Messy: How to Be Creative and Resilient in a Tidy-Minded World

by Tim Harford

So Lee Green and his colleagues developed a simple decision tree, throwing away most of the detail in the diagnostic table and focusing on a few obvious clues. The decision tree asks three yes/no questions: first, does the patient display a particular anomaly on a heart monitor? If so, straight to the coronary care unit. Otherwise, question two: is the patient’s main complaint about chest pain? If not, then there is no need for coronary care. But if so, then a third question tells the doctor to look for one of five obvious clues – any one of them is enough to send the patient to coronary care. The decision tree can be written on a postcard. Unlike the sophisticated algorithm, it requires no calculator. But it works. In fact it seems to work better than either the doctors or the complex diagnostic tool, referring almost every single patient with a heart attack – even more than the defensively minded doctors – yet generating fewer false positives than the diagnostic. More important, it saves time and effort in an urgent situation – which means that unlike the complex diagnostic, it gets used.

Use of heuristic decision tree behavioural economics