Why Your Predictions Are Bad: Reason #37
December 15, 2011 Leave a comment
Society’s highly esteemed economic forecasters have not done a good job of late. In fact, they tend to almost always be overly optimistic about economic growth. The usual cognitive bias suspects (with Kevin Spacey as the confirmation bias) surely play some role, but a series of experiments by a USC researcher highlights an issue that hasn’t received much attention: Probability estimates are influenced by distance, and that causes people to think unlikely events are more likely to happen in distant contexts and likely events are more likely to happen in near contexts.
In one experiment subjects were given the opportunity to bet on a boxing match featuring a 20-1 underdog fighting against an overwhelming favorite. Subjects who were told the fight would be 3,000 miles away rather than 3 miles away bet a significantly greater amount of money on the underdog. The imagined distance made an unlikely underdog victory seem more likely.
A follow up experiment asked subjects whether a certain person was likely to utilize renter’s insurance. Subjects were told that theft was likely or unlikely, and that the person would be moving into a new apartment in either a year or a day. When told theft was unlikely and the person was moving in a year, subjects were more likely to say the insurance would be used. They were also more likely to say the insurance was needed when told theft was likely and the person was moving in a day. Envisioning an unlikely event far in the future, or a likely event in the present, made the events seem more likely.
The influence of distance presents a challenge for policy makers who have a lot riding on their predictions. When considering an unlikely event — such as unemployment rising to 10% — the event seems less likely to happen in the country they live in or at the present moment than in a distant country or at a point far in the future.
Unfortunately, this is one of those prediction problems without a great solution. The only things I can think of are forcing economists to make predictions from Siberia or electing an Asimov enthusiast who will delegate all important economic decisions to Mulitvac.
Wakslak, C. (2011). The where and when of likely and unlikely events Organizational Behavior and Human Decision Processes DOI: 10.1016/j.obhdp.2011.10.004