Overconfidence Is Overrated

Don Moore with book

Source: Sarah Moore

Men are so simple and so much inclined to obey immediate needs that a deceiver will never lack victims for his deceptions.
—Machiavelli

The idea that overconfidence breeds success is widely accepted. It is, however, almost certainly false. There are positive correlations between confidence and success, but these correlations mostly stem from the fact that ability causes both confidence and success. Hence, the correlation between confidence and success is spurious. As every undergraduate learns, correlation is not sufficient for causation. The inference that if one only increased one’s confidence success would follow finds little support in the psychological literature. Researchers have tried, and some have claimed ‘success,’ but careful inspection suggests that these researchers were overconfident. Try as you might, you cannot wrest causality from correlational data.

A study by Vitanova (2020) is a good example. She used sophisticated econometric statistics to conjure causality. Supposedly, overconfidence among CEOs predicts (causes?) firm performance when the endogeneity of the variables is checked. Were it only so! Quasi experiments are more quasi than experiment. Experimental interventions cannot be faked (Pearl, 2000). The title of Vitanova’s article signal’s the author’s own overconfidence in her story: "Nurturing Overconfidence." As overconfidence is defined as a discrepancy between confidence and reality, any reality that correlates with a discrepancy will also correlate with the variable from which something is subtracted. If overconfidence predicts performance, so does simple confidence, and perhaps that is all there is (Krueger, Heck, & Asendorpf, 2017).

Don Moore (2020), in his new book Perfectly Confident, demythologizes overconfidence (see Krueger & Heck, 2021, for a review). Mind you, Moore does not knock confidence, but the false idea of being able to prevail if one were only more confident. Moore begins by recalling William James’s (1874/1978) famous thought experiment designed to show that in a state of heightened confidence one could do what one could otherwise not do. James felt that he could jump across an Alpine crevasse if he psyched himself into believing that he could. Mens agitat molem. James, writing from Harvard, did not actually try this. For good reason. Being a pragmatist, James brushed aside the logical paradox of overconfidence; if he jumped and lived to tell the tale, he would, by definition, not have been overconfident. Moore and colleagues actually did an intervention experiment where they instilled optimism (a proxy for exogenous confidence) in some participants, and found that these participants performed no better on a test than participants who were left to their own devices (Tenney et al., 2015).

It is clear that the myth dies hard. I venture to predict that it will not die at all, no matter how much counter-evidence the Moores among us will amass. Science will know, but the people will not follow (and certainly not the CEOs). I am at peace with this conclusion. Science searches truth; it has no obligation to convince everyone of it. It only has an obligation to share its findings. Have we not shown ad nauseam that correlation does not amount to causation? Yet, we see that even researchers ignore this simple principle. Why should people in the street do better?

It becomes a meta-question for psychological science to figure out why unlearning certain myths is so hard. I see three reasons. First, there is wishful thinking. It is pleasant to imagine that one could get something for nothing if one ‘only put one’s mind to it.’ CEOs might be especially vulnerable to this idea as it would feed their egos and justify their handsome compensation packets. Even more troubling is the possibility that the overconfidence myth can be moralized in that CEOs can tell themselves that their signaling of high confidence, however unjustified by reality, will motivate the workforce. These CEOs would be humbly overconfident for the sake of others. Thinking they are generously deceiving others, they are deceiving themselves. To cite a thoughtful friend of mine, who need not be named, “followers typically need to be assured that a strategy or set of actions—if endorsed by the leader—will lead to positive outcomes. Followers need to believe in order to become converts to the cause. They need to buy the myth. So the leader is often accorded the deus ex machina role.”

Second, there is the fallacy of the illusory correlation. If people only look for affirmative evidence, that is, cases of overconfidence paired with success, they may mistakenly infer that heightening one’s confidence above realistic levels will do good (Rothbart, 1981).

Third, Moore’s (2020) analysis actually reveals how a strengthening of confidence can be beneficial. If confidence is initially too low, keeping individuals from trying where they could succeed, a confidence boost can have a beneficial causal effect. But then it is perfectly calibrated confidence that does the trick, and not overconfidence. Two examples might convince you. Moore tells the story of how he, following the irrepressible Tony Robbins in a moment of darkness, walked, as told by Robbins, across a bed of burning coals. He got to the other side without dying. The secret is that there is no secret. Anyone can do it. The logic is similar to the one underlying the old ‘move your finger through the candle’s flame’ trick, which always regales small children. For the other example, recall David’s victory over Goliath (1 Samuel 17). David knew the sling to be a terrible and effective weapon. In fact, this was common knowledge at the time, but Goliath chose to ignore it. David, it seems, was appropriately confident, whereas Goliath was overconfident. We know how it ended.

Moore concludes that confidence is a great asset if properly calibrated. Let wisdom overcome braggadocio.