One Instrument, Four Numbers: What the UCLA Loneliness Scale Does and Does Not Measure in Young Adults

Four recent sources report loneliness prevalence among young adults, and no two agree. Cigna’s 2020 workplace survey found 73% of workers aged 18-22 reporting loneliness. Harvard’s Making Caring Common project put serious loneliness among 18-25-year-olds at 61% in its 2021 report. The Surgeon General’s 2023 advisory cites roughly half of all U.S. adults as lonely, without breaking the youngest cohort out at a comparable figure. AARP’s 2018 national survey, which did not focus on young adults but used the same instrument across its full 45-plus sample, found one in three lonely overall.

These numbers are not measuring the same thing, even though every report uses the word “loneliness.” The difference is not primarily about when each survey was fielded or how bad things had gotten. It is about which instrument each survey used, and what that instrument assumes a lonely person looks like.

The instrument that anchors the academic literature

The 20-item UCLA Loneliness Scale, developed in the late 1970s and revised since, is the instrument most consistently used in the peer-reviewed literature on loneliness and health, including much of the mortality research synthesized in Julianne Holt-Lunstad’s 2015 meta-analysis. AARP’s 2018 survey of adults 45 and older used it directly, which is worth noting because most large public surveys do not. The scale asks respondents to rate agreement with statements such as feeling isolated from others or lacking people to turn to, summed into a continuous score rather than a single yes-or-no answer. It treats loneliness as a disposition that can be measured at any point in time using stable, general statements, not as something tied to a specific week or event.

That design choice is deliberate and defensible. John Cacioppo’s 2008 account of loneliness as an evolved biological signal, comparable to hunger or thirst, depends on treating loneliness as a state with some persistence, not a mood that flickers with a bad Tuesday. A scale built from durable trait-like statements fits that theoretical picture. It also means the UCLA instrument tends to produce more conservative prevalence estimates than instruments built around a single blunt question, because agreement with several abstract statements is a higher bar than agreeing “yes” to one.

What the bespoke surveys ask instead

Cigna’s loneliness index, which produced the 73% figure for 18-22-year-old workers, is not the UCLA scale. It is a proprietary instrument built for a corporate wellness report, designed to be short enough to embed in an employee survey and to generate a headline number. Harvard’s Making Caring Common survey, which produced 61% for 18-25-year-olds reporting “serious loneliness,” used its own bespoke items as well, asking directly about frequency and intensity in ways calibrated to the pandemic period it was fielded in, rather than to decades of psychometric validation against mortality or health outcomes.

Neither approach is illegitimate. A short, direct question is easier for a respondent to answer honestly and quickly, and both surveys were transparent about their methods. But a bespoke instrument built to capture the acute, self-described experience of an isolated pandemic year will pick up more people than a scale built to detect a stable underlying disposition. The two are not interchangeable, and treating 73% and 61% and “one in three” as three data points on the same trend line is a mistake the underlying reports do not make but that summaries of them often do.

The Surgeon General’s 2023 advisory sidesteps some of this by reporting the general finding that roughly half of U.S. adults experience loneliness, framed as a synthesis across the literature rather than a single new survey figure, and by comparing the mortality risk of disconnection to smoking up to 15 cigarettes a day, a comparison drawn from the broader body of epidemiological work rather than from any one youth-specific instrument. That framing is more defensible precisely because it does not try to reconcile incompatible bespoke numbers into a false single trend.

Why this particular gap matters for young adults

Older adults are disproportionately studied with the UCLA scale or close variants, in large part because gerontological and health-outcomes research has used it for decades and funders in that space know it. Young adults, by contrast, are disproportionately studied through newer, bespoke, often commercially commissioned instruments, built quickly to answer a specific question, such as “how lonely are our youngest employees” or “how has the pandemic affected college-age adults.” That is not a coincidence of research funding; it means the instrument gap tracks the age gap. When a report says loneliness is worse among 18-25-year-olds than among older adults, some portion of that gap may be a real generational and life-stage difference. Some portion may simply be that the instruments used on young adults are structurally more likely to produce high numbers than the instrument used on the older comparison group.

The BMC Public Health review of the field, published in June 2023, names inconsistent measurement as one of the central barriers to comparing findings across the loneliness literature, and this is exactly the kind of case it has in mind. A reader comparing Cigna’s 73% to AARP’s 33% is not comparing young people to older people. The reader is comparing a proprietary corporate wellness index to a 43-year-old validated psychometric scale, applied to different age groups, in different years, for different institutional purposes.

What a cleaner comparison would require

The corrective is not complicated in principle, even if no one has done it yet at scale. A study designed to compare loneliness by age cohort would need to administer the same validated instrument, ideally the UCLA scale or another instrument with an established track record against health outcomes, to a single sample stratified across age groups, in the same survey wave, with the same reference period. That would isolate age as a variable instead of confounding it with instrument choice. Longitudinal administration of the same instrument to the same cohort as it ages, something closer to what the JACSIS study attempted for isolation in Japan over 2020-2021, would go further still, distinguishing a cohort effect specific to people who came of age during a particular period from a life-stage effect that any 20-year-old would experience regardless of when they turned 20.

Until that exists, the honest reading of the current evidence is narrower than most summaries suggest. Young adults report loneliness at high rates on nearly every instrument that has asked them, and the AARP finding that having spoken with a neighbor roughly halves the likelihood of loneliness at all ages suggests the underlying phenomenon responds to similar structural conditions regardless of age group. But the specific claim that young adults are lonelier than older adults, at a specific numerical gap, is not something the current mix of instruments can support. The instruments were not built to answer that question. They were built to answer other questions, and the age comparison is a byproduct that the data cannot cleanly bear.