Wednesday, October 09, 2002

The New York Times reports today that two Americans were awarded the Nobel Prize in economics. (username/password: sabertooth1) There appears to be a large error in the article:

In an article published in Science in 1981, they reported results of a study in which 152 students were given hypothetical choices for trying to save 600 people from a disease. Using one strategy, exactly 200 people could be saved. Using another, there would be a one-third chance that everyone would die, and a two-thirds chance that no one would be saved. Seventy-two percent of the subjects, preferring the less risky strategy, chose the first option.

Does that make sense? Isn't that another way of saying there's a 100% chance everyone would die? I think the New York Times meant to write "there would be a one-third chance that everyone would live, and a two-thirds chance that no one would be saved." The reason for this is because the expected value of both situations would be the same: 200 people saved. (Choice 1: 200 x 100% = 200; Choice 2: 600 x 33.33% = 200.)

The way the Times writer wrote the article, of course Choice 1 is less risky, and everyone would pick it. There's no need to award a Nobel Prize over something that stupid. We should be asking why 28% of the subjects chose the option where everyone dies. The point of the economic study is to find out how people evaluate risk when the risk is the same. Apparently the evaluation of risk isn't linear. There's value in the sure thing.

The great Armchair Economist, Steven Landsburg, discusses the rational evaluation of risk in his regular Slate column. And to show how useful the evaluation of risk is in everyday life, Landsburg applies this economic theory to the evaluation of risk when committing crimes in another Slate column.

In a quick search of my permalinked economics blogs, no one else mentioned this story.

UPDATE: The New York Times just fixed the error on the website. Does anyone know how to find and link to a cached copy of the article with the error?