4.7/5 - (48 votes)

You have probably said it yourself: the handsome ones are never nice, and the nice ones are never handsome. The pattern feels real, yet it can appear even when the two traits have nothing to do with each other.

The trick hiding in your dating pool

Mathematician Jordan Ellenberg describes a dating pool measured on two axes, niceness and handsomeness. A person looks at their own experience and concludes “the handsome ones tend not to be nice, and the nice ones tend not to be handsome.”

Now imagine Alice, who only dates men whose niceness plus handsomeness passes a certain threshold. A very nice man does not need to be as handsome to get in. A very handsome man does not need to be as nice. Inside her pool, the nicer men are therefore less handsome on average, and the reverse.

The trade-off is real inside the pool. It does not exist in the population.

Alice actually has high standards

Here is the twist: the men Alice dates are better than average, not worse. The average nice man in her pool is more handsome than the average man out there, because even among nice men, the least handsome ones were skipped.

The negative link comes from the filter alone. The rude men who made it in had to be even more handsome, and the plainest ones had to be even nicer.

The stamp collector’s display case

The same effect shows up with numbers. A collector owns 1,000 stamps. Of these, 300 are pretty, 100 are rare, and 30 are both. So 10% of all stamps are rare, and 10% of the pretty ones are rare too. Prettiness tells you nothing about rarity.

See also  Four datasets share the same averages and correlation, yet look nothing alike

The collector then puts the 370 stamps that are pretty or rare on display. Just over 27% of the display is rare (100 out of 370), yet only 10% of the pretty stamps are rare (30 out of 300). And all 70 of the not-pretty stamps in the case are rare.

A visitor who only sees the display would swear that plain stamps are the rare ones. The case is the filter, and the filter invents the pattern.

Where it was first spotted

The effect carries the name of Joseph Berkson, an American biostatistician who trained as a physicist, a physician and a statistician. He identified how selection effects can bias observational studies, and the bias eventually became known as Berkson’s paradox.

His original illustration came from hospital data. Take diabetes as a risk factor and cholecystitis as the disease. A hospital patient without diabetes is more likely to have cholecystitis than a random member of the general population, because that patient needed some other reason to be admitted in the first place. The result appears whether or not the two conditions are linked outside the hospital.

The same shape, everywhere

A person may notice that attractive celebrities tend to be untalented, and talented ones unattractive. But people who are neither attractive nor talented rarely become celebrities, so they never enter the observer’s view.

The rule behind all of these cases is simple. Two independent events become dependent once you only look at cases where at least one of them occurs. With 100 cases where A and B each happen half the time, A occurs in 50 of the 75 cases where A or B happens: 2 out of 3, instead of 1 out of 2.

See also  No, the first known census did not count people at all

It is a close cousin of survivorship bias, where only the winners are counted.

Before you trust a trade-off

Next time two good things seem to come in opposite packages, ask one question: who got left out of the sample? If the answer is “everyone who had neither,” the trade-off may live in the filter, not in the world.