Home > Social Media > Testing the Facebook ‘bandwagon…

Testing the Facebook ‘bandwagon liking’ theory

The Internet is an ocean of public opinion – on any topic really, from the high priority, like how the government is spending our tax money, to something as trivial as Jennifer Aniston allegedly switching plane flights to avoid Angelina Jolie – and the only way to sift through all the noise is to gravitate toward those that offer the most commentary. Following the crowd is a tried-and-true response in real life, and in the virtual world as well, according to a study recently published in the latest edition of Science.

To gauge how reliant online users are on other people’s views, researchers went on an unnamed social news discussion website and randomly manipulated user comments  – they made four percent of the comments “lucky” by awarding them a +1 rating when they were created (0 being the normal rating) and two percent “unlucky” by giving them a -1 rating. They then observed how users who viewed the lucky, normal, and unlucky comments rated the comments in order to evaluate if the spike created did indeed cause a snowball effect on their popularity.

“There are a number of findings, but the headline result is that the up-vote treatment yielded long run ratings which were 25 percent higher than the control group,” shares Sean Taylor, one of the researchers involved in the study currently working on his PhD in Information Systems at NYU’s Stern School of Business. “Early positive ratings snowballed into larger effects due to users herding on the positive scores.” Taylor adds that in contrast, the down-vote treatment caused some users to be more positive (which he defines as a “correction effect”) and others to be more negative (a “herding effect”). “They cancel out in aggregate and the down-vote treatment didn’t change long run ratings.”

“People are more skeptical of negative social influence,” explains Sinan Aral, an associate professor at the MIT Sloan School of Management who’s also Taylor’s advisor and part of the research team. “They’re more likely to ‘correct’ a negative vote and give it a positive vote.”

So do people like or comment on things that they have genuine interest in, or are they mostly driven to check out ones that have high popularity? “I think our research shows that it’s a combination of both social cues (such as counts of likes and comments) and content quality and interesting-ness that drives attention and ratings,” shares Taylor. “We saw more rating interactions on the comments where we manipulated the ratings in either direction, even though they should have been equivalently interesting content.”

“However, if it were purely social cues and not genuine interest driving these outcomes, then we would have seen a much stronger effect. Actually, we were unable to detect an effect on the number of replies to comments from our rating manipulations, indicating that deciding whether to reply to a post is probably driven by the content instead of social cues.”

The case of the bandwagon ‘like’

With the social news discussion website the researchers used for their study unknown (though many outlets have already applied the research to Facebook), we wondered if this herd or hype mentality can be measured on Facebook as well. I set up an experiment: Two links and two photo posts were separately posted yesterday, and for one link and one photo post, I asked some of my friends to like or comment on it, to see if it would catch on; I left the other posts be, without soliciting attention for them. For one particular post, I made sure to at least choose one of the topics that were found to cause a positive bandwagon response:

facebook comment experiment

Several hours later, for posts I peddled, only the friends I asked to like actually liked it – there really was no snowball effect. Despite comments saying “I liked this to spread the word, not because I like the topic” or any comment hoping to provoke a reaction, nobody else followed suit, which sort of proves what Taylor said that the urge to reply to a post is driven more by topic (which was utterly horrible) rather than the inclination to follow the crowd’s point-of-view. As for the posts I didn’t ask support for, those only got likes mostly from close friends and/or people who commonly like and comment on my Facebook posts on a daily basis.

Needless to say, my mini-experiment failed, and here’s why: It’s really hard to gauge hype mentality on Facebook from a personal profile – I am not famous. Despite having over a thousand contacts on the social networking site, with the way Facebook and its News Feed is structured now, it’s impossible to ensure that all my friends see my posts. We don’t really have a concrete idea on how the site behaves and how it decides which friends get more exposure. “Facebook’s News Feed changes how many people are shown a post based on the number of likes,” Taylor states. “In our experiment, the comments weren’t ordered or hidden based on the number of ratings.  So in that sense our results are conservative assessment for a system which also changes content distribution based on ratings.”

Is “liking” still a reliable action, or are we better off disregarding it all together? “It’s probably somewhere in between. We conclude that early ratings affect subsequent ratings, so it’s probable that when a post has a lot of likes, some of them were caused by the early likes. On the other hand, content quality is an important component of the rating decision – likes are still genuine in the sense that you can’t reliably make someone click ‘like’ by showing them something with lots of likes,” says Taylor.