Generalizing and Mean Regression

• Chapter 16: People are unlikely to learn anything they can use from surprising statistical facts. They can, however, generalize when presented with surprising individual cases. In short, we are unwilling to deduce the particular from the general, but we can infer the general from the particular.
• Chapter 17: If you don’t understand the concept of regression to the mean, go strait to this chapter. System 1 has the bad habit attributing cause the changes that are only random fluctuations. This is part of the reason System 2 finds it difficult to understand mean regression and correlation, which are both aspects of the same concept. (Dr. Doug: Changes in a student’s NCLB scores from one year to the next will vary just like someone’s golf score. Just because it goes up or down doesn’t mean the teacher was good or bad. Regression scores are a common source of trouble in research. Get ready for more trouble as teachers are judged by regressive test scores.)

Making Better Decisions

• Chapter 18: Kahneman gives great advice on how to correct your intuitive predictions using System 2 with the available evidence when you face a decision. The goal is unbiased decisions, reasonable assessments of probability, and moderate predictions of numerical outcomes. If you follow his advice you will not know the joy of correctly predicting and extreme outcome, but you will most likely be closer to the truth more often.

Have You Been Lucky of Good?

• Chapter 19: Humans consistently fool ourselves by constructing flimsy accounts of the past and believing they are true. We can’t help dealing with the limited information we have as if it’s all there is to know. We tend to ignore our ignorance. Hindsight bias is our tendency to believe we knew something was going to happen before it did. Once something happens, we exaggerate the probability that it would happen. We revise our beliefs in light of what happens. Outcome bias interferes with evaluating decisions properly. We also evaluate success without considering the impact of luck and how successful people and companies tend to regress to the mean. Kahneman uses this to question the conclusions in Collin and Porras’s famous.Built to Last.

Dealing With Long-Term Predictions

• Chapter 20: Even though our ability to predict things in the long term is poor, we continue to feel that our predictions are valid and make excuses when we are wrong. The illusion of validity was Kahneman’s first cognitive illusion. People who pick stocks for a living do no better than chance in the long term due to the efficiency of the market. The same is true for political pundits. Facts that challenge one’s skills are often not absorbed. Experts hate to be wrong. Such illusions are more stubborn than visual illusions. Everything makes sense in hindsight.
• Chapter 21: Many studies over the years show that simple formulas either match or exceed expert predictions in fields as diverse as the evaluation of credit risk, winners of football games, prediction of future grades, and even the future price of Bordeaux wine. When asked to evaluate the same information twice, experts frequently give different answers. Experienced radiologists contradict themselves 20% of the time. An excellent example of the power of formulas was created in 1953 to judge newborns. The so-called Apgar score is credited with reducing infant mortality and is still in use. You can create your own formula for making decisions using the advice at the end of this chapter. You don’t need to ignore your intuitive judgment, just don’t simply trust it.

Share this: