One in Three, One in Six, One in Four: Why Loneliness Estimates for Older Adults Never Match

Ask what share of older adults is lonely and the answer depends almost entirely on who is asking and how. AARP’s 2018 survey of adults 45 and older, using the 20-item UCLA Loneliness Scale, found one in three lonely. Gallup’s 2023 world poll, using a single self-rating question across 142 countries, found adults 65 and older to be the least lonely age group at 17%. The World Health Organization’s 2025 Commission on Social Connection report puts the global figure affected by loneliness at one in six. None of these numbers is wrong. They are measuring different things, in different populations, with different instruments, and the gap between them is the story.

The instrument problem, concretely

The UCLA Loneliness Scale asks respondents to rate agreement with statements like “I feel left out” and “I lack companionship,” and sums them into a continuous score, which researchers then bucket into categories. AARP’s Foundation surveys, in both 2018 and the 2025 follow-up, used this instrument consistently, which is what makes the two AARP reports directly comparable to one another and to much of the academic literature that also relies on UCLA-derived measures.

Gallup, by contrast, asks a single direct question — something closer to “did you experience loneliness yesterday” — across a much larger and more diverse set of countries, translated into dozens of languages. A single-item self-report and a 20-item validated scale will not produce the same prevalence even when applied to identical populations. The single item tends to run lower, partly because a person can answer “no” to a direct question about feeling lonely while still scoring in the lonely range on a scale that asks about companionship, social connectedness, and feeling understood without ever using the word “lonely.” This is not a flaw in either approach. It is a difference in what is being asked.

The National Academies’ 2020 consensus report on older adults sidesteps the loneliness-measurement problem partly by focusing on isolation instead, estimating that roughly one quarter of U.S. adults 65 and older are socially isolated — a structural count of network size and contact frequency rather than a subjective rating. That 25% figure is not comparable to AARP’s one-in-three loneliness figure or Gallup’s 17%, because isolation and loneliness are not the same construct. A person can have a small network and not feel lonely; a person embedded in a large network can feel intensely alone. Recent work on socially isolated older adults specifically has tried to identify which isolated individuals go on to become lonely, treating the two as related but distinct outcomes rather than interchangeable labels for the same problem.

What US federal surveillance adds, and complicates

Two federal-adjacent U.S. data sources, published years apart from their underlying fieldwork, add further texture. The CDC’s Morbidity and Mortality Weekly Report, published in June 2024 but drawing on 2022 survey data, estimated loneliness and lack of social and emotional support among U.S. adults using its own item set embedded in existing federal surveillance infrastructure. The 2025 study using 2022 HINTS-6 data found roughly one in seven U.S. adults reporting severe loneliness and one in five reporting moderate loneliness, for a combined moderate-to-severe rate above 37%. The 2026 analysis of the 2024 Household Pulse Survey breaks prevalence down by demographic and socioeconomic factors using yet another instrument embedded in a different federal vehicle. Three federal-adjacent sources, three different survey mechanisms, three different numbers — collected in different years and published even later, which means a reader citing “the CDC figure” or “the government figure” on loneliness needs to specify which survey, which item, and which year of fieldwork, because there is no single authoritative federal number.

Japan and the UK: measuring isolation as policy infrastructure

Two governments have gone further than surveying loneliness episodically and have tried to build it into permanent measurement systems, which changes the comparison problem again. The UK’s 2018 loneliness strategy embedded loneliness measurement into the Office for National Statistics, meaning the UK now tracks loneliness the way it tracks unemployment or inflation — as a standing national indicator rather than a one-off survey finding. Japan’s JACSIS study followed cohorts longitudinally through 2020 and 2021, which is a structurally different kind of evidence than any single cross-sectional snapshot: it can show change within the same people over time, rather than comparing different samples at different moments. Neither the UK’s ONS-embedded measure nor Japan’s longitudinal panel is directly comparable to Gallup’s cross-national single item or AARP’s UCLA-scale surveys, but each is more useful for its own purpose. The UK can track whether loneliness is rising or falling nationally over years. Japan can show whether the same individuals became more isolated or lonelier across the pandemic period specifically. Gallup can compare countries against one another using a common (if blunt) yardstick. AARP can compare 2018 to 2025 using the same instrument on a comparable age group. None of the four can do what the others do.

The comparability problem is now a recognized research gap

A 2023 review in BMC Public Health mapping the state of loneliness and isolation research names inconsistent measurement as a central barrier to comparing findings across studies — not a minor caveat but one of the field’s defining limitations. That review’s conclusion matters for anyone using these numbers in a policy document or a news story: a claim that loneliness among older adults is “rising” or “falling” globally requires specifying which instrument, in which country, over which years, because the four approaches above would not necessarily agree even if applied to the same population in the same year.

What would settle this

The single most useful thing the field could produce is not a bigger survey but a validation study: administer the UCLA scale, a Gallup-style single item, and a structural isolation count to the same sample of older adults in several countries simultaneously, then examine how they diverge and by how much. Absent that, any cross-country comparison of “loneliness among older adults” is really a comparison of survey instruments wearing loneliness as a label. The WHO Commission’s global estimate of one in six affected by loneliness is a defensible summary figure for advocacy purposes. It is not a number that can be reconciled, item for item, with AARP’s one in three or Gallup’s 17% among the over-65s. Reconciling them was never really the point of any of the three surveys.