Built on 6.9M flights · 50 airports · 300 routes

Flight Check

A Flight Check Investigation

Spurious Flight Correlations

We analyzed 119M+ BTS flight records across 12 years and 50 airports, then compared monthly patterns to drowning deaths, watermelon shipments, beer production, and lemonade searches. None of these correlations mean anything.

12

suspicious correlations

0.47

avg |r| value

0

causal relationships

Jan 2022 – Nov 2025 · 47 months · Bureau of Transportation Statistics · 50 airports

How to read these charts: Each card shows a real U.S. flight metric (blue or red line, left axis) month by month from Jan 2022 to Nov 2025, plotted against a completely unrelated dataset (colored line, right axis) with its seasonal pattern tiled across the same period. The r value measures how closely the two lines move together. An r of ±1.0 = perfect correlation. An r of 0 = no relationship. And none of it is causal — it's just what happens when seasonal patterns line up.

r = 0.64Moderate correlation

U.S. drowning deaths predict how late your flight will be

More drownings = longer delays. The waves repeat every year.

Avg delay when late (BTS)
US drowning deaths
Source: CDC WONDER Database (2023)
r = 0.59Moderate correlation

Google searches for "sunscreen" predict flight delays

The more people Google sunscreen, the longer your delay.

Avg delay when late (BTS)
"Sunscreen" search interest
Source: Google Trends (2024)
r = 0.61Moderate correlation

Watermelon supply predicts flight delay severity

Peak watermelon = peak delay minutes. Year after year.

Avg delay when late (BTS)
Watermelon supply
Source: USDA AMS Movement Reports (2024)
r = 0.56Moderate correlation

Lemonade search interest tracks delay minutes

Big Lemonade is coming for your travel plans.

Avg delay when late (BTS)
"Lemonade" search interest
Source: Google Trends (2024)
r = 0.59Moderate correlation

Your electricity bill predicts how late your flight will be

Higher electric bills, longer delays. The grid knows.

Avg delay when late (BTS)
Avg residential electricity bill
Source: EIA Electric Power Monthly (2024)
r = -0.63Moderate correlation

U.S. beer production rises as on-time performance drops

Breweries ramp up. Flights fall apart. Coincidence?

On-time arrival rate (BTS)
Beer produced
Source: TTB Beer National Report (2024)
r = 0.45Weak correlation

National Park visits mirror flight delay severity

When America goes to the parks, delays get longer.

Avg delay when late (BTS)
National Park visits
Source: NPS Visitor Use Statistics (2024)
r = 0.40Weak correlation

Flu season predicts flight cancellations

Peak flu = peak cancellations. Sick passengers or sick coincidence?

Cancellation rate (BTS)
Influenza positive tests
Source: CDC FluView (2024–25 season)
r = 0.39Weak correlation

U.S. ice cream production correlates with flight delays

When America makes more ice cream, delays get longer.

Avg delay when late (BTS)
Ice cream produced
Source: USDA Dairy Products Annual Summary (2024)
r = 0.38Weak correlation

Pumpkin spice searches vs. on-time arrivals

PSL season arrives, and so do the planes. Kind of.

On-time arrival rate (BTS)
"Pumpkin spice" search interest
Source: Google Trends (2024)
r = 0.27Basically nothing

Heating degree days correlate with flight cancellations

The colder it gets, the more flights get cancelled. At least this one makes sense.

Cancellation rate (BTS)
Heating degree days
Source: EIA Monthly Energy Review (2024)
r = 0.14Basically nothing

Shark attacks vs. flight delays

You'd think this would work. It really doesn't.

Avg delay when late (BTS)
US shark attacks
Source: International Shark Attack File, Florida Museum

So what's actually going on?

All of these correlations are driven by the same hidden variable: seasonality. Flight delays peak in summer because of thunderstorms, heat-related ground stops, and higher passenger volume. Ice cream, shark attacks, sunscreen purchases, and pool drownings also peak in summer.

This is why “correlation does not imply causation” exists as a phrase. Two things can move in perfect lockstep without one causing the other. The next time you see a scary-looking chart with a trend line, ask: is there a lurking variable?

Inspired by Tyler Vigen's Spurious Correlations. Flight data from the Bureau of Transportation Statistics. Everything else is illustrative — seasonal patterns are real, exact values are approximate.

Get the next data report when it drops.