Why Young Adults Report More Loneliness Than the Old

The 2021 Harvard survey from Making Caring Common found that 61% of adults aged 18 to 25 reported serious loneliness, against 36% of Americans overall. Cigna’s 2020 workplace report found 73% of workers aged 18 to 22 reporting loneliness sometimes or always, and over 80% among employed Gen Z specifically. Both figures run in the same direction and against the folk assumption that loneliness is chiefly an old-age problem.

That assumption was not baseless. Isolation and loneliness were studied for decades primarily as geriatric concerns, tied to bereavement, retirement, and physical decline. The AARP Foundation’s 2018 survey of adults 45 and older, using the 20-item UCLA Loneliness Scale, found one in three lonely — a substantial figure, but well below what the youngest cohorts now report in separate surveys. The National Academies’ 2020 consensus report put isolation, specifically, at roughly a quarter of adults 65 and older. Isolation and loneliness are not the same measurement, and the reports are not always comparable instrument to instrument, but the direction across them is now consistent: younger cohorts report more loneliness than older ones, not less.

This brief takes that pattern as established and asks what explains it, because the explanations on offer carry different policy consequences.

The pattern is not new, but it is underappreciated

The Harvard survey’s headline number — 61% of 18-to-25-year-olds seriously lonely — sits alongside a subtler finding: 43% of young adults said their loneliness had increased since the pandemic began, meaning a majority did not attribute their loneliness to the pandemic at all. Roughly half of lonely young adults in that survey said no one had taken more than a few minutes recently to ask how they were doing in a way that felt genuine. That is a measure of perceived relational quality, not just contact frequency, and it points toward something that predates 2020.

The American Enterprise Institute’s Survey Center on American Life reported in mid-2021 that 12% of Americans have no close friends at all, up from 3% in 1990. The decline is sharpest among men: the share with at least six close friends fell from 55% in 1990 to 27% in 2021, and 15% of men reported no close friends whatsoever, a fivefold increase over three decades. That survey was not limited to young adults, but it establishes that the erosion of close friendship networks has been underway across cohorts for a generation, not manufactured suddenly by any single event.

Four candidate explanations, and what distinguishes them

Life-stage disruption. Young adulthood involves repeated network rupture — leaving secondary school, leaving university, changing cities for work, the collapse of proximity-based friendship that Robert Putnam’s earlier work on associational decline documented at the population level. Under this account, elevated youth loneliness is partly mechanical: young adults have had less time to accumulate the stable, redundant social ties that buffer older adults. This explanation predicts loneliness should decline with age as networks stabilize, and the Harvard data are consistent with that shape, though a single cross-sectional survey cannot show the trajectory of the same cohort over time.

Structural isolation versus subjective loneliness. The National Academies’ report and the AARP survey both measure meaningfully higher isolation among older adults — reduced network size, reduced contact frequency — even where reported loneliness is lower. Older adults may be more isolated by network structure while feeling less lonely about it, and young adults may be more socially embedded by measures like weekly contact while feeling lonelier regardless. This is the isolation-versus-loneliness distinction doing real work: they are different constructs, independently predictive of outcomes, and a policy response built on one will not necessarily address the other. Holt-Lunstad’s 2015 meta-analysis in Perspectives on Psychological Science found that both isolation and loneliness independently predicted early mortality — odds ratios of 1.29 and 1.26 respectively — even after adjusting for health status, and found the effect was more pronounced in samples averaging under 65 than in older samples. That last finding cuts against a purely developmental story: if loneliness were simply a stage young people age out of, it would not track this strongly with mortality risk in younger samples specifically.

Comparison and disclosure effects. Younger respondents may simply be more willing to report loneliness on a survey instrument, whether because of reduced stigma, greater familiarity with mental-health vocabulary, or generational differences in what counts as an admission. This is a measurement-artifact explanation, and it is difficult to rule out with existing data. No source in this review directly tests whether willingness to disclose loneliness varies by age independent of experienced loneliness, and this is a genuine gap. It would require, at minimum, a validated behavioural or physiological correlate measured alongside the self-report across age groups, which none of the surveys cited here attempt.

A cohort effect distinct from a life-stage effect. The AEI friendship data suggest something happening within generations over calendar time, not just within individuals over their own lifespan: the decline in close friendships since 1990 affects people who were themselves young adults in 1990 as well as those who are young adults now. If today’s young adults are lonelier than young adults were three decades ago — and the AEI comparison implies they may be — then the explanation is not that youth is inherently lonely, but that the conditions of being young have changed. Fewer third places, thinner associational membership, and the compounding effect described in Robert Putnam’s account of declining civic participation would all sit under this heading. This explanation predicts that today’s older cohort, when they were young, would not have reported loneliness at 61%, and there is no directly comparable historical instrument to test that. The 1990 AEI comparison point is suggestive but concerns friendship counts, not the UCLA scale or a loneliness-specific measure, so the comparison across decades is approximate rather than exact.

Why the distinction matters for policy

These four explanations are not mutually exclusive, and the true account is almost certainly some mixture. But they point toward different interventions.

If the pattern is life-stage disruption, the policy response is transitional support — structured ways to rebuild networks after a move, a graduation, a job change — and the expectation should be that loneliness resolves somewhat on its own as networks stabilize, without discounting the acute risk in the interim.

If it is a genuine cohort shift tied to the erosion of associational life that Putnam and, separately, Eric Klinenberg’s account of social infrastructure both describe, the response has to be structural: rebuilding the physical and institutional spaces — Ray Oldenburg’s “third places” — where casual, recurring contact happens, rather than treating loneliness as a problem to be solved one individual at a time.

If it is a measurement artifact of disclosure norms, the urgent task is not intervention but better instrumentation, since the apparent age gradient may be partly or wholly illusory.

Former Surgeon General Vivek Murthy’s 2020 book argued for treating loneliness as a public health issue with population-level causes rather than a personal failing, which implicitly favors the structural and cohort-based explanations over the disclosure-artifact account. That is a reasonable prior given the AEI trendline, but it is a prior, not a settled finding. Holt-Lunstad’s 2021 review in the American Journal of Lifestyle Medicine argues for treating social connection as a modifiable preventive factor on par with diet and smoking — language that fits a life-stage or structural account better than a pure measurement story, since disclosure artifacts are not obviously modifiable in the same sense.

What would settle this

The clean test is a longitudinal panel that follows the same individuals from age 18 into midlife using a single consistent instrument — the UCLA scale, applied identically at each wave — paired with repeated measurement of the same age band across calendar years to separate cohort effects from life-stage effects. None of the surveys reviewed here does this: the Harvard, Cigna, and AARP instruments are each single cross-sections, and the AEI friendship comparison spans decades but changes what it measures loosely enough that it is suggestive rather than conclusive. Until that panel exists, the honest position is that young adults report more loneliness than older adults on every current instrument, and that the leading candidate explanations imply substantially different policy responses that the cross-sectional data cannot yet arbitrate between.