How we spend our time.

I think a lot about time. How quickly it seems to pass as you get older. How I use it. Time use and what gets our attention have been part of the national conversation for several years now. We’ve seen a two-decade shift in work and life as broadband, 4G, 5G, and satellite internet penetrated the globe while we acquired supercomputers for our pockets. Then tech companies developed attention-grabbing algorithms. But what do those changes in time and attention look like nationally?

Since 2003, the Bureau of Labor Statistics has asked a rotating sample of Americans to walk an interviewer through a single day of their life, minute by minute, from 4 a.m. to 4 a.m. A quarter-million of those diaries later, the American Time Use Survey is the closest thing we have to a national ledger of ordinary life, making it a useful instrument to trace changes in our daily habits through the arrival of the smartphone, algorithmic social media, and the pandemic. I wanted to plot these out across age, gender, and education level.

The big changes

Comparing twenty-two years of changes, from the first two years of diaries to the most recent two, the reallocation is stark. The average American gained about 45 minutes a day of non-work time spent entirely alone, and nearly half an hour of sleep and personal care. That time is largely replacing leisure away from home, travel, in-person socializing, shopping, and reading.

The work decline is partly demographic. The top line numbers are averages for everyone 15 and older, and the US population is older now than at the beginning of the century, with Baby Boomers aging into retirement; filter to the 25–44 age group and the work decline all but disappears. And the “games & computer leisure” gain is the most undercounted number on the chart, for reasons covered below.

The “male loneliness crisis” jumps out when you drill in. While time spent alone has increased across every demographic group, men aged 25–44 without a college degree are spending 62 more minutes per day alone. College-educated women in the same age group are spending just 12 more minutes a day alone. Most alarmingly, similar increases in time spent alone are seen among young men 15–24 (60 minutes), with young women not far behind (50 minutes).

Tracking the changes over time

The chart below traces every year of the survey through notable milestones: the launch of the iPhone (2007), Instagram’s wide adoption after the Facebook purchase (2012), TikTok (2018), and Covid (2020).

There is some year-to-year noise in the data, but the long-term trends are clear and steady: socializing in person has been on a long-term decline while time spent on screens and games has unsurprisingly grown, with jumps after 2008, 2016, and, of course, 2020.

Ranking our leisure activities

Rank the things Americans do for fun and TV holds first place wire to wire — still around two and a half hours a day — though that masks massive shifts from linear TV to streaming. The changes are seen below it: games and computer leisure (which covers focused screen time, but not passive scrolling) climbs from fifth to second, passing reading around 2011–12 and finally overtaking in-person socializing itself by the most recent sample.

Nobody reading this will be surprised to see reading fall from third to sixth over these twenty-two years. Filter and you will find that the oldest Americans cut the most: while it remains their second leading activity, readers 65+ went from nearly an hour a day to 34 minutes. For the middle-aged (45–64), the drop began right when smartphones arrived. Once upon a time I used to read a few books a month; it was my primary activity while commuting or on vacation. At this point I only read before bed, and to make that possible, I had to charge my phone outside the bedroom.

Notes on the data

(Honest caveat: Claude wrote this part)

The American Time Use Survey samples one person aged 15+ from households completing the Current Population Survey; each respondent reports one 24-hour diary, with weekends oversampled by design. Estimates here are survey-weighted average minutes per day across all days of the week (weight WT06; the special WT20 weight for 2020). Pooled “era” bins combine two adjacent years. Every plotted cell has at least 300 unweighted diaries.

Two collection gaps to know about: 2020 diaries stop in mid-March and resume in mid-May, so that year leans away from the strictest lockdown weeks; and 2025 has no diaries from September 30 to November 11 (the federal shutdown), with a sample about half its usual size. Education groups are restricted to ages 25+.

The deepest caveat is definitional. The ATUS records primary activities — what you say you were mainly doing. There is no “social media” code, and scrolling while doing something else is invisible. Screen numbers here are therefore floors, not totals, and the true crossing of screens past socializing almost certainly happened years before the diaries show it. The story this data tells best is not how much we look at phones — it’s what happened to everything else.

Sources: Sarah M. Flood, Liana C. Sayer, and Daniel Backman, American Time Use Survey Data Extract Builder: Version 3.3 [dataset] (College Park, MD: University of Maryland and Minneapolis, MN: IPUMS, 2025), https://doi.org/10.18128/D060.V3.3; U.S. Bureau of Labor Statistics, American Time Use Survey, 2003–2025.