Social Influence and the Wisdom of the Crowds
During the early 1900s, British scientist Francis Galton was intrigued by a contest involving 800 people and their behavior in the “guess the weight of the ox” contest. Each person would submit a guess, and the closest guessers received prizes. After the guessing contest was over, Galton averaged all the votes and got a very surprising result. The average guess was 1,197 pounds while the true weight of the ox was 1,198 pounds. Galton’s experiment highlights the “wisdom of the crowds” theory: When under the correct circumstances, intelligent crowds can possibly be smarter than even the most knowledgeable individual (Galton).
Although we are still unsure about the correct “circumstances” for this phenomena, one thing is for sure: the “wisdom of the crowds” certainly doesn’t exist in the presence of social influence effects between members of the “crowd.” Very recently, a group of researchers from Zurich demonstrated that even mild social influence can undermine the “wisdom of the crowd” effect in simple estimation tasks. In their study, subjects were given simple estimation questions, such as “What is the population density of Switzerland?” Some subjects, however, were allowed to reconsider their initial estimates after seeing the guesses from previous subjects (Lorenz). Their results can be summarized as follows:
(1) Smaller Variation / Convergence of Estimates – FIG 1: Subjects who received no information had guesses all over the place, and their guesses never converged to a final value. Their guesses, however, turned out to be more accurate. Subjects who had access to previous guesses experienced rapid convergence and significantly less variation. However, their final estimates turned out to be significantly far from the true value in most cases (Lorenz).
(2) Increased Confidence – Subjects who did/didn’t have access to previous guesses were also asked to rate their confidence for their individual guesses. Subjects with access to the previous guesses had an average confidence several times greater than subjects without knowledge of the previous guesses (Lorenz). This illustrates how just a tiny bit of social influence can make a huge change in not only the response accuracy, but also each individual’s confidence in their guess.
(3) Decreased Accuracy – FIG 2: In the presence of social influence, the initial estimates caused an information cascade for the rest of the subjects, resulting in rapid convergence, yet significant loss in accuracy. The table below highlights how just a tiny bit of social influence caused such as drastic decrease in accuracy (Lorenz).
To summarize, recent studies from Swiss researchers showcase how social influence can undermine the “wisdom of the crowds” phenomena. A possible explanation for this interesting result could be the idea of an “information cascade,” which happens when people behave the way others have behaved and ignore their own beliefs/ideas. Looking back onto this study, it is quite amazing how such a minor amount of social influence can produce herding behavior (which can be a good or bad thing). Given how highly connected our world is today, these results hinder the effectiveness of the “wisdom of the crowds” theory, and also showcase the increasing need to consider social network effects in our society today.
References