George and Jill are a nice young couple. He’s 30 and works at the local hospital; she’s 28 and is just finishing her PhD in molecular biology. Are they happily married? Their families think so but then they don’t really see them very often. You could interview them at great length and try to form an impression through a combination of “blinking” and “thinking”. That’s what marriage counselors do – but here’s an alternative method.
Psychologists John Howard and Robyn Dawes trained one partner from each of 27 couples just like George and Jill to monitor their own behavior for 35 consecutive days. The monitors counted two types of behavior and also rated the couples on a seven-point scale of marital happiness. Howard and Dawes used a simple decision rule to predict marital happiness: the difference between the frequencies of the two types of behaviors across the 35 days. The result: the simple decision rule they discovered was a valid albeit imperfect predictor of marital happiness.
The point here is that, in a domain as complex as marital happiness, predictions based on elaborate theories typically miss the mark (remember Chapter 9 and the problem of noise) – although, in hindsight, they can make great stories! On the other hand, simple decision rules can have better – even though limited – predictive validity. As Robyn Dawes and another of his research partners, Bernard Corrigan, put it: “The whole trick is to know what variables to look at and then to know how to add.” You do need judgment for the first task, but you can delegate the second to a calculator.
© Spyros Makridakis, Robin Hogarth and Anil Gaba, 2009