Twitter Confirms Late-Day Slump

Twitter analysis reveals the late-day energy low is nearly universal.

THE GIST

Sociologists at Cornell University analyzed a half billion Twitter updates from around the world for positive and negative feelings.

They found that regardless of country and culture, Twitter users' moods followed similar patterns of highs and lows during the day.

The 2.4 million Twitter users analyzed even had late-day slumps on weekends.

It's not just you: That late-day slump is nearly universal, at least in the world of social network users.

Sociologists at Cornell University studied updates from millions of Twitter users, analyzing tweets on an hourly and daily basis. After breakfast, moods from Dubuque to Dubai start going downhill.

"We found a very similar pattern in India, Africa, Europe, the UK, Canada, North America, Australia, New Zealand. The patterns are very, very similar even though all those cultures are different," said Michael Macy, a sociology professor at Cornell who, along with sociology graduate student Scott Golder, co-authored an article about the study in the latest issue of Science.

NEWS: Parody Blooms on Twitter

The sociologists analyzed more than half a million public updates from 2.4 million Twitter users in 84 countries over a two-year period. They divided the updates into groups by hour of the day and day of the week. Then they ran them through a program developed by psychologists to measure feelings. It uses a positive and a negative lexicon that each contain hundreds of words like "happy," "excited," "fearful" and "sad."

"People are most upbeat around breakfast time," Macy said. "Their mood deteriorates over the course of the day and then rebounds in the evening. Around dinnertime on through the time they go to bed it picks back up again."

At first the sociologists wondered if this was due to work-related stress because overall moods tended to be elevated on weekends. But they noticed that even then, tweets expressed late-day mood dips, even in the United Arab Emirates where weekends are Fridays and Saturdays.

"It could be that the elevated mood on the weekends is because they did not get woken up by an alarm clock, which would have disrupted their natural sleep patterns," Macy notes.

He cites strong evidence from previous research how important sleep is as a factor influencing feelings throughout the day.

But how can searching for words in tweets tell the scientists how a person is feeling -- don't they need more context? Macy says it doesn't matter who or what someone is tweeting about. The psychological lexicons, which have been validated, determine general mood based on whether the words Twitter users choose to use are particularly positive or negative.

"Social and behavioral scientists have these wonderful opportunities now to study human behavior and human interaction in ways they could not do before," Macy said.

Next, Macy says they are focused on using the enormous Twitter data to study behavior. Golder recently created a public website, Timeu.se , so anyone can do a keyword search on their data and plot the results. One of the sociologists' early findings: "happy hour" starts around 4 or 5 Monday through Thursday, but on Fridays begins after lunch and keeps going until 10 at night.

David Lazer is an associate professor of political science and computer science at Northeastern University. His group has determined countrywide moods based on Twitter updates, but the study didn't look at individuals' daily variations.

"The new thing here is that they're clearly finding some sort of signal," Lazer said of the Cornell article. "Part of what's beautiful about these data is that they truly offer a picture of the system as opposed to detached observations you might get in a standard survey."

When paired with traditional questionnaires, the Cornell results will give social scientists a better understanding of daily human rhythms, said Thomas Streeter, a sociology professor at the University of Vermont who studies media culture and technology.

"Definitely there's a pattern there," he said. "But of course they're talking about people who use Twitter. It's not farmers in Zimbabwe."

Social media can be ideal for expressing certain kinds of human expressions but not for others, he added. "We have to be careful to not read them as hot lines to our hearts."