Skyscraper Cats and High-Rise Syndrome

In 1987, a fascinating study was published in the Journal of the American Veterinary Medical Association (JAVMA) by researchers Whitney and Mehlhaff. They examined 132 cats that had fallen from high-rise buildings in New York City.

The data revealed a shocking, counterintuitive pattern: cats that fell from higher floors (7 to 32 stories) had a higher survival rate and sustained fewer severe injuries than cats that fell from lower floors (2 to 6 stories).

Of the cats that fell from 9 stories or higher, 95% survived. One cat even fell from 32 stories onto concrete and walked away with only a chipped tooth and a collapsed lung.

The physics theory

To explain this bizarre result, the researchers proposed a physics and biology hypothesis:

• During the first few stories of a fall, a cat is accelerating under gravity and feels fear, causing it to tense its muscles, which increases the impact force.
• After falling about 5 to 6 stories, the cat reaches its terminal velocity (about 60 mph or 97 km/h) where gravity is balanced by air resistance.
• At this point, the cat no longer accelerates, its vestibular system relaxes, and it extends its limbs horizontally like a flying squirrel. This distributes the impact force across its entire body and allows it to absorb the shock much better.

For years, this theory was accepted as scientific fact and cited in physics textbooks.

The statistical flaw

In 2002, statisticians and skeptics pointed out a massive flaw in the study’s design: survivorship bias in the data collection.

The study’s data was collected exclusively from cats brought to emergency veterinary clinics. Consider the difference in owner behavior based on the height of the fall:

• If a cat falls from the 3rd floor, it has a high chance of surviving but may have broken bones. The owner will almost certainly take the cat to the vet.
• If a cat falls from the 25th floor and dies instantly on impact, the owner does not take the dead cat to the emergency clinic. They contact animal control or dispose of the body directly.
• The vet clinic only sees the cats that fell from the 25th floor and miraculously survived.

Because the dead cats from high floors were never recorded in the veterinary dataset, the database was missing the failures. The apparent safety of falling from high floors was a statistical illusion caused by a highly skewed sample set.

Key takeaway

A dataset that relies on voluntary reporting (like visiting a clinic or answering a survey) will almost always suffer from selection bias. Before drawing conclusions from a correlation, check if the worst-case outcomes are simply being excluded from the records.