Conventional wisdom holds that people prefer advisors who offer certainty. For example, in his best-selling book Thinking, Fast and Slow, Daniel Kahneman writes, “An unbiased appreciation of uncertainty is a cornerstone of rationality—but it is not what people and organizations want.”
But research by Chicago Booth’s Celia Gaertig and University of Pennsylvania’s Joseph P. Simmons suggests otherwise, finding that while people judge confident advisors favorably, they don’t mind when advice itself is uncertain. “Our results challenge the belief that advisors need to provide false certainty for their advice to be heeded,” the researchers write.
Gaertig and Simmons conducted a series of studies both on Amazon Mechanical Turk and in the laboratory. In some of the studies, participants predicted the outcomes of upcoming sports games on the day they were played. Participants received advice on the basis of data from betting markets to help them make their predictions.
Uncertain advice wins consistently
After receiving advice on how to predict baseball-game outcomes from two advisors, study participants were more likely to say they preferred the second advisor when that advisor provided uncertain advice—for example, in the form of a percentage probability of an outcome.
Participants who preferred the second of two advisors when that advisor had given:
Gaertig and Simmons, 2018
While the advice was always good, the researchers manipulated the confidence of the advisors who delivered it by having some of them preface it with “I am not sure”—for example, “I am not sure, but I think that the Chicago Cubs will win this game.” Other advisors simply said, “The Chicago Cubs will win this game.”
The researchers also manipulated the certainty of the advice itself. To do this, they used statements of probability, such as: “There is a 57 percent chance that the Chicago Cubs will win this game”; nonnumerical statements of uncertainty, as in: “The Chicago Cubs are more likely to win this game”; and a range of outcomes, such as: “The Bucks and the Cavaliers will score between 197 and 217 points.”
The participants then rated the quality of the advice, indicating how persuasive it was, and how knowledgeable, competent, credible, and trustworthy they perceived the advisor to be.
In line with prior work, Gaertig and Simmons find that participants evaluated confident advisors more positively than advisors who lacked confidence. Importantly, however, they did not find that participants evaluated uncertain advice more negatively than certain advice. In fact, participants either rated uncertain and certain advice the same or they rated uncertain advice more positively than certain advice.
Statements of probability, nonnumerical statements of uncertainty, and imprecise advice in the form of a range of outcomes all yielded similar results. The one exception was that participants didn’t like when confident-appearing advisors qualified their advice with a “probably.”
In further experiments, these results held when participants were asked to directly choose between advisors. Most participants were inclined to choose advisors who included uncertainty in their advice over those who provided certain advice.
“Advisors benefit from expressing themselves with confidence, but not from communicating false certainty,” the researchers conclude.