Our Co Founder Spencer Greenberg Writes Op-Ed for The New York Times:
We overestimate and underestimate our abilities in weird ways.
By Spencer Greenberg and Seth Stephens-Davidowitz
Do you think you are an above-average driver, as most people do? How do you compare with others as a parent? Are you better than most at dancing? Where do you rank in your capability to save humanity?
Many of you will answer these questions incorrectly. For some of these skills, you will think you are better than you actually are. For others, you will think you are worse.
We have long known that, for particular skills, people tend to rate themselves imperfectly. In a famous study from 1981, researchers asked people to rate their driving ability. More than 90 percent considered themselves above average.
Of course, some people who think they are above-average drivers really are. But the 90 percent statistic shows that many people inflate how they compare with others. By definition, only 50 percent of people can rate above the median.
Studies like these led social scientists to conclude that people systematically exaggerate their own capabilities, that they have what researchers call “illusory superiority.”
But that’s not the whole story.
More recent studies have found examples in which people tend to underestimate their capabilities. One found that most people thought they would be worse than average at recovering from the death of a loved one. Another study reported that people thought they were worse than most at riding a unicycle. Here, they exhibit illusory inferiority.
So when are people likely to be overconfident in how they rank? And when are they underconfident?
One of us has conducted research on this topic. Spencer and his collaborators used Positly, a new platform he started that allows researchers to conduct a large number of studies rapidly. Instead of asking people where they rank on just a few skills, they asked where they ranked on 100 skills.
For each skill, participants were asked how they thought they compared with others on the platform who shared their age and gender, and lived in their area. If, on average, people thought they could outperform more than 50 percent of others at the task, that suggests systematic overconfidence. If, however, people thought they would outperform less than 50 percent, that’s evidence of
There was great variation in how people assessed their relative skills at a task. On average, people rated themselves better than 75 percent of others in their ability to use a computer, for instance. But people rated themselves better than only 32 percent of others in their ability to knit a sweater.
People tended to overestimate how they compared with others in their ability to dodge a fraud, win a trivia contest or cuddle. But they tended to underestimate how they ranked in their ability to predict the outcome of a sporting event, win a fistfight or dance.
The researchers next investigated what traits the skills that people were overconfident in have in common. To do this, they asked a different group of people to rate each of the 100 skills on 21 dimensions. They asked, for example, whether the skill involves a lot of luck, whether being bad at the skill would be considered embarrassing or whether the skill is usually performed by experts.
Four factors consistently predicted overconfidence. (If you want to try this yourself, go here.)
First, people tend to be overconfident on skills that reflect one’s underlying personality or character. This helps explain why people overestimated how they compare with others in their ethics, their reliability as a friend and their value as a human being.
And since many people feel pressure to conform to gender norms, this may help us understand why men and women tend to be particularly overconfident on different tasks. Across the 100 skills tested, men are a bit more overconfident overall in how they compared themselves with members of their gender. But men’s overconfidence is particularly noticeable in stereotypically male tasks. Men think they can best the majority of other men in poker, fixing a chair and understanding science. Women are far less confident that they can outperform other women in these tasks.
In contrast, women think they are better than most other women in understanding other people’s feelings, cooking a delicious meal and child-rearing. Men are less confident that they outrank other men in these tasks.
Fun fact: The average man thought he would be better than 63 percent of other men if he had to survive a zombie apocalypse. The average woman thought she would be better than 47 percent of other women at this task.
Another factor that predicted overconfidence is how much a person’s skill level at a trait is a matter of opinion. Give people more wiggle room in how they can define the skill, and they will tend to rate themselves higher. People were slightly more overconfident in how they ranked their intelligence (somewhat subjective) than their performance on an IQ test (seemingly more objective.)
Next, the researchers found that people tend to be overconfident on tasks that are perceived as easy and underconfident on tasks that are perceived as hard. People overestimate how they compare with others in chopping vegetables (easy) but underestimate where they rank in their ability to recite the alphabet backward (hard).
You can see the effect of difficulty on overconfidence clearly in a series of questions the researchers asked about lifting weights. On average, people thought they could outperform 71 percent of others in lifting 10 pounds; 64 percent in lifting 30 pounds; and 55 percent in lifting 50 pounds.
The final factor that influenced confidence is experience. The more experienced people are at a task, the more people tend to be overconfident. People tend to be overconfident in their skill at making scrambled eggs, which most people have done multiple times, and underconfident in their ability to paint a portrait, which most people have rarely tried.
Doing something often will, of course, tend to increase your skill. But it seems that as people gain experience, their confidence goes up faster than their skill.
One of the implications of this research is that people may systematically underestimate their ability to do really hard things that they have never tried before — a notable exception being men’s rating of their ability to handle a zombie apocalypse.
People are indeed overconfident in their ability to drive. (In our sample, people thought they would outperform 66 percent of others in driving.) But people think they are better than 52 percent of others at driving on ice, something that is more difficult and that they do less frequently. And they think they would be better than only 42 percent of others in driving a racing car, something that is really difficult and that most people never try.
Other difficult, relatively rarely tried activities in which people think they would be worse than most include running marathons, making a billion dollars and saving humanity.
What does this research tell us about human nature?
Some of the early work on confidence presented a picture of human beings as comically cocky. Most people, we were told, walked around falsely convinced that they were better than other people. This new research gives us a more nuanced picture.
Sure, many people still traipse around deluded that they outshine others in their driving on non-icy roads, vegetable-chopping and cuddling. But when they imagine doing something difficult or something that they haven’t tried before, people tend to be timid and doubtful of their capabilities. When they go outside their comfort zone, people systematically sell themselves short.
Why Choose Machine Learning Over A Traditional Financial Advisor?
Interview with Astronaut Scott Kelly
Supersonic Travel: The Future of Aviation
On Black Holes: Gateway to Another Dimension, or Ghosts of Stars’ Pasts?
Lockheed Martin Confirms the SR-72 – Son of Blackbird Will Reach Anywhere in the World in One Hour
Shedding Light on Dark Matter: Using Machine Learning to Unravel Physics’ Hardest Questions
Interview with the Inventor of Amazon's Alexa
Aquaponics: How Advanced Technology Grows Vegetables In The Desert
The World Cup Does Not Have a Lasting Positive Impact on Hosting Countries
The Implications of Machine Learning on Condensed Matter Physics & Quantum Computing
Original Article in New York Times: