Friday, June 23, 2017

Thinking, Fast and Slow II

Part II of the book, which runs from where I left off to the midpoint, is titled Heuristics and Biases. These chapters look at various errors made in System 2, and Kahneman's research usually consists of written tests given to volunteer college students. The main result is that they do not think well statistically, and they tend to produce, for example, causal explanations for events even when there is insufficient information to support their conclusions. Rather than apply valid statistical methods, they manufacture story lines which add a coherence that may not truly be present. Slight differences in the way in which a question is presented have a major impact on how it is answered even if those differences have no relevance to the answer. There is glee in Kahneman's tone when graduate students with training in statistics make the exact same kinds of mistakes that less-educated people do.

Having previously read a little on this topic, I am somewhat familiar with the errors he cites. The most interesting one to me is regression to the mean, which, he says, is beyond the comprehension of most people. I think about this in relation to investments when deciding what the future value of an investment might be given the history of its valuation. The crux here is randomness, and Kahneman correctly notes that people have a hard time with that idea. Rather than accept events as random, they produce stories which provide the illusion of coherence, hence their predictions tend to be inaccurate. It is an important point that luck and chance are undervalued compared to skill and talent in ordinary discourse. This becomes the subject of Part III, Overconfidence, which I've just begun.

Part II was a little too professorial for my liking. The questionnaires administered by researchers usually had an artificial quality which I thought detracted from their usefulness. They intentionally invoked various innuendos to prompt incorrect answers. As far as I could tell, Kahneman was mainly interested in proving the existence of specific kinds of mistakes, and he was indifferent to how important or unimportant getting the right answer may have been to the test participants. I felt that, because there were no negative consequences for study participants who didn't think clearly, they had little incentive to make much of an effort. As a practical matter it may be difficult to design studies in which the participants are engaged, but I think the results would be more meaningful if there had really been something at stake, and there wasn't. It is one thing to speculate on what the nonexistent "Linda" may be doing with her life based on a very sketchy description of her, and something else entirely to choose a spouse or a career, even though the same kinds mistakes might occur in all of these decisions. "Linda" is an imaginary being, which hardly places her on equal footing with a spouse, someone who might produce your children and occupy a significant fraction of your life.

So far in the book, Kahneman has mainly been explaining the research and concepts in his field and hasn't seemed interested in its implications. He is friends with Cass Sunstein, the law professor and behavioral economist who advocates rational policy-making, but he refrains from siding with him on the injection of rationality into public life. He will probably expand on his ideas later in the book, but for the moment he seems content to let poor judgment wreak havoc on the world in the name of democratic principles. If he retains this attitude, I will have to part company with him at some point.

No comments:

Post a Comment