Two New Thoughts on the Wisdom of Crowds

| By TMA World

Crowd of Crowds

Most of us know about the wisdom of the crowd (WOTC) – the aggregation of estimates from a random group of people is far more accurate than the estimates from the individuals that make up the crowd.  Averaging the diversity of views in a crowd has been used to improve economic forecasts, doctors’ decisions, and weather predictions.

Some sociologists have argued that allowing participants in a crowd to talk to one another before giving WOTC answers would reduce the accuracy of the final answer because it would lead to a herd mentality – the crowd converging on a consensus at the expense of accuracy.

New research throws doubt on this conclusion

A team of researchers affiliated with institutions in Argentina, the U.S., and Germany asked 5180 people several general knowledge questions such as estimating the height of the Eiffel Tower.  Each respondent was asked to give an answer privately and then to join with a group of four others to discuss the questions.  After one minute on each question, each group was asked for mutually agreed-upon answers.  

The researchers reported that the average answers of the 280 groups of five (not all who answered the questions individually were willing to join a group) were 49.2% more accurate than the average crowd response (as a whole).

Based on this research, authentic deliberation in a smaller group (a crowd within a crowd), and not simply voting independently, can lead to better decisions.

Drawing on Your Inner Crowd

What if you don’t have access to a crowd? Can your own different perspectives be utilized?

Researchers have been able to utilize data from three separate contests at a Dutch casino.  Each contest offered a prize of 100,000 euros to the person who could come closest to guessing the number of pearls held in an oversized Champagne glass.  About 160,000 people participated in each contest, and they could guess repeatedly over a two-month period.

What the researchers found was that people’s errors tended to get smaller if they averaged over many guesses.  It wasn’t that people were learning over time by, for example, consulting with others.  The later guesses showed no significant improvement in accuracy – it was only the average that became more accurate.  Also:

  • Estimates were better when people took more time between guesses.  It could be that this extra time allowed them to have different thoughts or forget their previous line of thinking (increasing the independence of subsequent guesses). 
  • Accuracy also improves if people are encouraged to make estimates in different ways – using different evidence and strategies.
  • Guessing a dozen times can improve one person’s estimate by about 40%; the same result can be achieved by averaging the guesses of two different people.
  • People’s biases can lead them astray; they account for 50% of errors.  People tend to stand by their own initial views.  In cognitive psychology, this is known as Choice-Supportive bias – when you choose something you tend to feel positive about it even if that choice has flaws.

In short, seeking the views of others is still a lot better than going it alone.

We hope you enjoyed this article, if you did we encourage you to share it with your peers.

 Click here to view our most recent newsletters and to sign up to receive future editions.