Professor Frederick discovered striking systematic patterns in how people answer questions about risk and patience, including those above. This short problem-solving test, he found, predicts a lot:
1) A bat and a ball cost $1.10 in total. The bat costs $1 more than the ball. How much does the ball cost?
2) If it takes five machines five minutes to make five widgets, how long would it take 100 machines to make 100 widgets?
3) In a lake, there is a patch of lily pads. Every day, the patch doubles in size. If it takes 48 days for the patch to cover the entire lake, how long would it take for the patch to cover half the lake?
The test measures not just the ability to solve math problems but the willingness to reflect on and check your answers. (Scores have a 0.44 correlation with math SAT scores, where 1.00 would be exact.) The questions all have intuitive answers—wrong ones.
Professor Frederick gave his ″cognitive reflection test″ to nearly 3,500 respondents, mostly students at universities including M.I.T., the University of Michigan and Bowling Green University. Participants also answered a survey about how they would choose between various financial payoffs, as well as time-oriented questions like how much they would pay to get a book delivered overnight.
Getting the math problems right predicts nothing about most tastes, including whether someone prefers apples or oranges, Coke or Pepsi, rap music or ballet. But high scorers—those who get all the questions right—do prefer taking risks.
″Even when it actually hurts you on average to take the gamble, the smart people, the high-scoring people, actually like it more,″ Professor Frederick said in an interview. Almost a third of high scorers preferred a 1 percent chance of $5,000 to a sure $60.
They are also more patient, particularly when the difference, and the implied interest rate, is large. Choosing $3,400 this month over $3,800 next month implies an annual discount rate of 280 percent. Yet only 35 percent of low scorers—those who missed every question—said they would wait, while 60 percent of high scorers preferred the later, bigger payoff.
The problem with the temperament checks in the last two paragraphs is that they’re still testing roughly the same thing that’s tested earlier on-- competence at word problems.
And possibly interest in word problems—I know I’ve seen versions of the three problems before. I wouldn’t be going at them completely cold, but I wouldn’t have noticed and remembered having seen them decades ago if word problems weren’t part of my mental univers.
I recall reading a study once that used a test which I am almost certain was this one to try to answer the cause/correlation question of whether philosophical training/credentials improved one’s critical thinking or whether those who undertook philosophy already had good critical thinking skills; when I recently tried to re-find it for some point or other, I was unable to. If anyone also remembers this study, I’d appreciate any pointers.
(About all I can remember about it was that it concluded, after using Bayesian networks, that training probably caused the improvements and didn’t just correlate.)
I don’t believe the article says “reflective”:
The problem with the temperament checks in the last two paragraphs is that they’re still testing roughly the same thing that’s tested earlier on-- competence at word problems.
And possibly interest in word problems—I know I’ve seen versions of the three problems before. I wouldn’t be going at them completely cold, but I wouldn’t have noticed and remembered having seen them decades ago if word problems weren’t part of my mental univers.
Somewhat offtopic:
I recall reading a study once that used a test which I am almost certain was this one to try to answer the cause/correlation question of whether philosophical training/credentials improved one’s critical thinking or whether those who undertook philosophy already had good critical thinking skills; when I recently tried to re-find it for some point or other, I was unable to. If anyone also remembers this study, I’d appreciate any pointers.
(About all I can remember about it was that it concluded, after using Bayesian networks, that training probably caused the improvements and didn’t just correlate.)