direction reverses after excluding 10,156 records stamped exactly 00:00.
10,156
Exact-midnight records
8.7% of all 116,814 accepted records share an exact-midnight start time.
One timestamp decision changes the direction of the answer
Run the timestamps exactly as recorded and the answer seems obvious. Across three years and 116,814 accepted records, night hours produced 4,771 offenses per 1,000 night-hours against 4,187 per 1,000 daylight-hours — a night rate 14.0% higher. Start coverage at sunset, apparently.
Now look at the timestamps themselves. 10,156 records — 8.7% of everything — are stamped at exactly 00:00. Some are real midnight events. A pile-up that size usually means the true time was never recorded. Exclude those rows, or mark them unknown, and the night rate falls to 3,898 per 1,000 night-hours: 6.9% below daylight. The conclusion doesn’t just weaken. It flips.
Bottom line
This extract does not support a simple rule that coverage should begin at sunset.
The midnight records are not a footnote
When a tenth of the sample decides the answer, that decision belongs in the headline — so we ran the comparison four ways. Keep the midnight records as recorded. Exclude them. Treat them as unknown. Redistribute them across the clock in proportion to records with trustworthy times.
All three careful treatments erase the apparent nighttime elevation. Even redistribution — the handling friendliest to the after-dark theory — lands at 4,269 per 1,000 night-hours, roughly level with daylight. That agreement is stronger evidence than the tidy confidence interval around the literal reading.
Evidence visual
Night records per 1,000 night-hours under timestamp treatments
Daylight remains the comparison rate. Excluding, treating as unknown, or redistributing exact-midnight rows changes the conclusion.
A useful schedule needs more than sunrise and sunset
So when should coverage start? Honestly, this dataset can’t tell you — and that is the finding. Hour, season, weekday, and offense type all carry more information than a daylight label, and Seattle’s long summer days mean “after dark” slides hours across the clock depending on the month.
The practical move is to stop scheduling from a slogan. Take the hourly pattern, overlay your property’s own rhythms — closing time, deliveries, occupancy — and review it a couple of times a year instead of freezing one rule. The checklist below is the working version.
One more nuance: darkness is never the only thing that changes after sunset. Occupancy falls, deliveries stop, staff go home, and the mix of activity shifts — any of which can move exposure more than the sun does. That is why the hourly and seasonal pattern belongs beside a site owner’s own operating rhythm, not against a single solar boundary that slides across the clock with the seasons.
- Separate exact times from imprecise or windowed times.
- Compare clock hour and time relative to sunset.
- Check whether the pattern changes by month and offense subtype.
- Add property hours, deliveries, occupancy, and access activity.
- Review the schedule periodically rather than freezing one rule.
The clock pattern is real, but it tracks activity more than darkness
Strip the midnight rows out and a within-clock shape survives. Records climb from 3,322 in the last twelve hours before sunset to 6,577 in the first hour after it, peak at 7,234 in the three-to-four hours after sunset, then fall back to 3,098 by the final pre-sunrise window. That shape does not depend on the 10,156 exact-midnight placeholders. It also does not isolate darkness: the evening peak is when occupancy, traffic, and business activity are also highest, so the curve is as consistent with routine activity as with the sun.
Where the evidence gets stronger is light as a control, not light as a schedule. A Campbell systematic review of improved street lighting found modest, consistent reductions in night-time crime — the guardianship effect again — but the effect sizes vary by area and offense, and the most recent CCTV meta-analysis shows cameras cut some offense types, notably in car parks, far more reliably than others. Neither result supports a blanket claim that light or cameras stop after-dark crime.
So the actionable version is smaller and more specific than a sunset rule: audit which approaches, lots, and entrances are genuinely dark during the hours the property is exposed, and pair that with the coverage map to see where records already recur. The study’s own numbers say the clock hour matters — not a solar boundary that slides across the clock with the season.
Better question
Ask which specific dark areas carry exposure — not whether crime rises after sunset.
Questions property teams ask
Does 51.9% outside daylight mean most property crime happens at night?
No. That share combines night and civil twilight and ignores how many calendar-hours each phase contains. Rate comparisons are more informative, but they remain sensitive to timestamp precision.
Why normalize by clock-hours?
Seattle has unequal daylight and darkness across seasons. Dividing by phase-hours avoids comparing raw counts accumulated across different amounts of time.
Does the result prove darkness causes offenses?
No. The study is descriptive and does not control for occupancy, activity, season, property exposure, or other factors that change with time.
Should coverage begin at sunset?
This analysis does not support that rule by itself. Exact-midnight handling reverses the observed comparison, so a property schedule should use local hourly and operating patterns.
If darkness isn’t the driver, does lighting still matter?
Possibly, modestly, and locally. Lighting changes guardianship and visibility at specific places; the evidence supports targeted improvements at exposed spots, not a blanket after-dark rule.
What should drive a coverage schedule?
The property’s own hourly exposure — closing times, deliveries, occupancy, and access activity — reviewed against current data and refreshed periodically, rather than a fixed sunset start.
Our methods, limits, and sources
How we calculated this
This is original 911 Sentinel research — we gathered the records, ran every calculation below, and published the aggregate dataset.
We matched accepted Seattle property-offense start timestamps to USNO sunrise, civil-twilight, and sunset boundaries using one central Seattle coordinate.
- We classified 116,814 accepted records as daylight, civil twilight, or night.
- We calculated available calendar-hours in each light phase and reported records per 1,000 phase-hours.
- We compared night with daylight using a rate ratio and interval.
- We repeated the calculation with exact-midnight rows retained, excluded, treated as unknown, and redistributed.
What this analysis cannot establish
- Recorded start time may be estimated or may begin a broader occurrence window.
- Exact-midnight values can encode real events or missing precision; the source does not distinguish them.
- Calendar-time is not a denominator for property, population, vehicle, or operating exposure.
- One central coordinate supplies solar boundaries for all Seattle records.
- The association does not establish that darkness caused an offense.
Sources
The raw records come from the sources below; the study design, analysis, charts, and conclusions are our own.
- Seattle Police Department — SPD Crime Data: 2008-Present
- U.S. Naval Observatory — Astronomical Applications API v4.0.1
- Published aggregate result (JSON)
- Source manifest (JSON)
- Brandon C. Welsh & David P. Farrington (Campbell Systematic Reviews) — Effects of Improved Street Lighting on Crime
- Eric L. Piza, Brandon C. Welsh, David P. Farrington & Amanda L. Thomas (Criminology & Public Policy) — CCTV surveillance for crime prevention: A 40-year systematic review with meta-analysis
- Lawrence E. Cohen & Marcus Felson (American Sociological Review) — Social Change and Crime Rate Trends: A Routine Activity Approach
Related questions and practical guides
No obligation · free property walk
Test the hours that matter at your property
A property review can combine public time patterns with lighting, deliveries, access activity, vacancy, and current operating hours.