No difference

"Full moon nights are wild" — the view from the field

Police officers, paramedics, and ER staff around the world have passed down the same observation: full moon shifts get rough. The term "lunar effect" exists for a reason, and plenty of first responders still brace for full-moon nights. The instincts of people who have spent decades on the street don't deserve to be laughed off from a desk. Which is exactly why this page tests the belief on the field's own turf — actual crash records.

On paper, you can build a plausible story: full-moon nights are brighter, maybe more people go out; moonlight might catch a driver's eye; some studies suggest people sleep worse. But the question is not whether a story can be told. It is whether hundreds of thousands of crash records actually carry its fingerprint.

Some research actually found an increase

In 2017, the BMJ published a study finding about 5% more fatal motorcycle crashes in the US on full moon nights (Redelmeier & Shafir) — the authors' hypothesis being not moonlight itself but the glance toward the moon. Most studies of overall traffic accidents and moon phase, on the other hand, find no significant difference. In other words, this is not a long-settled question: conclusions have swung with the population and method studied. Our verdict casts the widest possible net — all injury accidents in Japan and all fatal crashes in the US.

How this verdict is computed

• Japan: daily counts of all injury accidents from the National Police Agency's open data (since 2019)
• United States: daily counts of all fatal crashes from NHTSA FARS (since 2000). Note the metrics differ: injury accidents for Japan, fatal crashes for the US
• Accident counts have strong weekday rhythms (commutes, weekends) and seasonality. The expected value is therefore the average for the same weekday × month, and we compare observed ÷ expected — the same adjustment used on the births and FX pages
• Each day is classified as a full moon day (±24h around the instant) or new moon day by the moon's age at local noon (JST for Japan, US Eastern for the US), and each group's mean index is compared to baseline (1.00)

See the methodology page for the verdict criteria.

What we can't see yet

This result doesn't mean "the moon has nothing to do with it." Here is what the current method can't see:

• Night hours: the folklore is about the full-moon night. Our data is daily, so a rise confined to a few night hours wouldn't show
• Type of crash: the BMJ study's increase was in fatal motorcycle crashes. Counting every crash together dilutes a movement like that
• Whether the moon was visible: nobody's eye is caught by a full moon behind clouds. For now we count every night regardless of the sky

What we want to check next

• Use crash times to compare night hours only
• Check whether US fatal motorcycle crashes alone reproduce the BMJ result
• Compare only full-moon nights with clear skies

Perhaps the field's "full-moon nights are wild" holds something we haven't yet turned into numbers. We'd like to find it in the data one day.

Sources

• National Police Agency of Japan, traffic accident open data
• NHTSA FARS (Fatality Analysis Reporting System)
• Redelmeier DA, Shafir E. "The full moon and motorcycle related mortality" BMJ 2017;359:j5367
• Moon phases computed in-house from Jean Meeus, "Astronomical Algorithms" (UTC)