Irrational Economist_ Making Decisions in a Dangerous World - Erwann Michel-Kerjan [27]
For example, if a senior in high school has to decide which college to attend after having been accepted at five schools, the first step in decision analysis would be to systematically list the pros and cons of each school. Then the student would have to think deeply about key tradeoffs, such as tuition cost versus the quality of the school, proximity to home versus exposure to new experiences, or, say, academic challenge versus time for social activities. Next, the student would link the pros and cons of each school to these personal values in order to produce a ranking. If the schools at the top of this ranking are close, the decision analyst might recommend getting more information or perhaps ask the student to rethink the tradeoffs or other subjective inputs. And if none of the schools turn out to be very attractive, the analyst might suggest that other alternatives be explored.
Decision analysts prefer this kind of disciplined approach because it is very systematic. They are generally distrustful of relying solely on intuition or casual analyses of problems that entail conflicting goals, complexity or uncertainty, and empirical data. Decision analysts believe that problem decomposition, thoughtful tradeoff analysis, probability assessments, and sensitivity analysis in general will help people to both gain insights and, ultimately, make better decisions. Indeed, this is how they define a good decision, at the risk of becoming tautological (i.e., that an optimal choice is one that follows the dictates of decision analysis). They may also cite descriptive research from the psychology of judgment and decision making to support their belief that people, when left to their own devices, tend to act inconsistently, make elementary mistakes of logic, or fall victim to a long list of potential decision traps. And indeed, the field of decision psychology—also known as behavioral decision theory—has catalogued a depressingly long list of traps, from myopic framing and overconfidence to anchoring and wishful thinking.2
At first glance, the case for decision analysis seems very strong and in many cases it can really help reveal the best course of action. But there are some concerns as well. Intuition can in some cases outperform analytical approaches. Although humans may seem like bumblers in psychological experiments, they can be very smart in the real world (see Gigerenzer, 2002). Just as rats may look stupid in maze learning experiments, they may outsmart their human masters in the kitchen when it comes to finding food. Another important consideration is that decision analysis is not easy to master, may not fit a person’s preferred thinking style, and may be hard to apply in practical domains, which explains why its impact has been modest on the whole.
WHAT IS EXPECTED UTILITY?
To appreciate these issues fully, we first need to introduce some of the basic ideas of both decision analysis and behavioral decision theory. Suppose the student in the above example is left with two top schools to choose between. To simplify, assume further that all relevant considerations have been mapped into a single overall attractiveness measure, called utility. Since there is uncertainty about how much utility either school will actually yield after four years, a decision analyst might express this risk in terms of three levels of utility (High, Medium, and Low) with subjective probabilities attached to each. School A entails the following profile (45 percent, H; 30 percent, M; 25 percent, L), meaning that there is 45 percent chance of scoring High eventually, a 30 percent chance of Medium, and a 25 percent chance of Low. If we now assign the following utility values on a scale from zero to ten, H = 10, M = 5, and L = 0, we can compute the expected utility score for school A as: .45*10 + .3*5 + .25*0 = 6. Next, assume that School B has a different profile, namely (60