An analysis of 240,000 Americans reveals that the rise in alone time began long before the smartphone boom. It points to age, life stage, and social change as the real forces behind America’s loneliness epidemic.
Study: In-person social isolation in the age of smartphonies: Examining age, period, cohort effects by gender. Image credit: Valebru/Shutterstock.com
In a recent study in PLoS One, researchers examined how birth cohort, time period, and age shape the time that people spend alone when social media and smartphone use are widespread.
Their findings show that social isolation has increased sharply over the past two decades, accelerating since the mid-2010s. However, smartphones alone cannot explain these changes, with generational differences and aging contributing to isolation.
Background
The U.S. Surgeon General described isolation and loneliness as a national epidemic in 2023, pointing to online entertainment, remote work, smartphones, and social media as contributing factors. While these technologies allow new ways to stay connected, critics argue they reduce face-to-face connections and weaken real-world support networks.
Evidence cited for rising isolation often comes from time-trend studies, such as the American Time Use Survey (ATUS), which shows a steady rise in the time that people spend alone since 2003. However, these analyses may confuse three separate influences: age (isolation tends to rise later in life), period (broad societal changes such as new technologies), and cohort (differences between generations).
Research shows that age may have a U-shaped effect, with social isolation rising notably after midlife due to retirement, loss of peers, and other life transitions. Cohort effects may also play a role: younger generations who grew up with smart devices may rely more on digital than in-person interactions. In comparison, older generations maintain different social patterns.
Prior evidence on generational differences is mixed, with some studies suggesting younger cohorts are less socially engaged offline, and others showing stability in family and friend connections.
About the Study
Researchers analyzed ATUS data from 2003 to 2022, covering 240,576 respondents aged 15 to 79. The ATUS, conducted by the Census Bureau, collects nationally representative 24-hour diaries of daily activities through telephone interviews.
Respondents report what they did, how long activities lasted, where they occurred, and with whom. Social isolation was measured as minutes spent alone during non-work activities, since work-related “who” data were inconsistently available before 2010 and could bias comparisons.
Cohort, age, and period groups were created in five-year intervals: age groups (15–19 through 75–79), four periods (2003–2007 through 2018–2022), and sixteen birth cohorts (1924–1932 through 1999–2007). The models also controlled for weekends versus weekdays and holidays, since social contact patterns vary by day type.
The researchers applied the Fosse and Winship age-period-cohort (APC) method to separate the influences of cohort, age, and period. This approach overcomes the statistical challenge that age = period – cohort by distinguishing nonlinear effects and estimating a “canonical solution line” of possible linear effects.
Weighted regression models, adjusted for sampling design and non-response, estimated the net contributions of cohort, age, and period. Analyses were stratified by gender, with results visualized using two-dimensional APC plots.
Key Findings
The researchers found that Americans are spending increasingly more time alone in non-work activities. For men, time spent alone rose from 268 minutes in 2003 to 312 minutes in 2022, while for women it increased from 282 to 297 minutes.
This growth accelerated around 2013–2014. Age patterns followed a U-shaped curve, with social contact highest in the mid-30s before declining steadily into older age. Women’s time alone rose sharply to 500 minutes by age 79, while men plateaued at around 451 minutes.
Cohort effects revealed that older cohorts (born before 1940) spent the most time alone, with levels dropping significantly among those born in the 1970s–1980s and rising again for the youngest cohorts.
However, the authors emphasize that it is difficult to fully disentangle cohort effects from age effects in the ATUS data, meaning that some generational differences may reflect life-course patterns rather than actual cohort change.
Advanced APC modeling confirmed that period effects (societal changes such as the spread of smartphones) did contribute to rising isolation, particularly after the mid-2010s. Yet cohort and age effects were much stronger influences. Quantitatively, the age effect was about five times larger than the period effect, roughly a 150-minute difference between mid-30s and late-70s adults, compared to a 30-minute rise over the 2003-2022 period.
Gender differences emerged, with women showing steeper increases in isolation later in life, partly reflecting higher widowhood rates. Sensitivity tests (e.g., weekends vs weekdays, pre-pandemic vs post-pandemic, leisure time vs nonwork time) confirmed the robustness of these patterns.
Conclusions
This study demonstrates that while societal shifts, such as smartphone adoption, have contributed to rising social isolation, age and cohort dynamics are far more influential. The strongest finding is the U-shaped age effect, where isolation is lowest in the mid-30s and peaks in older adulthood, particularly for women.
This suggests that aging processes, including retirement, family changes, and widowhood, drive much of the isolation crisis. Cohort effects also indicate that older generations experience substantially higher isolation than younger ones, although these generational interpretations are tentative, given the overlap between age and cohort influences.
Key strengths include using two decades of nationally representative ATUS data and applying advanced APC modeling, which allows a clearer separation of cohort, age, and period influences.
However, limitations remain, including reliance on cross-sectional data, difficulties distinguishing age from cohort effects, and modest assumptions required in APC modeling. Overall, the findings reinforce the Surgeon General’s warning and highlight older adults as a priority group for interventions to reduce social isolation.
Chronic insomnia linked to faster cognitive decline and brain changes