Hello, and welcome to the Q&A for Episode 9 of Think Like a Debater.

A：Consider successful people again. Steve Jobs, Bill Gates, and Mark Zuckerberg left university to start businesses. Among five founders of the world’s most successful technology companies, three dropped out: 60%.

“That percentage is enormous!”

“It must be telling us something!”

“University is useless; I should drop out and start a company too.”

The conclusion intuitively feels biased, extreme, and based on exceptional cases.

It is survivorship bias. Across the full population, university graduates have advantages in employment and average income. For an individual seeking the best chance of work and earnings, attending university remains better than not attending.

A subtler example appears in bestsellers explaining corporate success. The 1982 book In Search of Excellence sold over three million copies after identifying eight shared traits among forty-three outstanding companies.

Years later, economics professor Gary Smith noted in his 2014 book Standard Deviations that thirty-five were public companies whose performance could be tracked. More than half—twenty—later performed below average.

Finding common traits among forty-three excellent companies sounds reasonable, but it repeats the logic of studying Jobs, Gates, and Zuckerberg: survivorship bias.

Successful people and companies form a slanted, unrepresentative sample. Ignoring failures makes data incomplete and nonrandom, biasing conclusions.

A：I recently stayed at a highly rated homestay. Its strengths and weaknesses were obvious: quiet but inconvenient; artistic but impractical, with a window that leaked cold air; two lovely cats, but a shared bathroom with an eccentric neighbor.

It was decent, but such pronounced drawbacks should have lowered the average rating.

Two days later I understood. When I mentioned the drafty window, the host said the property could not meet my needs and suggested I leave for a negotiated refund.

That revealed the selection bias: whenever an unsatisfied guest raised an unsolvable problem, the host encouraged cancellation, excluding that guest’s rating from the average. No wonder reviews were uniformly positive.