Social media users swayed by previous ratings
When rating things online, people tend to follow the herd. A single random “like” can influence a comment’s score at a social news site, researchers report in the Aug. 9 Science.
Users of the site discuss news articles and rate each other’s comments with “up votes” (positive ratings) and “down votes” (negative ratings). Votes affect each comment’s overall score. To test whether previous ratings sway users, Sinan Aral of MIT and colleagues randomly assigned all comments submitted to the site over a five-month period an up vote, a down vote or no vote.
An unearned up vote packed a surprising punch. The first person to view a randomly liked comment was 32 percent more likely to rate it positively than to do the same with a comment that had received no vote. In the long run, boosted comments’ final scores were 25 percent higher than scores of untouched comments. Random negative votes did not affect a comment’s final rating because users compensated with extra up votes.
The findings may help researchers analyze herding behavior or manipulation in other kinds of rating systems, including electoral polls and stock market predictions, the authors suggest.
Editor's Note: A misstatement about down votes in this story was corrected on August 9, 2013.
L. Muchnik et al. Social influence bias: a randomized experiment. Science. Vol. 341, 9 August 2013, p. 647. doi: 10.1126/science.1240466. [Go to]