We all have cognitive biases. Most often, these biases stick us with nothing more than bad cellphone service or a little gambling debt. Occasionally, though, they cause us to accuse our neighbors of espionage.

The reason why is called the Clustering Illusion. Randomness isn't obvious, and there's a way to demonstrate that. An interesting exercise, given by many math teachers, involves two groups of people making two records of the results of a series of coin flips. One group actually flips a coin fifty times and records the results, and the second group fakes a record. It's remarkably rare for the false record to look anything like a genuine one. The false record will have runs of two or three tails, before the fakers get nervous and switch back to heads. They think that clusters of similar outcomes only happen when the data isn't random. The real record will have long runs, often of seven or eight tails in a row. Although if you look at those clusters, the coin looks weighted, it's just the result of random chance. Clustering of certain outcomes is just another aspect of randomness.

It's tough to separate out the difference between actual random events and events that seem random, but follow certain trends. In the 1940s, Londoners learned this to their cost. The city was being regularly bombed by rockets. People were afraid, and looking for anything that might help them stay alive. When papers started publishing maps of the bomb sites, people noticed certain clusters, and certain empty spaces, and decided that the Germans had targeted some neighborhoods and ignored others. Analysts, then and now, don't believe that to be the case. The bombs fell randomly, and randomly clustered in certain areas.


Believing the clustering, and the neglected areas, to be a trend, people fell to speculating on why certain neighborhoods were spared. They decided that those neighborhoods housed German spies. The idea festered, and people lived — unjustly — under the suspicion of treason during wartime.

Today we're not focused on spies so much as on health. The Clustering Effect can keep us from seeing that clusters of cancer cases can crop up through random chance, and nothing else. We believe that people develop cancer due to proximity to power lines or big cities, and avoid places unnecessarily. This isn't to say that clusters of diseases aren't often caused by outside effects. But if you find yourself saying, "It can't be a coincidence," take a second to realize that, yes, it can.


Top Image: U.S. Air Force.

[Via Forget Mori, Evidence-Based Technical Analysis]