06.03.2016 Views

Thinking, Fast and Slow - Daniel Kahneman

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

correct answer is 41%: Applying Bayes’s rule in odds form, the prior odds are the odds for the<br />

Blue cab from the base rate, <strong>and</strong> the likelihood ratio is the ratio of the probability of the<br />

witness saying the cab is Blue if it is Blue, divided by the probability of the witness saying the<br />

cab is Blue if it is Green: posterior odds = (.15/.85) × (.80/.20) = .706. The odds are the ratio of<br />

the probability that the cab is Blue, divided by the probability that the cab is Green. To obtain<br />

the probability that the cab is Blue, we compute: Probability (Blue) = .706/1. 706 = .41. The<br />

probability that the cab is Blue is 41%.<br />

not too far from the Bayesian: Amos Tversky <strong>and</strong> <strong>Daniel</strong> <strong>Kahneman</strong>, “Causal Schemas in<br />

Judgments Under Uncertainty,” in Progress in Social Psychology, ed. Morris Fishbein<br />

(Hillsdale, NJ: Erlbaum, 1980), 49–72.<br />

University of Michigan: Richard E. Nisbett <strong>and</strong> Eugene Borgida, “Attribution <strong>and</strong> the<br />

Psychology of Prediction,” Journal of Personality <strong>and</strong> Social Psychology 32 (1975): 932–43.<br />

relieved of responsibility: John M. Darley <strong>and</strong> Bibb Latane, “Byst<strong>and</strong>er Intervention in<br />

Emergencies: Diffusion of Responsibility,” Journal of Personality <strong>and</strong> Social Psychology 8<br />

(1968): 377–83.<br />

17: Regression to the Mean<br />

help of the most brilliant statisticians: Michael Bulmer, Francis Galton: Pioneer of Heredity<br />

<strong>and</strong> Biometry (Baltimore: Johns Hopkins University Press, 2003).<br />

st<strong>and</strong>ard scores: Researchers transform each original score into a st<strong>and</strong>ard score by subtracting<br />

the mean <strong>and</strong> dividing the result by the st<strong>and</strong>ard deviation. St<strong>and</strong>ard scores have a mean of<br />

zero <strong>and</strong> a st<strong>and</strong>ard deviation of 1, can be compared across variables (especially when the<br />

statistica {he deviatiol distributions of the original scores are similar), <strong>and</strong> have many desirable<br />

mathematical properties, which Galton had to work out to underst<strong>and</strong> the nature of correlation<br />

<strong>and</strong> regression.<br />

correlation between parent <strong>and</strong> child: This will not be true in an environment in which some<br />

children are malnourished. Differences in nutrition will become important, the proportion of<br />

shared factors will diminish, <strong>and</strong> with it the correlation between the height of parents <strong>and</strong> the<br />

height of children (unless the parents of malnourished children were also stunted by hunger in<br />

childhood).<br />

height <strong>and</strong> weight: The correlation was computed for a very large sample of the population of<br />

the United States (the Gallup-Healthways Well-Being Index).<br />

income <strong>and</strong> education: The correlation appears impressive, but I was surprised to learn many<br />

years ago from the sociologist Christopher Jencks that if everyone had the same education, the

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!