Med thumb 3796559729 72527e0d3f o  2

This week, the Bureau of Labor Statistics reported that Americans sleep 8.8 hours a night in its annual American Time Use Survey (ATUS). The data is consistent with last year's survey, which offered nearly identical findings. The high estimate has prompted claims that "we're getting more sleep than we think" and that "we're not actually an underslept country." But these claims fail to recognize the ATUS sleep data for what it is: an unexplained outlier. 

The ATUS is hardly the only national survey with intel on American sleep habits. The Centers for Disease Control (CDC) asks about sleep in the National Health Interview Survey, the Behavioral Risk Factor Surveillance System and the National Health and Nutrition Examination Survey. The National Sleep Foundation has published its annual Sleep in America poll since 2002. Gallup released a sleep poll in 2013. And the list goes on. These surveys don't phrase their sleep-related inquiries in exactly the same way, but they more or less ask people how much sleep they get on weekday and weekend nights. And across the board, they report consistent findings: Americans sleep somewhere around seven hours a night, and slightly more on weekends. 

For the sake of health outcomes, it's more important to know how much sleep people are getting than how much time they reserve for Zzzs.

Then, there's the ATUS, whose nearly nine-hour figure deviates considerably from the herd. The ATUS findings dominate headlines because they challenge the notion that sleep is the epidemic it's made out to be. As we wrote previously, news stories referencing the ATUS have basically reported results without much analysis or argued that ATUS numbers are the unbiased, accurate estimates. Last fall, the BLS declined to discuss the discrepancy between their numbers and everyone else's. So, we came up with our own hypothesis to explain the deviant data: The survey's primary focus on time use, and method of collecting information about daily activities, broadens the category of sleep to include time spent in bed.

Here's how the ATUS works: Rather than estimate nightly sleep duration, ATUS respondents recount what they did between 4am and 4pm on the previous day. Their answers get coded as one of 12 activities, including sleeping, grooming, working, preparing meals, eating, doing laundry and "none of your business." If respondents say they were doing more than one activity at the same time, they must either assign the activities different time slots or choose the main activity. 

It seems possible, if not likely, that snoozing and attempting to sleep are considered the same activity, either by respondents or the BLS. After all, the ATUS is an economic survey, not a health survey. And its goal is determining how people use their days, rather than gaining insight about population-level health trends. 

It's not clear that asking people how much they sleep is more or less biased than asking them what they're doing in the middle of the night. But, the two questions are different. For the sake of health outcomes, it's more important to know how much sleep people are getting than how much time they reserve for Zzzs. Of course, we're speculating. But we should question outliers before we accept them as true.