Which Young Adults the Loneliness Average Hides

Sixty-one percent of adults aged 18 to 25 reported serious loneliness in the Harvard Making Caring Common survey conducted in the early months of the pandemic, against 36% of Americans overall. That gap between a subgroup figure and the national average is the single most durable finding in the youth-loneliness literature. It is also the number most often repeated without asking what is inside it.

“Young adults” is not a homogeneous population, and the instruments used to measure loneliness in this group rarely break it down further than age band and, occasionally, sex. That leaves several sizeable subgroups compressed into a single headline figure that may not describe any of them well.

The gap by gender is wider than it looks

The Survey Center on American Life’s 2021 friendship study found that 15% of men reported having no close friends at all, a fivefold increase from 3% in 1990. Across all men, the share with six or more close friends fell from 55% in 1990 to 27%. These are lifetime trend figures rather than a young-adult-specific breakdown, but the trajectory matters for interpreting the current cohort: young men are entering adulthood with a thinner friendship baseline than young men thirty years ago, not because they were surveyed differently, but because the underlying network size appears to have shrunk.

This is worth separating from the loneliness question proper. A shrinking friend network is a structural, isolation-type measure. Whether a given young man reports feeling lonely is a separate, subjective question, and the two do not move in lockstep. A young adult with two close friends and a demanding job may report low loneliness; one with six casual acquaintances and no one to call at short notice may report high loneliness. National loneliness surveys that ask about subjective feeling state, without also asking about network size and composition, will not surface this distinction, and most of the surveys reviewed here ask about one or the other, not both.

Workers versus students versus neither

The Cigna loneliness index found in 2020 that 73% of workers aged 18 to 22 reported loneliness, and more than 80% of employed Gen Z workers overall did. That is a workplace-specific figure, drawn from employed respondents, and it sits well above the youth-loneliness figures drawn from general population samples. Gallup’s 2024 workplace survey adds a mechanism: fully remote employees reported loneliness at 25%, hybrid workers at 21%, and fully on-site workers at 16%, with loneliness elevated among employees under 35 specifically. Put those together and a plausible, though not proven, story emerges: young workers in remote or hybrid arrangements may carry a heavier loneliness burden than young workers on-site, and the aggregate “young adult” loneliness figure blends both groups along with young adults not in the labor force at all.

The college-specific literature complicates this further rather than resolving it. A 2025 study in Frontiers in Psychology examining college students after the acute pandemic period found that the kind of social connection mattered more than its volume in predicting wellbeing — a finding consistent with the broader isolation-versus-loneliness distinction, but one that is specific to a campus population with built-in structures for contact that a young remote worker or a young parent does not have. Comparing a college freshman’s loneliness score to a 24-year-old remote customer service worker’s score, then reporting both under “young adults 18-25,” discards exactly the context that would explain the difference.

Race and ethnicity are usually invisible in this figure

Few national loneliness surveys break results down by race and ethnicity at all. The KFF Survey of Racism, Discrimination and Health is one of the only ones that does, and it links experiences of discrimination directly to the size of a person’s social support network. Because most youth-loneliness reporting, including the widely cited Harvard figure, does not disaggregate this way, it is not possible to say from that data alone whether the 61% figure for 18-to-25-year-olds is evenly distributed across racial groups or concentrated in ways that discrimination and network-size findings would predict. The absence of that breakdown is itself worth naming, because it means a policy response built on the aggregate number could miss the subgroup most affected by it.

The global comparison cuts the other way

Gallup’s 142-country survey found that adults aged 19 to 29 report the highest loneliness rate of any age band worldwide, at 27%, against 17% among those 65 and older. That confirms the age pattern seen in the U.S. figures, but the level is far lower than the American numbers: 27% globally, versus 61% in the Harvard survey and 57% in Cigna’s 2025 report on American adults generally. Part of the gap is instrument mismatch — Gallup asked about feeling lonely “a lot of the previous day,” a narrower and more immediate question than the Harvard survey’s broader framing of “serious loneliness.” Part of it may be a genuine difference between the U.S. youth experience and the global one. The data available cannot cleanly separate those two explanations, and it would be a mistake to treat the 61% and 27% figures as measuring the same thing just because both mention young adults.

What the aggregate figure is for, and what it isn’t

None of this means the headline number is wrong. Sixty-one percent, or 57% in Cigna’s most recent report, is a legitimate description of a national sample at a point in time, and it is useful for tracking whether youth loneliness is rising or falling year over year. What it cannot do is tell a health system, a university, or an employer which young adults to reach, because it averages over living situation, employment status, remote-work arrangement, and network size — variables that the subgroup data suggest matter a great deal.

A study designed to fix this would need to ask about subjective loneliness and structural isolation in the same instrument, applied to the same respondents, broken out by employment status, living arrangement, and race, and repeated longitudinally rather than as a single cross-sectional snapshot. Several of the surveys cited here do one or two of these things. None does all of them at once, which is why the 61% figure keeps getting cited as if it described a single kind of person, when it describes several.