Sleep Deprived? Twitter Usage Data Can Map How Screwed Up Your Sleep Is

Some counties have it worse than others.

It’s hard to imagine the scientific benefit to be gained from a collection of tweets sent between 4 and 6 a.m., but a team of scientists at the University of Chicago has found one. They’re using the morning tweets sent by thousands of Americans to illuminate the patterns behind why our society struggles with sleep loss, as well as the places in the country where it may be easier to combat its effects.

In a paper published Thursday in Cell, University of Chicago Associate Professor of Molecular Genetics and Cell Biology Michael Rust, Ph.D., and his colleagues turned to anonymized Twitter data to help illuminate patterns of wakefulness across the country. He tells Inverse that they were searching for places where our internal “biological clock” is misaligned with the scheduling demands of modern society. While this disparity between biological and societal clocks has been known to trouble astronauts on NASA missions, it also hits close to home. Here, it’s sometimes called “social jet lag” and is associated with poor health outcomes, including depression and cardiovascular disease.

“So we know we have an internal biological clock. If you don’t have to set an alarm clock, then that clock will give the body signals to wake up and go to bed at certain times,” Rust says. “It’s now possible for people to work schedules that are in conflict with their internal rhythms.”

Rust’s study used Twitter data to help determine these two schedules by analyzing people’s people’s patterns of tweeting (called “tweetograms”) on weekdays and weekends.

The difference between internal clocks and the schedules people keep, as demonstrated by their Twitter data.


Tweets aren’t a perfect way to calculate an internal clock, but Rust isn’t the first to turn to Twitter to gather data about human activity. He explains that the theory behind using social media data to explain sleep patterns is that firstly, it can solve the issue of asking people to self- report their hours of sleep, which is not always reliable. And second, it’s a record of wakefulness — even if just for that brief moment it takes to write 240 characters.

His team’s research found that there’s roughly a 75-minute gap between most Americans’ internal clocks and the clocks determined by their schedules — almost as if we’re living in different time zones from ourselves. As an example of the team’s process, the paper illustrates data from four counties in New York, California, Louisiana, and Minnesota, where tweet frequencies follow a similar curve, peaking around noon and midnight, then dipping into a “trough,” usually in the early morning. To arrive at the final calculation, they applied this logic to 1,500 US counties.

The paper refers to these graphs as tweetograms. 


Going by the tweetograms, the west coast tended to have the longest periods of time in which people weren’t tweeting — 5.5 hours in Orange Country, California, for example. This may indicate that people there are adhering to a slightly normal sleeping schedule. Counties in the east coast, however, were less on-schedule. Suffolk County, New York, which the paper says is a representative example, only had 4.4 hours of solace. In Wayne County, Minnesota, the tweet break was only 3.6 hours. The team also noted that in counties with low tweeting activity, there was a correlation with concurrent data from the CDC’s Behavioral Risk Factor Surveillance System: People in those places reported sufficient levels of sleep.

To take it one step further, Rust used this analysis as a basis to calculate the number of minutes of social jet lag that people in these counties likely experienced, as demonstrated by their late-night tweeting habits. He did this by comparing the troughs on weekends (when most people tend to pick their own waking hours, a loose approximation of what a biological clock might be) and weekdays.

That difference is what yielded his approximation that most Americans experience 75 minutes of social jet lag. But again, the west coast tended to show a less severe pattern: Counties in the Pacific time zone had an average of 56 minutes of social jet lag, compared to counties in Eastern and Central time zones, where people had 77 minutes of social jet lag on average. Rust isn’t sure why this happens, but he can hazard a guess:

“I’m tempted to speculate that somehow with the lifestyle and the environment on the west coast, people are actually being exposed to more daylight, and this is actually helping their clocks be better aligned to the sun,” he says. “I don’t know if that’s actually true, but we do know that by far, the bright light that comes from the sun is a very important cue for your internal rhythm,” he adds.

So in short, he can speculate that it’s not necessarily that west-coasters are making any huge lifestyle changes to help realign their internal clocks with the demands of modern life. Instead, it does seem that the environment may help offset the balance that may continuously plague the rest of us.

Related Tags