Data Shows that Positive Content Does Better on Social Media
Positive tweets are favorited as much as five times more than negative or neutral ones.
Image via Pixabay user kaboompics
Within minutes of logging onto Facebook and Twitter, users are typically confronted with all manner of content. A good chunk of that content is negative—violent stories, dire political predictions, reports on controversial figures or terrorism—and acts like lighter fluid to a slow burning fire in the way it triggers nasty exchanges. Though it may seem that the varieties of negative content dominate social media, a new data-based study suggests otherwise.
Published September 30 in the PeerJ Computer Science journal, the study reports that positive content spread via social media is shared more often and reaches a larger audience than negative content.
The study’s researchers, Emilio Ferrara and Zeyao Yang, took Twitter as their subject, analyzing all public tweets produced during September 2014 that featured URLs or media content (photos, videos, etc.). In total, the two sent 19,766,112 tweets through SentiStrength sentiment analysis (or opinion mining) algorithm, known for its advantages in annotating “short, informal texts, like tweets, that contain abbreviations, slang, and the like,” as well as analyzing emoticons, negations, and booster words like “VERY happy.”
Ferrara and Yang focused their efforts on finding out what effects sentiment (emotion) has on the spread and popularity of social media posts. They then shifted their attention to entire conversations, categorizing them into different classes depending on their evolution over time.
While the researchers found that negative content spreads faster than other content, they also found that Twitter users preferred positive content.
“The positivity bias (or Pollyanna effect) rapidly kicks in when we analyze how many times the tweets become retweeted or favorited,” Ferrara and Yang report. “[I]ndividuals online clearly tend to prefer positive tweets, which are favorited as much as five times more than negative or neutral ones; the same holds true for the amount of retweets collected by positive posts, which is up to 2.5 times more than negative or neutral ones.”
Ferrara and Yang also discovered that in general, highly anticipated events (movies, sports matches, etc.) generate positive reactions, while “unexpected events are often harbingers of negative emotions.” Elections and some sports events, which may trigger “flame wars,” are a couple of exceptions. More transient events, whose durations are very brief, “represent the norm on social media like Twitter and are not characterized by any particular” emotion.
“Recent events, going for tragic episodes of terrorism, to the emergence of pandemics like Ebola, have highlighted once again how central social media are in the timely diffusion of information, yet how dangerous they can be when they are abused or misused to spread misinformation or fear,” Ferrara and Yang conclude. “Our contribution pushes forward previous studies on sentiment and information diffusion and furthers our understanding of how the emotions expressed in a short piece of text might correlated with its spreading in online social ecosystems.”