Take your best guess for the questions below. Without looking up the answers, jot down your guess in your notes app or on a piece of paper.
- What is the weight of the Liberty Bell?
- Saudi Arabia consumes what percentage of the oil it produces?
- What percent of the world’s population lives in China, India, and the European Union combined?
Next, we want you to take a second guess at these questions. But here’s the catch, this time try answering from the perspective a friend whom you often disagree with. (For us, it’s the colleague with whom we shared an office in grad school, ever the contrarian.) How would your friend answer these questions? Write down the second guesses.
Now, the correct answers. The Liberty Bell weighs 2,080 pounds, and, when we conducted the study in 2021, Saudi Arabia consumed 32.5 percent of the oil it produced, and 43.2 percent of the world’s population lived in China, India, and the European Union combined.
For the final step, compare your first guess with the average of both your guesses.
If you’re like most of the participants in our experiment, averaging the two guesses for each question brings you closer to the answer. Why this is has to do with the fascinating way in which people make estimates and how principles of aggregation can be used to improve numerical estimates.
A lot of research has shown that the aggregate of individual judgements can be quite accurate, in what has been termed the “wisdom of crowds.” What makes a crowd so wise? Its wisdom relies on a relatively simple principle: when people’s guesses are sufficiently diverse and independent, averaging judgments increases accuracy by canceling out errors across individuals.
The same principles underlying wise crowds also apply when multiple estimates from a single person are averaged—a phenomenon known as the “wisdom of the inner crowd.”
Interestingly, research suggests that the same principles underlying wise crowds also apply when multiple estimates from a single person are averaged—a phenomenon known as the “wisdom of the inner crowd.” As it turns out, the average guess of the same person is often more accurate than each individual guess on its own.
Although effective, multiple guesses from a single person do suffer from a major drawback. They are typically quite similar to one another, as people tend to anchor on their first guess when generating a second guess.
Previously, at least two methods have been tried to improve on the inner crowd. The first can be thought of as a “sleep on it” effect. Basically, the more time introduced between guesses, the more diverse and independent both guesses become, and as a result, the better the average. Another way is to prompt people to purposefully base their second estimates on different assumptions. Named “dialectical bootstrapping” it has been shown to improve the average too.
This research prompted us to think whether another prompt could help people tap into the wisdom of their inner crowd—through disagreement. While disagreeing with others is often portrayed as undesirable, we hypothesized it would help people de-anchor from their own, initial estimates—making the estimates of a single person more diverse and independent.
When tapping into the wisdom of the inner crowd … taking the perspective of a friend you disagree with could help your own estimates look more like the aggregate estimates of a diverse, independent group of people.
In a set of five studies, recently published in Psychological Science, we tested this approach with over six thousand online participants from the United States and the United Kingdom. We found that disagreement and perspective taking may offer an effective way to improve individual estimates.
Each study used a similar setup. First, we asked people to provide estimates for a series of questions (like the ones you’ve seen above). The crucial intervention came when we asked people to make a second estimate. For some, the instruction was to simply make a second guess, while others were told to make their second guess from the perspective of a friend they disagree with. We then averaged the guesses to test how accurate the two guesses were on their own and together.
We found that, overall, the benefit of averaging two estimates was consistently higher when people took a disagreeing perspective.
But we couldn’t be sure that it was disagreement that mattered and not simply taking a friend’s perspective. So, in a follow up study, we added a third condition where people were asked to answer from the perspective a friend who they often agree with. Taking an agreeing perspective did not improve estimates, performing similarly to simply making a second guess.
Why does taking a disagreeing perspective improves estimates? To find out we prompted participants to provide us with the most extreme estimate they, their agreeing friend, or their disagreeing friend would consider when making the second guess. We reasoned that that taking the disagreeing perspective should mitigate the tendency to anchor a first guess.
This is what we found—those in the disagree condition made more extreme estimates than either of the other two groups. These results suggest being prompted to incorporate a disagreeing perspective can lead to more diverse and independent estimates, seeding the conditions necessary to make the wisdom of the crowds effective within an individual.
But incorporating a disagreeing perspective doesn’t guarantee a better estimate.
In our final study, we identified an important situation where taking a disagreeing perspective may backfire—when a question’s answer lies close to the lower or upper end of a scale. In this case, there is a danger the average will become worse. For example, if the answer is 2 percent on a scale from 0 to 100 percent.
When we asked participants to estimate what percentage of U.S. adults own a cell phone (answer: 95 percent) using the three perspectives—self, agree, disagree—those who adopted the disagreeing perspective produced worse average estimates. Participants in the disagreeing condition were much more likely to make extreme estimates as their second guess, making the average of the two guesses less accurate.
We also identified an important boundary condition—for cases where the answer is close to some natural end point, averaging your guess with a disagreeing perspective likely makes you worse.
So across our studies, when tapping into the wisdom of the inner crowd, we found a new technique—taking the perspective of a friend you disagree with—could help your own estimates look more like the aggregate estimates of a diverse, independent group of people. But we also identified an important boundary condition—for cases where the answer is close to some natural end point, averaging your guess with a disagreeing perspective likely makes you worse.
When thinking about how these results might generalize, it’s also worth mentioning a few nuances. We did limit ourselves to estimates that are numeric, meaning that the generalizability of these findings to other judgments remains to be tested. There is also something to be said about whether people would, even if prompted to make multiple guesses, intuitively average them versus simply taking their first guess, as some research shows that people are rather averse to aggregate judgments.
The wisdom of crowds phenomenon dates back to 1907, when the (in)famous polymath Sir Francis Galton who, while attending a livestock fair where visitors guessed the weight of a butchered ox, observed that the aggregate of visitors’ guesses produced an incredibly accurate estimate of the ox’s weight, even outperforming the guesses of expert butchers and farmers.
Our research shows that fairgoers might have produced more accurate judgments on their own had they tapped into the wisdom of their inner crowd. So whether you’re guessing the weight of livestock or something else, the next time you need to make an estimate, try making one yourself and one as that person you disagree with. Your combined efforts could result in an improved judgment.
