57%, 36%, One in Three: The Same Question, Answered Differently

The Cigna Group’s Loneliness in America 2025, released June 1, reports that 57% of American adults are lonely. Fieldwork ran from May 29 to June 14, 2024, across more than 7,500 U.S. adults, so the report is a year old by the time the number reaches print — a gap worth naming, though not the interesting one. The interesting gap is with everyone else’s number.

Harvard’s Making Caring Common project put U.S. serious loneliness at 36% in 2021. The AARP Foundation, surveying adults 45 and older in 2018, found one in three. The Surgeon General’s 2023 advisory cited “approximately half.” Gallup’s 142-country survey with Meta found 72% of people worldwide felt socially connected, which is not quite the inverse of loneliness but sits in obvious tension with a 57% domestic loneliness figure. None of these studies is wrong. They are measuring different things and calling them the same word.

What each instrument actually asks

The AARP survey used the 20-item UCLA Loneliness Scale, the closest thing the field has to a gold-standard instrument, which is why its one-in-three figure is directly comparable to the academic literature that also uses it. Harvard’s Making Caring Common survey asked about “serious loneliness” using its own items, oriented toward the pandemic disruption of 2020-21 rather than a validated multi-item scale. The CDC’s 2024 surveillance report, drawing on 2022 data, used yet another set of survey questions embedded in federal health monitoring rather than a clinical loneliness instrument. Cigna’s own methodology has shifted across its report series; the 2025 figure of 57% is not a simple continuation of the company’s earlier loneliness index, and comparing it year-over-year within the Cigna series is safer than comparing it across organizations.

This is not a minor technical quibble. A 2023 review in BMC Public Health singled out inconsistent measurement as one of the field’s central barriers to comparison, and the Cigna-to-Harvard-to-AARP spread from 33% to 57% illustrates exactly why. Ask “do you feel isolated from others” and ask “do you often feel that no one truly understands you” and ask a full UCLA battery covering both, and three different numbers come back from populations that may feel identically lonely. The instrument is doing as much work as the underlying reality.

A genuine international comparison

Gallup’s cross-national data offers a cleaner test, because at least the same question was asked the same way in 142 countries between June 2022 and February 2023, of roughly 1,000 people aged 15 and older per country. That survey found 24% of people worldwide reporting loneliness “very” or “fairly” often, with young adults 19-29 at 27% and adults 65 and older at just 17% — the reverse of the age pattern many assume. Gallup’s companion analysis explicitly cautions that connectedness and loneliness are not simple inverses; a country can score high on felt social connection and still carry meaningful loneliness in a demographic subgroup.

Set the Gallup figure of 24% worldwide against Cigna’s 57% for the United States, and the U.S. appears to be an outlier of striking severity, roughly double the rate. Some of that gap may be real: American social infrastructure, workplace culture, and household structure differ from the international average in ways that plausibly matter. But some of it is almost certainly the instrument. Gallup asked about frequency of loneliness in the recent past. Cigna’s survey design is not the same question, administered to a different population, on a different timeline, using different response categories. Reporting the two side by side as though they describe the same thing is the single most common error in loneliness journalism, and it happens because both numbers are true and neither is comparable to the other.

Why this keeps happening

Three things are being conflated across these reports, routinely and without acknowledgment. First, isolation and loneliness are different constructs — one structural, one subjective — and some surveys ask about the felt experience while others measure network size or contact frequency and report the result as “loneliness” regardless. Second, “serious” or “chronic” loneliness is not the same threshold as “sometimes” or “occasionally” lonely, and surveys collapse these differently into headline figures. Third, publication date and fieldwork date routinely diverge; the CDC’s 2024 report describes 2022 conditions, and Cigna’s 2025 report describes mid-2024 conditions. A reader comparing “the CDC number” to “the Cigna number” as though both describe June 2025 is comparing two different years to begin with, before the instrument problem even enters.

None of this means the underlying phenomenon is invented. Every major U.S. survey of the past decade — AARP’s one in three, Harvard’s 36%, the Surgeon General’s roughly half, Cigna’s 57% — describes loneliness as common enough to be a population-level concern rather than a marginal condition, and that convergence on “common” is itself meaningful even where the exact figures diverge sharply.

What would resolve the discrepancy is not another survey with another headline number, but a single validated instrument — ideally the UCLA scale or an equivalent — fielded on the same timeline across multiple countries with comparable sampling, repeated at fixed intervals. The AARP survey and the Gallup multinational work each get partway there within their own scope. No source in the current literature does both scale validation and true cross-national comparability at once, and until one does, headline percentages should be read as describing the instrument as much as the population.