- Most individuals are not very good at judging their own skills, leading to either overconfidence or underconfidence.
- In difficult tasks, people often overestimate their performance but expect to do worse than average. The reverse is true for easy tasks.
- This “hard-easy” effect suggests that people can be overconfident and underconfident at the same time.
- The effect may be explained by the existence of different types of overconfidence and the reliance on prior expectations.
Let’s start off with a little trivia test. Write down your best guesses for the following questions (no Google allowed!):
- What is the capital of Azerbaijan?
- How many nanoseconds are there in a second?
- Before becoming a Canadian province, Saskatchewan was part of what other entity?
- How many feet are there in a mile?
- What is the most popular first name in the world?
- Which U.S. state instituted the nation's first mandatory seatbelt law in 1984?
- Which North American city has the following subway stops: Kendall Square, Central Square, and Porter Square?
- How many countries were members of the European Union as of June 2003?
- Who was the Greek God of War?
- Who ruled Iraq before Saddam Hussein?1
Without checking the answers (below), how many of the 10 items do you think you got right? How many of the 10 questions do you think the typical reader got right?
Yes, I know—those trivia questions weren’t easy! They were carefully selected by psychology researchers, led by UC Berkeley's Don Moore, to study the puzzling human bias of overconfidence. Overconfidence refers to a feeling of confidence that is unjustified based on someone’s actual performance or ability.
Given your personal predictions of the quiz scores above, is there a chance that you were overconfident? Let's find out.
Being Both Better and Worse Than You Think You Are
Research suggests that overconfidence isn’t always straightforward. The trivia quiz above was part of a research study that teased apart different aspects of overconfidence. The experiment presented 255 U.S. students with a general knowledge quiz. Half of the participants received the comparatively difficult set of questions that you answered just now. The other half received 10 much easier questions, with an example being “Berlin is the capital of what country?” After completing the quiz, participants were asked to judge their own and other people’s expected performances in the quiz.
Here are some key results of the experiment:
- In the difficult trivia quiz, participants received an average test score of 1.39 correct answers.
- When asked to predict their own scores on the difficult quiz, the participants’ average prediction was 2.62 correct answers.
- When asked to predict other people’s scores on the difficult quiz, the participants’ average prediction was 3.54 correct answers.
Let’s try to make sense of these numbers. First, the low quiz results confirmed the overall difficulty of the trivia questions. On average, people got more than 8 out of 10 questions wrong. Hence, there is certainly no need to beat yourself up if you didn’t score a perfect 10 (phew)!
Taking a more detailed look at participants’ score predictions, however, several puzzling findings emerge. Despite the tough task, most people overestimated their own performance. On average, they expected to score 1.23 points higher than they actually did.
At the same time, they thought others would perform even better. Indeed, they estimated that other participants would beat their own expected score by approximately one point.
It thus seemed that while people expected to score better than they actually did, they also expected to do worse than average. Does that mean they were both overconfident and underconfident at the same time?
To make things even more complicated, the study demonstrated a surprising influence of the decision context. The researchers found that the task type affected people’s confidence.
In the difficult trivia quiz, people tended to overestimate their personal scores yet underestimate their performance compared to others. By contrast, the reverse was true when participants completed the easy trivia questions.
When presented with the easier quiz, participant scores reached an impressive average of 8.53, but they underestimated their performance and expected lower scores (average of 8.26) than was actually the case. Yet, they also thought they were smarter than average, predicting other participants’ scores (average of 8.05) to be lower than their own.
Making Sense of the Results
The experimental results on overconfidence are complex and may appear contradictory at first. It seems like people are both underconfident and overconfident at the same time—but which direction their predictions go depends on the difficulty of the task. How is this possible?
Firstly, it is important to recognise that overconfidence is a multi-faceted concept. Psychologists suggest that several different types of overconfidence exist. Two distinct types of overconfidence are overestimation, which refers to the belief that you are better than you actually are, and overplacement, which refers to the belief that you are better compared to other people. Since these two types are independent of each other, it is possible to be overconfident in one aspect and underconfident in another. (You can find more details on the different types of overconfidence in this related article.)
The puzzling influence of context is typically referred to as the “hard-easy effect.” Different theories have been proposed to explain the surprising differences in confidence for hard and easy tasks. One theory by Dan Moore and another psychologist, Paul Healy, is based on the assumption that people lack insight into their own performance, and are even worse at judging others’ performance. This is why feelings of confidence are not only affected by situational performance cues but also by prior experiences and expectations.
Here's an example of what this might mean in the context of our trivia quiz.
- Susan, an avid quizzer who never misses an episode of "Mastermind," is intrigued by the trivia quiz of this article. Before even looking at the questions, she expects to get a high score, because—well—she usually does! While going through the questions, however, she struggles to think of possible answers and takes longer than expected to finish. These negative performance cues are eventually integrated with Susan's initial, optimistic expectations. Despite her struggles during the quiz, she ends up overestimating her score.
- Susan is likely to be even more optimistic when trying to predict other people's performances on the quiz. Individuals typically use their own performance estimates as a point of reference when gauging others. Hence, Susan's judgement is likely to be influenced by both her own expected score as well as her prior expectations of average quiz performance. This means that her predictions for other people's performance on the quiz will fall somewhere in between and will be higher than her own. A logical consequence is worse-than-average beliefs or underplacement.
The example of Susan showcases the importance of prior beliefs when calibrating feelings of confidence with actual performance. In unusually hard tasks, prior expectations about performance are typically higher than actual performance, thus resulting in overestimation and underplacement. By contrast, in unusually easy tasks, prior expectations about performance are typically lower than actual performance, thus resulting in underestimation and overplacement.
How do your own results fit in with this theory? Compare your actual quiz scores with your predictions for personal and average performance, using the data cited in this article. Were you overconfident and underconfident at the same time? And what does this mean for your judgements and choices in real life?
1. Questions adapted from:
Moore, D. A., & Small, D. A. (2007). Error and bias in comparative judgment: on being both better and worse than we think we are. Journal of personality and social psychology, 92(6), 972–989. https://doi.org/10.1037/0022-3518.104.22.1682
Answers: 1) Baku; 2) 1 billion; 3) the Northwest Territories; 4) 5,280; 5) Mohammed; 6) New York; 7) Cambridge, MA; 8) 15; 9) Ares; 10) Ahmad Hassan al-Bakr