Twitter is great for staying up-to-date on, well, pretty much everything: the news, celebrity gossip, your roommate's best-friend's breakfast. But a new paper out today in the journal Science suggests that Twitter can also be used to track peoples' moods. The researchers found that, across the globe, tweets are predictably upbeat or cranky based on the local time of day.
Cornell University sociologists Scott Golder and Michael Macy spent two years collecting 509 million tweets from 2.4 million users in 84 different countries (albeit with a notable dearth of representation from Africa). Using a well-established text analysis tool, they scored tweets based on their use of hundreds of positive words (like "happy" or "enthusiastic") or negative words (like "sad" or "anxious"). When Macy and Golder plotted these scores against the tweet's time stamp, they found what should come as no surprise to anyone who works a nine-to-five: peoples' moods are best early in the morning, slowly deteriorate as the day wears on, then finally pick up in the evening (read: after happy hour). And, cultural differences be damned, the same was true worldwide, suggesting mood is hard-wired in the human psyche.
"Twitter is a goldmine for being able to observe human behavior," Macy said. "We all have basically the same biology, and the pattern we found was very robust."
[Read more in the Blue Marble blog]