Twitter Helps Determine Morning People and Night Owls
|Using data from Twitter, researchers analyzed 509 million messages from 2.4 million users in 84 countries to explore the daily, weekly and seasonal variations in people’s mood cycles. Courtesy of © 2011 Jupiter Images Corporation|
It’s true. The daily grind dealing with bosses, colleagues and repetitive work sours people’s moods. But researchers say the cause may be something more than the work itself; people’s biological clocks may be sending a message.
“Though it might seem intuitive to suggest that the decrease in mood level during the midday hours is a result of workday-related stress,” said Scott Golder, lead researcher for a study appearing today in the journal Science, “it turns out we see the same rhythmic shape on the weekends, when people typically are not working. This suggests to us that something more enduring is going on, such as the effect of biological processes and sleep.”
Golder, a graduate student in sociology at Cornell University, and sociology professor Michael Macy, recently analyzed text messages from 2.4 million users of the online social networking service Twitter to explore the daily, weekly and seasonal variations in the mood of people from 84 countries around the world.
Using Twitter.com’s data access protocol, Golder and Macy collected up to 400 public messages from each user in the sample for a total of more than 509 million messages authored between February 2008 and January 2010. The researchers excluded users with fewer than 25 messages.
Twitter’s 140 character limit on message length allowed them to chart the “positive” and “negative affect,” or mood, of user communications using a prominent lexicon for text analysis, the Linguistic Inquiry and Word Count (LIWC). LIWC permitted measurements of an individual’s spontaneous expressions, i.e. his or her positive affect — enthusiasm, delight, activeness and alertness — and negative affect — distress, fear, anger, guilt and disgust.
Golder and Macy found that individuals awaken in a good mood that deteriorates as the day progresses, which may be more consistent with a person’s 24-hour physiological and behavioral cycle than any specific activity.
“Positive affect has a valley during the middle part of the day, rebounding at night,” said Golder. “Because we see these trends every day of the week, for people across the globe, this suggests that there are biological processes driving these rhythms, and also that people are refreshed by sleep, overnight.”
Moreover, the findings were consistent across all the countries studied, with different cultures, backgrounds, and so on. “We were amazed to see just how much consistency there was in the shape of affective rhythms globally,” said Golder.
The researchers saw the same basic peaks in positive affect in the morning and evening and troughs in negative affects early in the morning that rise throughout the day.
“We also saw evidence of work week-related stress as the weekend exhibited higher levels of positive mood, regardless of which days actually constitute a culture’s weekends,” Golder said.
For example, diurnal, or daily, rhythms in four groups of countries: the United States and Canada; the United Kingdom, Australia, Ireland and New Zealand; India; and English-speaking Africa were statistically similar. People in these countries had higher positive moods on Saturday and Sunday compared to the rest of the week.
The finding suggests that despite the differences in time of day that an individual’s physical functions are active — hormone level, body temperature, cognitive faculties, eating and sleeping — most people have similar mood cycles across cultural norms.
The researchers established support for their conclusion by examining text messages from the United Arab Emirates (UAE) where the traditional work week runs Sunday-Thursday. They found people in the UAE had higher positive moods on Friday and Saturday and lower positive moods Sunday through Thursday.
Additionally, Golder and Macy assessed people’s moods related to changes in seasons. Previous research on seasonal mood changes relied on small samples within single countries and was severely constrained by the difficulty of collecting data over an entire year. However, prior research found higher occurrences of depressive anxiety in winter the further north people were situated.
Although originally attributed to insufficient exposure to light, more recent research on seasonal mood variation supports the “phase shift hypothesis,” which argues that seasonal affective disorders, such as loss of energy, depressed mood and daytime fatigue, are largely based on abnormal delays in synchronizing a person’s body clock with early bright light during winter.
Golder and Macy found a “striking effect” in mood related to changes in daylength. Average positive mood increased when daylength increased, as the summer solstice approached, but decreased as the winter approached. Average negative mood did not increase or decrease seasonally.
“This suggests that ‘winter blues’ is associated with decreased positive effect, not increased or decreased negative effect,” said Golder. In their report, “Diurnal and Seasonal Mood Vary with Work, Sleep and Daylength Across Diverse Cultures,” the researchers contend that the increased positive affect approaching the summer solstice may correspond to longer days and earlier light, thereby reducing the discrepancy between social and biological timing.
The National Science Foundation’s (NSF) Division of Behavioral and Cognitive Sciences in its Directorate for Social Behavioral Economics and Division of Information & Intelligent Systems in its Directorate for Computer and Information Science and Engineering funded the research.
“In addition to the substantive findings about mood rhythms, we think this research is interesting because it is an example of a relatively new methodological approach, the use of the Internet to do social science research,” said Golder.
“This well-focused study is at the vanguard of progress using new methods to study increasingly sophisticated scientific questions online,” said William Bainbridge of NSF’s Human-Centered Computing program, “and on the basis of the knowledge gained to design improved social media for the emerging information civilization.”
“People often critique the Internet and social media for being mundane,” said Golder. “However, we suggest that most of the conversations people have with friends, family, co-workers and so on, are likewise mundane–we just don’t take the time to record them. But the Internet records everything, and these “digital traces” of activity become important tools in analyzing everyday social behavior.”