A Cornell study published Friday in the journal Science used Twitter to study changes in peoples’ moods, discovering that seasonal variations in day length affect people in similar ways across cultures, indicating there is an underlying biological rhythm at work.
The authors of the study, Scott A. Golder grad and Prof. Michael Macy, sociology, analyzed two years’ worth of tweets by 2.4 million people around the globe. In total, the researchers analyzed 509 million messages.
While Macy notes there have been other studies that use Twitter to assess moods, he said many of these studies tended to focus on individuals’ moods on particular dates such as holidays, or instead “take the pulse of the entire population.” They do not provide insight into the daily rhythms of an individual, he said.
Golder said that the team assessed the moods of millions of twitter users by using Linguistics Inquiry and Word Count technology, which measured the percentage of positive and negative words at each hour of the day.
The study finds that across cultures, there are peaks in positive tweets in the early morning and late at night, while tapering occurs in the late afternoon. The research notes that overall moods were lowest at the beginning of the workweek and rose as the week continued, peaking at the weekend.
“People have a tendency to say, ‘Of course, these findings are obvious,’” Macy said. “People probably look at the morning peak of positive moods and how it deteriorates throughout the day and attribute that to working all day. But in fact, what we found is that a hard workday is not enough to explain the trend.”
The study also found that as days get longer, moods tend to be more positive. They become increasingly negative as days shorten.
“We found that it is not the absolute length of the day that was associated with mood, but whether the days were getting longer or shorter,” Macy said. “With regards to seasonal affective disorder, it is not the amount of sunlight that is correlated with mood. It is the amount of daylight relative to surrounding days.”
Macy said that even during the weekends, the same mood patterns hold true, signifying that factors including sleep patterns or underlying biological rhythms contribute to these findings.
Macy said that one of the most important implications of his research is the use of innovative methods to study human behavior.
“The methods in the past included surveys, field research and lab experiments that either failed to record human behavior in real time or were based on extremely small samples,” Macy said.
“What Twitter and other social media make possible for the first time is the chance for social scientists to observe human behavior in real time with an enormously large scale,” he said. “That is really an extraordinary opportunity.”
Golder echoed this sentiment, adding that the study reflects larger changes in social sciences research.
“Really what this study says to us is that more and more of the future of social science is going to be involving data from the Internet,” Golder said. “There will always be room for surveys and interviews, but these new methods of using Internet data are going to start to become really important. Students studying sociology, economics or political science have the incredible opportunity to enter these fields at a time when they’re completely changing.”