An analysis of over 500 million tweets over 2 years in 84 different countries confirms the common wisdom — people tend to be happiest first thing in the morning before the day begins, and on the weekends.
Researchers say this provides proof that scientists can use Twitter as a reliable research tool. Others are more skeptical, suggesting that such data is “low-lying fruit,” since most people express such simplistic emotional states readily.
Cornell University graduate student Scott Golder and sociologist Dr. Michael Macy analyzed English-language tweets from 2.4 million people — approximately 510 million messages of the brief, conversation-like exchanges that occur on Twitter. The researchers used a computer program that searched for words indicating positive mood and negative mood.
Words like “happy,” “enthusiastic,” “brilliant” indicated a positive mood, while words like “sad,” “anxious, “fear” indicated a negative mood.
The researchers measured the absolute amount of positive affect and negative affect in messages on the service, and how that varied by hour of the day, day of the week and season. Their findings were remarkably consistent across time zones and, by inference, across nationality and culture.
The new study found that a positive attitude peaks early in the morning, but starts to dip mid-morning before lunch. It appears to rise again as people approach bedtime.
Specifically, positive feelings peaked between about 7 and 9 a.m., then descended between 3 and 6 p.m. before rising again and peaking about midnight.
Negative feelings, on the other hand, were less commonly expressed by Twitter users. They also showed much less variability. They were lowest early in the morning, rising slightly throughout the day. And just like good feelings, bad ones rose at night, peaking about 10 p.m.
“It illustrates a new opportunity for doing social and behavioral science in ways that were really unimaginable even five years ago,” Macy told the Associated Press.
The study used a non-randomized sample of English-speaking users. Twitter users are furthermore more likely to be younger, more affluent, and more well-educated than the general population. There are no studies demonstrating whether a Twitter user who sends messages regularly — one who “tweets” — is in any way representative of the general population. Especially since most Americans don’t use Twitter.
Their study appears in Friday’s edition of the journal Science.