After analyzing two years' worth of tweets by 2.4 million people around the world, researchers at Cornell University have concluded that individuals wake up happy but that their mood deteriorates as the day progresses.
That discovery, among others reported Thursday in the journal Science, will interest researchers who are trying to understand how circadian rhythms and other natural influences shape our states of mind. But the study's primary significance may have more to do with its methods than its results.
"We now have the ability to view societies at a massive scale using the Internet," said study leader Scott Golder, a graduate student in sociology at Cornell. "This will open up opportunities for social scientists."
Golder said he intended to use Twitter to study behavior, not emotion. He and a fellow graduate student wrote a computer program that sampled all Twitter user accounts created between February 2008 and April 2009, collecting up to 400 messages from each account.
The program compiled more than half a billion Twitter messages, none longer than 140 characters. Most were written by English speakers and deemed good candidates for analysis with other software. The researchers looked at keywords in the tweets to figure out what people were doing and used timestamps embedded in the tweets to peg those activities to particular times of day and locations around the world.
They surmised that bacon is more popular than sausage (but eaten at the same time of day) and that a television show about someone named "Oprah" aired at 4 p.m. on weekdays. They estimated that it takes seven hours to become inebriated, based on the lag between tweets about "beer" and tweets about being "drunk."
They also figured out that they could search for mood-oriented keywords just as easily as they searched for behavior-oriented ones, Golder said.
The team employed a well-known text analysis program that is often used by researchers to sort words based on their emotional content; it seeks out words such as "happy," "awesome" and "fantastic" that have positive overtones as well as words like "afraid," "remorse" and "fury" that have negative ones. Sure enough, patterns emerged.
More Read: articles.latimes.com
