web analytics

If You “Like” Something Online, Others Might Do the Same

92eRwW3qHkPR-Gp67ZvUJZEmm1T3hlv1FIbtYBJaKy1KAQAA6wAAAFBO_260x196The “like” you give for a Facebook page or the restaurant review you post on Yelp.com might not be able to change the world, but they will actually influence, in one way or another, those who see these information. Lev Muchnik and his colleagues from Hebrew University of Jerusalem teamed up with a news aggregation website and conducted a study on the social influence of different online information.

 The study reveals that positive social influence tends to accumulate and usually lead to herding effect, while negative social influence is often neutralized by user correction. Such effect is significant in areas such as politics, culture and commerce.

 The news aggregation website used in this study allows users to post their own comments and also on other users ‘comments. The researchers rigged the website that every comment submitted by a user would automatically receive an “up” vote (positive), a “down” vote (negative) or no vote at all (control). Then the researchers observed the users’ rating and comments.

 During the continuous 5-month experiment, 101,281 comments were viewed more than 10 million times and obtained 308,515 ratings by other users. The researchers found that their up-voted comments were 32 percent more likely to get another positive vote by the first user to come across compared with their control group. The arbitrary up-voted comments resulted in herding effects that the overall ratings of these comments received a 25 percent increase by the end of the research.

In addition, the up-voted comments were no more or less likely to receive the first “down” vote by other users. This suggests that users tend not to correct the rating manipulation as long as it is positive. While, comments that received down vote by the researchers, on the contrary, were significantly more likely to get an up vote from the next user to come across, which indicates that the users were inspired to correct the rating manipulation when they were down (negative). The researchers noted that this is more likely to occur between friends.

This study is only a start. Muchnik and his co-workers hope that their study will inspire peers’ interests in more sophisticated analysis of social influences on electoral polling and stock market prediction.

Image sourceShutterstock