Why, and When, Are We Bad at Predicting Risk?

Would you risk riding to the airport in a self-driving car?

If you said no, you aren’t alone. In a 2017 Pew survey, 56 percent of Americans said they would not risk it. That proportion likely has increased in the aftermath of the self-driving Uber car killing a pedestrian on March 19, 2018. Meanwhile, so far this year, around 1200 other pedestrians have been killed by people-driven cars, and few of us have decided not to risk driving (or walking).

Time will tell whether, as experts assure us, self-driving cars, without distracted or inebriated drivers, really will be much safer. Even if it’s so, it will be a hard fact to embrace. Why? Because we fear disasters that are vividly “available” in our minds and memories—shark attacks, school shootings, plane crashes—often in settings where we feel little control. “Dramatic outcomes make us gasp,” Nathan DeWall and I conclude in Psychology, 12^th Edition, while “probabilities we hardly grasp.”  

We do a better job of grasping probabilities in realms where we have lots of experience. If a weather forecaster predicts a mere 30 percent chance of rain for tomorrow, we won’t be shocked if it does indeed rain—as it should about one-third of the time, given such a forecast. We have much less experience with presidential election predictions. Thus many people thought the pollsters and prognosticators had egg on their faces after Donald Trump’s upset win. Statistician and author Nate Silver’s final election forecast gave Trump but a 29 percent chance of victory. Although a 30 percent chance of rain and a 30 percent victory chance are the same odds, an ensuing rain comes as less of a shock.

With March Madness basketball games, as with weather forecasts, we fans have more experience. Tweets Silver:

Lesson learned? In domains where we have minimal direct experience, we often don’t get it because the cognitive availability of vivid, rare events may hijack our thinking: “Probabilities we hardly grasp.” But in realms where we do experience life’s uncertainties—as in daily weather variations and sports outcomes—we get it. We appreciate that probabilities calibrate uncertainties. Given enough happenings, anything, however improbable, is sure to occur.