Skip to content
FourthSpace

The friendship recession: 23 years of time use

How has the time Americans spend with friends, alone and socializing changed since 2003? Did it fall for everyone or only for some ages? And what lines up with it in time?

Analysis by Andrew Ly (opens in a new tab) · Updated

Key finding

60.0 (2003) to 26.5 (2025)

Average minutes per day spent with friends

Time with friends fell by more than half.

Average minutes a day spent with friends, all Americans 15 and older, including those who saw no friends that day. 2020 covers May to December only. 2025 has no interviews from October 1 to November 12.

  • 3-year average
  • Annual estimate
  • Smartphone owners

atus_with_whom_by_year_age.csv

Findings

  1. Time with friends fell by more than half.

    Americans 15 and older averaged 60.0 minutes a day with friends in 2003 and 26.5 in 2025. Most of the fall came before the pandemic. By 2019 it was already 34.3 minutes. Source: ATUS 2003-2025, TRTFRIEND, weighted.

  2. It fell in every age group.

    In 3-year averages, ages 25-34 fell from 56 to 32 minutes, 35-44 from 35 to 19, 45-54 from 35 to 17, 55-64 from 34 to 19 and 65+ from 42 to 20. An aging population cannot explain a drop that shows up inside each age band.

  3. Young people lost the most.

    Ages 15-24 went from 155 minutes a day with friends in 2003 to 54 in 2025. Their time alone rose from 239 to 299 minutes.

  4. Time alone rose by 49 minutes a day.

    The all-ages average went from 284.8 minutes in 2003 to 334.0 in 2025.

  5. Fewer people socialize on a given day.

    40.1% of Americans socialized or communicated on an average day in 2003. In 2025 it was 30.4%. Daily socializing time fell from 46.7 to 35.0 minutes.

  6. Games and computer leisure caught up.

    They rose from 18 minutes a day in 2003 to 37 in 2025, about the same as socializing.

  7. Evenings with neighbors fell the most in the GSS.

    44.0% of adults spent an evening with a neighbor at least monthly in 1974. In 2024 it was 26.9%. Evenings with friends outside the neighborhood fell less, from 42.4% to 36.4%. Evenings at a bar fell from 18.9% to 12.8%.

  8. The timing lines up with smartphones.

    Adult smartphone ownership went from 35% in 2011 to 91% in 2024 (Pew). The U.S. fell from 11th to 23rd in the World Happiness Report ranking between the 2011 and 2025 data years. These are parallel trends, which cannot show what caused what.

Options

  1. Option A. Lead with the friends number

    My pick

    "60 minutes a day with friends in 2003. 27 today." It is easy to verify and covers the whole country.

  2. Option B. Lead with young adults

    The 15-24 drop (155 to 54 minutes) is the starkest. It fits a younger early-adopter audience.

  3. Option C. Lead with time alone

    "49 more minutes alone every day." It pairs well with loneliness messaging in Project 04.

  4. Option D. Lead with neighbors

    The GSS neighbor drop (44% to 27%) is the most local finding and the closest fit to a place-based app.

  5. Option E. Build a friendship recession index

    Combine ATUS friends time, share socializing and GSS evenings into one indexed line for the deck. More work and less transparent.

Default pick: A, with D as the second slide. The friends number is easy to verify and needs no explanation. The neighbor finding then moves the story from national to local, which is where FourthSpace operates.

Method

  1. 01_atus_time_use.py reads the ATUS summary file. Socializing uses the BLS definition, activity codes 1201 (socializing and communicating) and 1202 (attending social events). Games are 120307 and computer leisure is 120308. Every estimate uses the ATUS final weight (TUFNWGTP), with the 2020 replacement weight (TU20FWGT) for 2020.
  2. 02_atus_with_whom.py uses the respondent-file totals for time alone (TRTALONE), with friends (TRTFRIEND) and with family (TRTFAMILY). These count time in activities where BLS asks who was present, which excludes sleep, grooming and most work. Averages include people who saw no friends that day.
  3. 03_gss_social_evenings.py computes the weighted share of adults (WTSSPS) who spend an evening with each group at least monthly and at least weekly.
  4. 04_pew_adoption.py and 05_whr_us_rank.py load the context series.
  5. 06_charts.py draws the charts. Age panels show a 3-year rolling mean over the annual points.
  6. Validation. The 2024 share of people who socialized or communicated on an average day comes out at 29.9%, which matches the 30% BLS publishes.

Tools

Python 3.13, pandas, NumPy, pyreadstat (GSS Stata file), pyarrow, lxml/pandas.read_html (Pew tables), openpyxl (WHR), matplotlib, Plotly. Downloads with curl on a Mac (fetch_data.sh).

Caveats

  1. Correlation only. Smartphone adoption, the WHR ranking and time with friends moving together does not show cause.
  2. 2020 data cover May through December only, after a pause for COVID-19. 2025 has no interviews from October 1 to November 12 because of the federal shutdown. The hero chart note flags both years.
  3. ATUS "with friends" counts time when a friend was present during a diary activity. It misses phone and online contact, which are coded as separate activities.
  4. Sample sizes have shrunk, from 20,720 respondents in 2003 to 6,146 in 2025. Single-age-group, single-year estimates are noisy, which is why age charts use 3-year averages.
  5. GSS moved to web and mixed-mode interviewing in 2021. Mode can shift answers, so the 2021-2024 points are shaded and should be read with care.
  6. WHR ranks depend on which countries are surveyed each year (137 to 158).

The debate in brief

The tech and isolation link is contested. Writers like Jonathan Haidt point to the overlap between smartphone adoption and falling social time as evidence that phones displaced in-person life. Critics such as Candice Odgers and Amy Orben note that many things changed in the same years, including work patterns, housing, the decline of civic groups and the pandemic, and that time-series overlap cannot separate them. This project shows the overlap and stops there. Project 02 tests the individual-level association with a specification curve.

Downloads

These files come from US federal agencies and are in the public domain. Every other source has its own terms, so it is linked at the publisher under Sources.

Sources

  1. BLS American Time Use Survey, multi-year microdata (respondent, activity, who, summary files)

    Years
    2003-2025
    Geography
    U.S., ages 15+

    https://www.bls.gov/tus/data.htm (opens in a new tab)

    Public domain

  2. General Social Survey cumulative file, Release 3a (items SOCFREND, SOCOMMUN, SOCBAR, SOCREL)

    Years
    1974-2024
    Geography
    U.S. adults

    https://gss.norc.org/get-the-data/stata (opens in a new tab)

    Publisher's terms

  3. Pew Research Center Mobile Fact Sheet (smartphone ownership) and Social Media Fact Sheet

    Years
    2011-2026
    Geography
    U.S. adults

    https://www.pewresearch.org/internet/fact-sheet/mobile/ (opens in a new tab)

    Publisher's terms

  4. World Happiness Report 2026, Data for Figure 2.1 (U.S. rank and life evaluation by year)

    Years
    2011-2025
    Geography
    147 to 158 countries

    https://files.worldhappiness.report/WHR26_Data_Figure_2.1.xlsx (opens in a new tab)

    Publisher's terms

Interactive charts

  • Americans spend about 49 more minutes a day alone than in 2003

  • Evenings with neighbors fell from 44% to 27% of adults since 1974

  • Time with friends has fallen in every age group since 2003

  • Games and computer leisure now match time spent socializing

Keep reading

Phones, social media and teen mental health: a specification curve

Is heavy social media use tied to worse mental health among U.S. high school students? How sure can we be, and how big is the gap next to other things in a teenager's life?

Next study
Heavy social media use lines up with worse teen mental health in 9 of 10 models

Be first through the door.

We're opening in Sacramento first. Join the list and we'll email you when we open.

Fields marked * are required.

Places you'd go (optional)
How we use your details

Study01/07Join the waitlist