In the 1990s, I was living in Chicago, where there are high dropout rates from the high schools. People often asked, "Is there a way to know which school has the lowest dropout rate?" There existed data measuring different cues of school performance: the pay of teachers, the number of English-speaking students in a class, things like that.No, this is an example of the superiority of computer analysis. Simple regressions often show that one or two factors are dominant, but it takes a statistical analysis to prove it. Those who rely on gut feelings often get fooled into thinking things like teacher pay are more significant.
I wondered: could one feed these into a computer, analyze them and obtain a prediction on which high school produced the fewest dropouts? We did that. And we were astonished to find that computer-based versions of Franklin’s bookkeeping method -- a program that weighed 18 different cues -- proved less accurate than going with the rule of thumb of "get one good reason and ignore the rest of the information."
Q: What was the "one good reason" that got you the right answer?
A: Knowing which school had high daily attendance rates. If two schools had the same attendance levels, you needed one more cue -- good writing scores -- and then you could ignore the rest.
Tuesday, August 28, 2007
New Blink book
Gerd Gigerenzer, a German social psychologist, has written a new book justifying making decisions by gut feeling. Here is a sample of his research: