Americans Are Sleeping More, If Not Necessarily Better

(Bloomberg Opinion) -- Last year was pretty terrible. Hundreds of thousands of Americans died of a scary new disease, tens of millions caught it and virtually all of us had our lives upended in some way. But hey, at least we seem to have gotten more sleep!

These data are from the American Time Use Survey, among the most wondrous of the statistical products of the U.S. government. The Bureau of Labor Statistics released 2020 numbers last week, and they’ve already gotten a lot of attention. The headline finding announced by the BLS was that the share of employed persons who did their jobs at least partly at home nearly doubled from 22% to 42%. Among the other things that journalists and researchers have discovered in the data so far are that employed mothers with children under 12 spent more time on child care than on their paid jobs, 12% of employed women reported working and taking care of their children simultaneously, and that men did more housework than in 2019 but still less than women.

Me, I was curious about the sleep. It’s not all that surprising that Americans got more of it in 2020, especially since the ATUS total “includes naps and spells of sleeplessness.”  I certainly did (especially if you include the spells of sleeplessness during the night). But given the constant drumbeat of warnings that Americans aren’t getting enough sleep, I wondered if the pandemic year was an anomaly.

It wasn’t, really. The jump from 2019 to 2020 was especially big, but hours spent sleeping have mostly been rising since the first ATUS was conducted in 2003.

Because of the pandemic, the Census Bureau staffers who conduct the time-use survey for the BLS missed almost two months of work, which is why the 2020 results only cover May through December. As a result the BLS isn’t including them in its historical annual time series, comparing them instead just with data from the same time period in 2019. For that and other reasons the pre-2020 trend is more meaningful than last year’s relatively sharp increase, but given that hours slept for May through December 2019 were the same as for the full year, indicating that there isn’t a big seasonal effect, I don’t think including the 2020 number as I’ve done here represents a chart crime.

Having a y-axis that starts somewhere other than zero also isn’t a chart crime (despite what some people say), but yes the percentage changes we’re talking about here are less significant than a quick glance might lead you to think. Sleeping time went up 1.9% from 2019 to 2020, and 3.2% from 2003 to 2019. Those changes are much bigger than the standard errors estimated by the BLS, though, and the rise over the past two decades is pretty evenly distributed by age, with every group under 65 getting more sleep in 2019 than 2003

How does one square this with the widespread view that ubiquitous smartphones and the hectic pace of modern life are resulting in an epidemic of sleep deprivation? One doesn’t, exactly. The ATUS is not the only available measure of how much sleep we’re getting, and not all the available evidence points in the same encouraging direction. The available evidence is, in fact, something of a muddle.

One thing that can be said with some confidence is that we are getting less sleep than in the age before the electric light — what is sometimes called “Edison’s anti-sleep revolution.” Quantitative evidence of this is limited, but Gallup does happen to have asked Americans about their sleeping habits in 1942, when home televisions were rare and many rural areas still didn’t have electricity, and average hours slept was about an hour longer than in recent polls.

The flattish trend in Gallup’s results since 1990 seems compatible with the ATUS results, given that they both show a modest increase in average sleep duration, but the average hours slept reported by Gallup is much lower than the ATUS average. That’s partly because Gallup’s accounting excludes daytime naps and “spells of sleeplessness,” but it may also be that asking people how much they usually sleep inevitably delivers different and presumably less-reliable results than collecting detailed 24-hour time diaries of the previous day, which is what the ATUS survey-takers do. Two recent studies that compared people’s self-assessments of how long they usually sleep with measurements (by actigraph) of how long they actually slept found that the self-assessments tend to err in the opposite direction from the one indicated by the Gallup/ATUS disagreement, though. That is, people overestimate how many hours they usually sleep, with the errors biggest for the shortest sleepers.

Where does that leave us? Well, a 2016 review of every study from 1960 through 2013 that used “objective” methods (either actigraphy or polysomnography) to measure sleep found no trend one way or another in sleep duration. A 2010 review that focused on U.S. time use surveys from 1975 through 2006, including the first four editions of the ATUS, found that average sleeping time was about the same in 2006 as in 1975, but was almost half an hour longer than in studies conducted in 1985 and 1998-1999. The latter review also found that the odds of being a “short sleeper” — that is, getting six or fewer hours of sleep a night — had risen significantly for full-time workers, although not for the population as a whole.

Two more recent studies based on the Centers for Disease Control and Prevention’s annual National Health Interview Survey, which asks respondents how many hours they usually sleep, have also found increases in the percentage of short sleepers over the past decade. One that focused on working adults found that the percentage getting less than seven hours of sleep a night rose from 30.9% in 2010 to 35.6% in 2018, with the prevalence highest among protective-service and military (50%) and healthcare-support occupations (45%). Another found a similar, if somewhat smaller, increase in short sleep among all adults, and that Blacks and Hispanics were more likely to report six or fewer hours of sleep than non-Hispanic Whites. 

One possible reading of this is that sleep inequality may be rising even as average sleep duration is. Protective-service and healthcare-support workers often have little control over their schedules, which in turn often don’t fit their circadian rhythms. Home environment is also an issue, with nearly 11% of Hispanic households and 3.6% of Black households having more than one occupant per room, compared with just 1.4% of non-Hispanic White ones. The financial distress and instability that accompanied and followed the Great Recession may have made sleep harder for many as well.

For Americans in more comfortable situations and in more control of their schedules, on the other hand, the messaging from the media and the medical establishment about the importance of sleep may actually be having an effect. One 2018 study of ATUS data found that the sleep increase since 2003 had been accompanied by a decrease in time spent reading and watching TV or movies (on any kind of device) just before going to sleep. The authors, University of Pennsylvania psychiatry professors Matthias Basner and David F. Dinges, concluded that:

This can be interpreted as a sign of increasing willingness in parts of the population to relinquish these popular prebed activities to obtain more sleep and may in part be attributed to educational campaigns, scientific reports of the health risks of reduced sleep duration, and the increasing interest in sleep loss and its consequences in both the scientific and the popular literature.

See, people do pay attention to the health advice they hear and read! It’s may just be that, in the case of sleep, not everyone is in position to act on it.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

Justin Fox is a Bloomberg Opinion columnist covering business. He was the editorial director of Harvard Business Review and wrote for Time, Fortune and American Banker. He is the author of “The Myth of the Rational Market.”

©2021 Bloomberg L.P.