The Instrument Built on Undergraduates Is Still Measuring Everyone

The AARP Foundation’s 2018 survey of adults 45 and older reported that one in three respondents was lonely. That figure is directly comparable to decades of academic research, for a specific and stated reason: the survey used the 20-item UCLA Loneliness Scale, the same instrument psychology has relied on since the late 1970s, rather than inventing its own question. Most of the loneliness statistics now circulating in policy documents cannot make that claim. The instrument that quietly underwrites much of the loneliness literature was built on a population — undergraduates — that no longer resembles the population policymakers are now most worried about, and the field has responded not by revalidating it but by walking away from it in pieces.

An instrument built for one age group, applied to all of them

The UCLA scale asks respondents to rate agreement with statements like “I have nobody to talk to” or “I feel left out,” summed into a continuous score. It treats loneliness as a stable trait-like disposition, measurable the same way in a 19-year-old and an 80-year-old, and assumes that a fixed set of items captures the experience of social deficit across radically different life stages, household structures, and technological environments. Cacioppo’s work extended the underlying model further, framing loneliness as an evolved biological alarm signal — akin to hunger — that motivates reconnection, which is a claim about mechanism, not measurement, but the two have become intertwined in how the scale gets used and cited.

The assumption that held up reasonably well for decades was that a college-student-derived instrument could travel. Holt-Lunstad’s 2015 meta-analysis, drawing on studies using varied measures including versions of the UCLA scale, found loneliness carried a 26% increase in odds of early mortality and social isolation a 29% increase, effects that persisted after adjusting for health status and were stronger in samples averaging under 65. That is evidence the underlying construct predicts something real, not evidence that the instrument travels cleanly across every age band it gets deployed on.

Where the field split from it

By the 2020s, most large surveys had stopped using the UCLA scale, for reasons that mostly had nothing to do with its assumptions about age. Cigna’s 2020 workplace report used its own loneliness index and found 61% of U.S. adults sometimes or always lonely, with 73% of workers aged 18 to 22 reporting loneliness. Harvard’s Making Caring Common survey found 36% of Americans seriously lonely and 61% of young adults aged 18 to 25, using its own measure. Gallup, surveying 142 countries with Meta, asked a single-item question — how much of the previous day the respondent felt lonely — and found 24% lonely worldwide, with adults 19 to 29 at 27%, the highest of any age band, and adults 65 and older at 17%, the lowest.

These are not competing estimates of the same underlying fact. They are different questions producing numbers that happen to cluster in a range readers assume is comparable. A 20-item trait scale summed into a continuous score, a single retrospective day-before item, and a bespoke workplace index are measuring related but distinct things: the UCLA scale asks about a standing disposition, the Gallup item asks about yesterday’s affect, and none of them was designed with a shared theory of what counts as “lonely enough” to cross a threshold. The UK’s 2018 loneliness strategy tried to address this at the country level by folding loneliness measurement into the Office for National Statistics rather than leaving it to one-off surveys, which improves consistency within Britain but does nothing to reconcile the UCLA-derived academic literature with the single-item surveys that dominate international comparison. The 2023 U.S. Surgeon General’s advisory, reporting that roughly half of American adults experience loneliness, drew on this patchwork rather than a single harmonized source, and said so.

A 2024 study in Scientific Reports examined the isolation-loneliness relationship directly by age band and found the two constructs relate to each other differently depending on age — evidence against the assumption, embedded in trait-scale designs, that a fixed relationship between structural isolation and subjective loneliness holds across the life course. A 2023 review in BMC Public Health went further, naming inconsistent measurement as one of the central barriers to comparing findings across the loneliness literature at all.

What this means for young-adult policy specifically

The instrument’s origin on college campuses should, in principle, make it well-suited to measuring loneliness in young adults now. It is not being used that way in most current young-adult research, which increasingly relies on single-item or study-specific measures instead — the same fragmentation affecting every other age group. That matters more for young adults than for older adults because the youth-loneliness debate is also a causal debate, and causal claims are especially sensitive to what the outcome measure actually captures.

Jonathan Haidt’s 2024 argument that a shift from play-based to phone-based childhood drove a rise in adolescent anxiety, depression, and loneliness beginning in the early 2010s rests partly on trend data assembled in the accompanying collaborative evidence review with Jean Twenge. That review is explicit that it includes dissenting findings, and the causal strength of the phone-based-childhood claim remains contested among researchers working with the same underlying data. Part of what makes the dispute hard to settle is that “loneliness” in these datasets is rarely the UCLA scale; it is whatever item happened to be on whatever survey collected in whatever year, which makes it difficult to know whether a rising line reflects a real shift in social experience or a shift in which instrument was fielded when.

What would settle more of this

A policy brief cannot resolve a measurement dispute, but it can say what would. The field needs either a shared short-form instrument validated across age bands and administered identically in the surveys that feed national policy, or — failing that — routine publication of item-level wording alongside every headline loneliness figure, so a reader can tell whether “36% lonely” and “24% lonely” are describing the same phenomenon or two different ones that happen to share a word. The UCLA scale’s origin on undergraduates was never disqualifying. What is disqualifying is a policy landscape that cites it, a single-item Gallup figure, and a bespoke corporate index in the same paragraph as though they are interchangeable evidence for the same underlying rate.