Opinion, Berkeley Blogs

Making better forecasts

By Don Moore

Trying to decide what to buy at the store? It depends on how much food your family will consume in the coming week. Where should our government set tax rates? The decision depends, among other things, on the consequences of that choice for economic growth rates and government revenues. How to invest in homeland security? It depends on where future risks lie. These are but three examples of forecasts that inform decisions. Just about every decision depends in some basic way on a forecast underlying it, but we rarely think systematically about how we make forecasts and the errors to which they are prone.

Discussions of policies often get hung up on philosophical arguments that can be clarified by examining the forecasts implicit in them. We can make progress resolving these policy debates, whether they be about how much milk to buy at the store or how high taxes should be, if we examine the forecasts that underlie our predictions and make their assumptions more explicit.

Psychology offers some useful lessons about the sorts of biases to which human judgment falls victim when making forecasts. We tend to zero in on the outcome we think is most likely and become too confident that our prediction will prevail. We neglect outside perspectives, such as the forecasts of other people. It is rare for people to give the opinions of others equal weight with their own, even when others’ opinions represent the collective wisdom of many other people. And we often neglect to consider historical guidance, focusing instead on the most novel and interesting current trends, even when they aren’t useful predictors.

There is a project underway whose goal it is to learn how to advise policy-makers on how to make better forecasts. Participants in the study, and we need thousands of them, will be making forecasts and getting feedback on the quality of their forecasts. They will get to experience a variety of different approaches to making forecasts and training on how to think about forecasting. The study will be employing the best insights from psychology to make forecasts as accurate as they can be. The scientists running this study are motivated by the abiding hope that by improving the quality of our forecasts we can improve the substance of policy debates and clarifying the terms of debate. Better forecasts should also help us make better decisions, about everything from national security to grocery store purchases.

If you are curious to find out more about the project or might consider signing up, please visit:

http://goodjudgmentproject.blogspot.com/