I’ve a lot here in this blog and in my books about how much I use teaching and learning inductively and how effectively I think it is — with both language learners (even Google Translate learns inductively!) and non-ELL’s.

Now, researchers have found that thinking inductively is more effective in predicting world events, too.

Jonah Lehrer has written a column for Wired titled Do Political Experts Know What They’re Talking About? In it, he interviews researcher Philip Tetlock about his examining of political pundit predictions. Here is an excerpt:

Tetlock: Some experts displayed a top-down style of reasoning: politics as a deductive art. They started with a big-idea premise about human nature, society, or economics and applied it to the specifics of the case. They tended to reach more confident conclusions about the future. And the positions they reached were easier to classify ideologically: that is the Keynesian prediction and that is the free-market fundamentalist prediction and that is the worst-case environmentalist prediction and that is the best case technology-driven growth prediction etc. Other experts displayed a bottom-up style of reasoning: politics as a much messier inductive art. They reached less confident conclusions and they are more likely to draw on a seemingly contradictory mix of ideas in reaching those conclusions (sometimes from the left, sometimes from the right).

We called the big-idea experts “hedgehogs” (they know one big thing) and the more eclectic experts “foxes” (they know many, not so big things).

Lehrer: Do these different styles correlate with levels of accuracy?

Tetlock: In assessing accuracy, it is crucial to make the “law of large numbers” work for you. Any fool can be lucky a few times. The key is consistency. So, in the first round of our studies, we assessed the accuracy of almost 30,000 predictions from almost 300 experts. We tested a lot of different hypotheses about the correlates of consistency and accuracy. Is ideology the key factor? Having a PhD? Having past access to classified information? And a lot of hypotheses bit the dust. The most consistent predictor of consistently more accurate forecasts was “style of reasoning”: experts with the more eclectic, self-critical, and modest cognitive styles tended to outperform the big-idea people (foxes tended to outperform hedgehogs).

Also in the article, Tetlock talks about an interesting forecasting project for which he is recruiting volunteers. The link in the article to learn more and volunteer is the wrong one, but I’ve found the correct site. It’s called the Good Judgment Team.