Sunday, March 15, 2009

Why We Make Mistakes


There are many books on 'behavioral economics', but Why We Make Mistakes by a journalist is a pretty good summary. Indeed, I was looking at Ariely's Predictably Irrational, and found this book much more succinct. Fun facts from the book:
  • There is a 1 in 1 million chance of finding a gun in an airport check
  • people don't remember names, as opposed to the jobs or families of a person
  • 80% of calls to a corporate help desk are for lost passwords
  • Simply changing pill colors from white to red and black makes them more distinctive
  • People recall their specific grades in school with an upward bias
  • Men report a median of 7 sex partners, women 4
  • 84% of doctors thought others were biased by self interest, whereas only 16% thought they were biased by self interest
  • The most common airplane accident is 'controlled flight into terrain', or flying a plane into the ground
  • When asked to pick a movie viewed later, more choose highbrow movies; choosing movies now we choose lowbrow movies
  • being first on the ballot adds about 3% to a candidates vote
  • As something becomes familiar, the more we tend to notice it less
  • We see things not as they are, but as they ought to be.
  • Experts make mistakes expecting patterns that aren't there.
  • Depressed people are realists, happy people are overconfident
  • People learn more from summaries than from reading entire chapters
  • A horse race handicapper does as well with 5 bits of information as having 10, 20, or 40, though his confidence increases with the number of information bits available
  • Hope impedes adaptation. Someone with a potentially reversible colonostemy is more unhappy after 1 year than someone with a 1 year irreversible colonostemy.
  • When you are trying to make judgments about complex systems, things that are easily observed are overweighted.
  • Money does not improve efficiency of large organizations (ie, giving everyone more money).
  • Money does increase the ability of individuals to withstand discomfort in tests.

4 comments:

Anonymous said...

Personally I am skeptical behavioral finance's results. Generally a lot of the experiments seem to discount the ability of the subjects to learn.

Take for instance the overconfidence claim relating to driving ability. If most people haven't been in a serious accident, why shouldn't they have high expectations of their ability? I think its a straight forward exercise in Bayesian learning. This is a good paper disputing the "Wobegon paradox."

Anonymous said...

Men report a median of 7 sex partners, women 4

This is easily explained: women prefer promiscuous hunks to monogamous average guys. Thus some men have many partners, and plenty of men have no partners. Only the most disfigured of women is going to end up with no sex partner.

Dangerhorse said...

Randian has it backwards. The data could be explained by a small number of promiscuous women and a larger number of chaste women, in contrast with a population of men whose distribution of partners has lower variance.

Anyhow, the point remains: this one is not necessarily a mistake on anyone's part.

David Foster said...

Most of the controlled flight into terrain accidents are in instrument conditions--the ground being flown into cannot be seen--so not as totally dumb as it may sound.