Why Younger Adults Report More Loneliness Than Older Adults

Gallup’s 2023 survey of 142 countries found that 27% of adults aged 19 to 29 report feeling very or fairly lonely, against 17% of adults 65 and older. That ten-point gap is not an anomaly of one survey. It shows up, in some form, across nearly every major instrument fielded in the past five years, in multiple countries, using different question wordings. It also inverts what most people assume: that loneliness is chiefly an old-age problem, tied to bereavement, retirement, and physical decline.

The pattern deserves scrutiny on its own terms, separate from any single explanation for it, because several explanations are circulating and they do not all point in the same direction.

The pattern, stated precisely

Harvard’s Making Caring Common survey, fielded in the fall of 2020 and published in February 2021, found 61% of young adults aged 18 to 25 reporting serious loneliness, compared with 36% of American adults overall. That figure is elevated by pandemic conditions and should not be read as a stable baseline. But the age gradient it captured predates and outlasts the acute pandemic period. Gallup and Meta’s 2023 global survey of roughly 142,000 respondents across 142 countries, fielded from June 2022 to February 2023, found the same ordering outside the U.S. and outside the crisis window: loneliness declines fairly steadily with age, and connectedness rises with it. Gallup’s October 2023 and 2024 releases confirm the direction is stable across two survey waves, not a single-year artifact.

Workplace data shows the same tilt. Gallup’s 2024 State of the Global Workplace report found loneliness higher among employees under 35 than among older workers, holding work arrangement constant. Cigna’s 2020 loneliness index found 73% of workers aged 18 to 22 reporting loneliness, versus much lower rates among older cohorts, in a U.S. workforce sample.

Set against this is the older-adult literature, which is extensive and serious but answers a different question. The National Academies’ 2020 consensus report put social isolation, a structural rather than subjective measure, at roughly one quarter of U.S. adults 65 and older. The AARP Foundation’s 2018 survey, using the 20-item UCLA Loneliness Scale on 3,020 adults 45 and older, found one in three lonely. Neither of these numbers is directly comparable to the youth figures above, because they measure isolation and loneliness in a population aged 45 to 100-plus, not benchmarked against the young. When the comparison is made explicitly, as Gallup does, older adults come out lower on loneliness, not higher. That contradicts the folk model that loneliness is primarily an old-age affliction, and it is worth stating plainly because so much loneliness policy, including much social prescribing infrastructure, was built with older adults as the target population.

Explanation one: this is a cohort effect, not a life-stage effect

The most-discussed explanation, associated with Jonathan Haidt’s 2024 book, is that a specific generation experienced a discrete change in childhood and adolescence, the shift from play-based to phone-based socializing beginning in the early 2010s, and that this cohort carries elevated loneliness into adulthood rather than young people in general being lonelier at every point in history. The companion evidence review Haidt maintains with Jean Twenge compiles the supporting studies but also, usefully, includes dissenting work. Several researchers dispute the strength of the causal claim, arguing the correlational data is consistent with alternative explanations, including economic precarity, delayed family formation, and the general disruption of the 2008 recession hitting this cohort’s formative years. The debate is genuinely open in the literature; it is not resolved by citing one side.

A cohort account predicts something specific: today’s elevated youth loneliness should persist as this generation ages, rather than resolving itself once its members turn 40. That prediction cannot be tested with cross-sectional data of the kind Gallup and Harvard have collected. It requires the same people followed over decades, and that panel does not yet exist at the scale needed.

Explanation two: this is a life-stage effect that has always existed

An older and less dramatic account holds that early adulthood has always carried elevated loneliness, because it is the period with the least accumulated social capital. Robert Putnam’s account of civic decline was framed around institutions, not age, but the underlying logic applies: young adults have had less time to build the dense, overlapping networks that lower loneliness at midlife, and they are simultaneously undergoing the highest rate of life transitions, moving cities, changing jobs, leaving school, that break existing ties without yet having replaced them. On this account, the 2023 Gallup gradient is not new information; it reflects a durable feature of the life course that earlier surveys, run with less international coverage, simply did not document as clearly.

The American Enterprise Institute’s 2021 survey on American friendship is relevant here without settling the debate. It found 12% of Americans report no close friends, up from 3% in 1990, with the largest shift among men, whose share reporting six or more close friends fell from 55% to 27% over three decades. That is a trend across all ages, not a youth-specific one, which weakens a pure life-stage explanation and lends some support to a cohort or period effect operating across the whole friendship-formation system, not just among the young.

Explanation three: isolation and loneliness are being conflated across the age comparison

A study in Scientific Reports, published in December 2024, addressed the isolation-versus-loneliness distinction directly and found the relationship between the two varies by age. This matters for the whole comparison above. Older adults, especially those 65 and up, often have smaller networks by objective measures, fewer contacts, more likely to live alone, and yet report less subjective loneliness than young adults with larger networks. Cacioppo and Patrick’s 2008 account of loneliness as a biological signal offers one reading of why: loneliness tracks the gap between desired and actual connection, not the absolute size of a network. Older adults may have recalibrated their expectations downward, while young adults, surrounded by visible peer activity through social platforms and by cultural expectations of an active social life, experience a larger gap between what they have and what they believe they should have.

If that is right, the age gradient in loneliness is partly an artifact of expectation, not evidence that young people have objectively worse social lives than the isolated elderly. This is a case where the isolation/loneliness distinction the Center returns to repeatedly is not a footnote; it may be the central mechanism.

What would actually distinguish these accounts

None of the three explanations is disproven by current evidence, and they are not mutually exclusive. But they make different predictions, and a study designed to test those predictions would look different from what exists now. It would need to be longitudinal rather than cross-sectional, following the same cohort from adolescence into midlife with a consistent loneliness instrument, so that cohort and life-stage effects could be separated rather than inferred. It would need to measure isolation and loneliness as distinct constructs in the same sample, not rely on one survey using UCLA-scale loneliness and another using network-size isolation and treat the results as commensurable. And it would need international coverage held constant over time, since the Gallup 2023 data, valuable as it is, covers a single 8-month fielding window and cannot yet show whether the youth-elevated pattern is widening, narrowing, or stable.

Until such a study exists, the honest description of the evidence is that young adults report more loneliness than older adults across nearly every recent survey that asks the comparable question, and that at least three distinct mechanisms could produce that result, with different implications for what, if anything, would change it.