When Two Surveys of Young Adult Loneliness Differ by Twelve Points

Two figures circulate in almost every policy discussion of young adult loneliness in the United States, and they do not agree.

Cigna’s Loneliness and the Workplace report, published in January 2020, put loneliness among workers aged 18 to 22 at 73%, and among employed Gen Z respondents at over 80%. Its headline figure for U.S. adults overall was 61% — sometimes or always feeling lonely — which the report described as a seven-point rise year over year.

The Harvard Graduate School of Education’s Making Caring Common project, publishing in February 2021, reported serious loneliness among 61% of young adults aged 18 to 25, against 36% of Americans overall.

For young adults the two estimates are twelve points apart. For the adult population as a whole they are twenty-five points apart. And the direction of the discrepancy is the opposite of what most people assume: the lower general-population figure comes from the survey fielded during a pandemic, and the higher one from a survey fielded before it. Any account that treats these as successive measurements of a worsening trend has the arithmetic backwards.

The gap is mostly in the question, not the population

Cigna’s threshold, as reported, is respondents who feel lonely “sometimes or always.” Harvard’s category is “serious loneliness.” Those are not the same construct and were never intended to be. A threshold that admits “sometimes” will capture a large share of any population, because occasional loneliness is close to universal; a threshold labelled “serious” is by construction a smaller and more clinically meaningful group.

This is the single most common failure in reading loneliness prevalence figures across sources. The number depends less on how lonely a population is than on where the analyst drew the line. A survey reporting 61% of all adults lonely and a survey reporting 36% of all adults seriously lonely are not in conflict; they are answering different questions and both may be right.

The useful calibration point sits between them. The AARP Foundation’s 2018 national survey of 3,020 adults aged 45 and older found roughly one in three reporting loneliness — and, unusually among advocacy-sponsored surveys, it used the 20-item UCLA Loneliness Scale rather than a bespoke set of items. That choice is what makes the AARP estimate comparable to the academic literature. It is also, notably, a survey of midlife and older adults, so it cannot be used to bound the young adult figure directly. What it does show is that a validated multi-item instrument applied to a national sample tends to produce something in the region of a third, not two-thirds.

Who was actually asked

The Cigna youth figures are drawn from employed respondents. The 73% applies to workers aged 18 to 22, and the 80%-plus figure to employed Gen Z. Harvard’s 61% applies to adults aged 18 to 25 regardless of employment status.

That distinction matters in both directions and it is not obvious which way it cuts. A workforce sample excludes full-time students, who make up a large share of 18-to-22-year-olds and whose social circumstances are structurally different. It also excludes the unemployed and those not seeking work, a group with elevated risk on almost every social measure. The net effect on a loneliness estimate is indeterminate, which is precisely the problem: the two samples are not nested, so the difference cannot be decomposed.

The age bands also differ. Eighteen to 22 and 18 to 25 overlap but are not the same cohort, and the three additional years span the transition out of full-time education for most respondents.

Publication date is not collection date

Cigna’s report appeared on 23 January 2020, which means the fieldwork predates the pandemic entirely. Harvard’s report appeared in February 2021 and describes data collected in late 2020. The pandemic sits between them.

This produces the inversion noted above, and it should discipline how the two are used. The Cigna figures describe the pre-pandemic baseline for employed young adults. The Harvard figures describe a mid-pandemic snapshot of all young adults on a stricter threshold. Neither is a follow-up to the other, and no trend line can be drawn between them. Harvard’s own within-survey trend measure is the more defensible one: 43% of young adults reported that their loneliness had increased since the pandemic began. That is a retrospective self-assessment rather than a repeated measurement, and it carries the recall problems that come with it, but at least it holds the instrument constant.

What neither survey measures

Both figures are subjective loneliness. Neither is a measure of social isolation, which is a structural property of a network — how many people someone is in contact with, how often, and through what ties. The two are independently predictive of health outcomes and are conflated constantly, including in the framing of young adult loneliness as an emergency.

The structural picture for young adults is thinner than the subjective one. The Survey Center on American Life’s 2021 friendship survey found that 12% of Americans report having no close friends, up from 3% in 1990, and that the share of men with at least six close friends fell from 55% in 1990 to 27% in 2021. Those are network measures, and the historical comparison is what makes them valuable. They are not, however, reported as young-adult-specific figures, which leaves a real gap: there is no widely cited structural isolation estimate for 18-to-25-year-olds comparable to the National Academies’ 2020 finding that roughly a quarter of adults aged 65 and older are socially isolated.

The asymmetry is institutional as much as empirical. The most authoritative document in the field — a National Academies consensus report, with the review process that implies — addresses older adults. The clinician-facing commentary that followed it in the American Journal of Geriatric Psychiatry likewise argues for routine assessment in geriatric settings and sets out what that would require in practice. Young adult prevalence, by contrast, rests substantially on a survey commissioned by a health insurer and a survey conducted by a school of education. Both are serious pieces of work. Neither has passed through anything like a consensus process, and neither was designed to support the population-level screening decisions it is now cited in support of.

Why the twelve points matter for policy

Prevalence figures do specific work in policy. They set the denominator for cost estimates, they justify universal versus targeted programmes, and they determine whether a screening instrument is worth deploying at scale.

A population in which 73% of young workers are lonely implies universal intervention, because targeting is pointless when nearly everyone qualifies. A population in which 36% of adults and 61% of young adults report serious loneliness implies a targeted approach with a defensible case-finding threshold. These are different budgets and different delivery models. Choosing between them on the basis of whichever figure was published most recently is not a method.

The health argument for taking either seriously does not depend on resolving the gap. Holt-Lunstad’s 2015 meta-analysis put the odds ratio for early mortality at 1.29 for social isolation, 1.26 for loneliness and 1.32 for living alone, with effects persisting after adjustment for health status — and, importantly for this topic, social deficits were more predictive of death in samples averaging under 65 than in older samples. Her 2021 review in the American Journal of Lifestyle Medicine extends the argument to prevention, positioning social connection as a modifiable protective factor belonging alongside diet, exercise and smoking. Murthy’s 2020 book made the parallel case for treating loneliness as a public health question rather than a private difficulty.

What none of that settles is how many young adults need what.

The measurement fix already exists

The UK’s 2018 loneliness strategy, the first national strategy of its kind, did something that no U.S. equivalent has done: it embedded loneliness measurement into the Office for National Statistics. That means a standing instrument, a fixed threshold, and repeated fielding — which is what makes a trend statement possible. Japan created a cabinet post for loneliness and isolation in February 2021, and the two governments held a first bilateral ministerial meeting on loneliness policy that June, signalling that this has become a formal cross-government area rather than a research interest.

The intervention evidence has not kept pace. Two 2021 systematic reviews of social prescribing — the main delivery mechanism funded under the UK strategy — found positive individual impacts across all nine studies in one review, with three reporting reduced use of GP, emergency, social worker or inpatient services, and improvements in self-esteem and self-confidence in the other. Both flagged limited trial evidence and substantial heterogeneity between programmes. Neither was specific to young adults.

A better evidence base for this cohort would look like the following. One instrument, applied repeatedly, with the threshold stated in the headline rather than the appendix. A sample not restricted by employment status, reported in age bands that match other national surveys. Structural network measures collected alongside subjective loneliness, so that isolation and loneliness can be separated in the same respondents. And collection dates in the headline, not the publication date.

Until then, the honest statement is that between a third and two-thirds of American young adults report loneliness depending on where the line is drawn, and that the field has not agreed where to draw it.