The Age Pattern That Doesn't Fit the Loneliness Narrative

Public health communication about aging has settled on a fairly consistent image: older adults living alone, disconnected from family, at risk of the kind of isolation that shortens life. That image is not wrong, exactly. But the survey data on who actually reports feeling lonely does not support the version of it that treats age as a straightforward risk gradient, with loneliness rising as people grow older.

Gallup’s 2023 survey of 142 countries, conducted with Meta between June 2022 and February 2023, found that adults 65 and older reported the lowest rate of loneliness of any age group measured: 17% said they felt very or fairly lonely, against 27% among adults aged 19 to 29. The overall global figure was 24%. Younger adults were the loneliest cohort, not the oldest. This is not a fluke of one survey. Harvard’s Making Caring Common project found, in a 2021 survey fielded during the pandemic, that 61% of young adults aged 18 to 25 reported serious loneliness, compared with much lower rates among older respondents. The pattern recurs across instruments that ask about subjective loneliness directly.

This sits awkwardly next to the National Academies’ 2020 consensus report, which put the share of U.S. adults 65 and older who are socially isolated at roughly one in four. That is a large number, and it is the number most often cited in coverage of aging and connection. So which is it: are older adults doing relatively well, or badly?

Two different questions, two different answers

The resolution is that isolation and loneliness are not the same measurement, and a body of research this size has been careful enough to say so even when public discussion has not. Isolation describes an objective property of a person’s social network: how many contacts they have, how often they see them, whether they live alone. Loneliness describes a subjective state, the felt gap between the connection a person has and the connection they want. The two are correlated but distinct, and they behave differently across the life course.

Structurally, older adults are more isolated on average. Retirement removes a daily source of contact. Widowhood and the death of peers shrink a social network in a way that is largely irreversible within a single life stage. Mobility limitations and hearing loss make maintaining existing relationships harder, not just forming new ones. The National Academies report treats this as a health-system problem precisely because it is measurable and durable: a smaller network is a smaller network regardless of how the person inside it feels about it.

Subjectively, though, older adults appear to adapt. Longstanding psychological literature on socioemotional selectivity — not covered in the sources this article draws on, but consistent with the pattern here — holds that people become more selective about relationships as they age, investing in fewer, closer ties rather than maintaining a large network. A smaller network chosen deliberately does not necessarily produce the felt deficit that defines loneliness. A large network lost involuntarily, or a life stage where a person expects more connection than they have, does.

Young adults may be the loneliest group for something like the opposite reason. AARP’s 2018 survey of adults 45 and older, using the 20-item UCLA Loneliness Scale rather than a bespoke instrument, found that the strongest predictors of loneliness were the size and diversity of a person’s social network and physical isolation — not age itself. Younger adults are frequently in periods of high transition: leaving school, changing cities, entering and leaving relationships, all of which disrupt networks faster than they can be rebuilt. That instability, rather than any deficit specific to youth, may be doing the work.

What this does not mean

None of this means older adults are fine, or that policy attention to isolation in later life is misplaced. The mortality evidence does not split neatly along the same line as the loneliness surveys. Holt-Lunstad’s 2015 meta-analysis in Perspectives on Psychological Science found that social isolation, loneliness, and living alone were all independently associated with roughly 26% to 32% higher odds of early death, and — notably — that these effects were often stronger in samples averaging under 65 than in older samples. That finding, on its own, complicates any tidy story that youth is protected and old age is at risk, or the reverse. The relationship between social deficit and mortality does not track cleanly with the relationship between social deficit and reported loneliness, because they are measuring different mechanisms operating over different time horizons.

The CDC’s 2024 surveillance report, drawing on 2022 data, is explicit that loneliness and lack of social and emotional support are linked to heart disease, stroke, dementia, type 2 diabetes, depression, and anxiety across the adult population, not specifically concentrated at either end of the age range. The U.S. Surgeon General’s 2023 advisory put the share of American adults reporting loneliness at roughly half, a figure drawn from multiple surveys with varying instruments and therefore not perfectly comparable to Gallup’s cross-national figure, but broadly consistent with loneliness being a majority-population problem rather than one narrowly located in later life.

What the age pattern in the loneliness data does mean is that isolation and loneliness cannot be used as proxies for one another in program design or public communication, especially for older adults. A city that measures success in eldercare by asking people whether they feel lonely may report improvement even while the underlying network — number of contacts, frequency of in-person meeting, presence of someone to call in an emergency — continues to thin. Conversely, a program that expands an older adult’s objective network without addressing whether the relationships in it are wanted or valued may show no change in loneliness scores at all, and get counted as a failure.

Gallup’s own analysis, published in November 2023, made a version of this point directly: connectedness and loneliness are not simple inverses of one another. A country, or an age group, can score relatively well on one and poorly on the other. Treating them as a single “social health” number obscures exactly the distinction that matters most for figuring out what to do about either one.

What would clarify this

The clearest gap in the current evidence is longitudinal tracking of the same individuals across both isolation and loneliness measures as they age, rather than repeated cross-sectional snapshots of different cohorts at different ages. Almost everything cited above, including the Gallup figures showing older adults as the least lonely age group, is cross-sectional: it compares different people of different ages at one point in time, not the same people followed as they grow older. It is possible that today’s 70-year-olds are less lonely than today’s 25-year-olds for reasons specific to their cohort — having come of age before the mobile phone, for instance — rather than for reasons intrinsic to being 70. A cohort effect and an age effect would produce an identical cross-sectional pattern and require different policy responses. Only panel data that follows the same people over decades can distinguish the two, and that data does not yet exist at the scale needed to settle the question.