Research · for research and professional bodies, older people's charities & befriending services

How Loneliness Is Measured, and Why Different Questions Count Different People

Published 2026-11-15

There is no single loneliness thermometer. Researchers ask about loneliness in different ways, and the same older people can be counted as lonely by one set of questions and not by another: in a US national survey of older adults, 57% of those who said they felt lonely when asked directly were not classed as lonely by a short scale that never uses the word[1].

Every loneliness figure in a funding bid, a newspaper or a government strategy rests on a few survey questions, a set of answer options and a decision about where to draw the line. Change any one and the figure moves. A measure that misses some people leaves their experience out of the case for services, and u3a groups, men's sheds, lunch clubs and befriending schemes are increasingly asked to prove their worth with these same numbers. This study explains how the main measures work and why they disagree. It describes the tools without reproducing them; nothing here can be used to score an individual.

57%of older adults in the US Health and Retirement Study (2002) who said they were lonely on a single direct question were not classed as lonely on a three-item scale[1]
43–47%of older people in a Swedish national study (aged 77 and over) whom a three-item scale counted as lonely were picked up by single direct questions[2]

Researchers measure three different things, and only one of them is loneliness

The first decision is what to measure. WHO's Commission on Social Connection, reporting in 2025, defines loneliness as “the painful feeling that arises from a gap between desired and actual social connections” and social isolation as “the objective lack of sufficient social connections”[3]. Loneliness can only be reported by the person who feels it; isolation can be counted from the outside. Living alone, often mixed in with both, is simply a household arrangement.

The two are related, but less closely than many people assume. When the authors of one of the most widely used short loneliness scales compared it with standard measures of objective isolation in two US population studies, they found the relationship “relatively modest”, and concluded that the quantity and quality of relationships are distinct[4].

Loneliness itself is measured in two ways. Indirect scales ask several questions about how people experience their relationships, without using the word “lonely”, and add up the answers. Direct questions simply ask how often a person feels lonely. As the sections below show, each counts somewhat different people.

The UCLA scales: a 20-item scale and a three-item version built for telephone surveys

The best-known family of loneliness scales comes from the University of California, Los Angeles. The full UCLA Loneliness Scale, now in its third version, has 20 items. Testing across samples of students, nurses, teachers and older people found it highly reliable, with internal consistency (coefficient alpha) ranging from .89 to .94 and a test–retest correlation of .73 over a year[5].

In 2004 Mary Elizabeth Hughes and colleagues published a short version built for telephone surveys. It has three items and “a simplified set of response categories”, and the authors found it “appears to measure overall loneliness quite well”[4]. Each item has three answer options, from least to most frequent, so total scores run from 3 to 9, higher meaning more frequent loneliness[6]. Its name causes confusion: in August 2025 the UK Office for National Statistics (ONS) noted that it had been “incorrectly referred to as ‘UCLA 3-item Loneliness Scale’” because, while derived from the longer UCLA scale, “it is not officially named as such”[6].

A score is not a yes-or-no answer, so studies that report a percentage “lonely” have to choose a cut-off. One English analysis of people aged 50 and over, for example, counted a score of 6 or more as lonely[7]. But the ONS guidance is blunt: “There is no standard accepted score for which a person would definitely be considered lonely.”[6] A prevalence figure from this scale is partly the researcher's choice.

The De Jong Gierveld scales separate missing a close bond from missing a wider circle

The second major family comes from the Netherlands. The De Jong Gierveld scale has 11 items and can be scored as a single loneliness measure, but its items were written with a distinction drawn by the researcher Robert Weiss in 1973 in mind: emotional loneliness, the absence of the close, intimate relationship a person wishes for, and social loneliness, having a wider circle of friends and acquaintances smaller than desired[8]. A shorter six-item version has three items for each kind. As with the UCLA scales, “the word loneliness has not been used in the set of items”[8].

The items have three response categories. Each answer is treated as lonely or not lonely, with the middle, hedged answer always counted as lonely, so each three-item subscale runs from 0, not lonely, to 3, intensely lonely[8]. That choice is one reason its figures are not interchangeable with those from other measures.

Its authors tested the six-item version in the United Nations Generations and Gender Surveys in France, Germany, the Netherlands, Russia, Bulgaria, Georgia and Japan, with samples of between 8,158 and 12,828 adults aged 18 to 79 per country, and found reliable and valid scales for emotional and social loneliness in each[8]. Scores differed markedly between countries (see loneliness across countries).

Asking “how often do you feel lonely?” counts fewer people, and different ones

The direct question is short and simple, but the word itself carries stigma for some: the ONS notes that “some people might be reluctant to admit to loneliness” and that “this might be particularly true of certain groups such as older men”[6].

Sharon Shiovitz-Ezra and Liat Ayalon compared a single direct item, taken from a depression symptom scale, with the three-item scale in about 2,000 observations from the 2002 US Health and Retirement Study. Even at the cut-off that best matched the two, 57% of respondents who reported being lonely on the direct item were classified as not lonely on the scale, and the two measures gave inconsistent answers about how loneliness related to age and education[1].

A Swedish national study of 1,599 people aged 77 and over, published in 2026, compared two single questions, on feeling lonely and on being bothered by loneliness, with two versions of the three-item scale. About 70% were placed at low levels of loneliness by the single items, against about 60% by the scale; measured against the scale, the single items picked up only 43–47% of people with loneliness. People answering “hardly ever” lonely had significantly poorer outcomes than those answering “never”. The authors concluded that single- and multi-item measures “are not equivalent and are optimally used in combination”[2].

The difference is not only one of size. In Norway’s Life Course, Generation and Gender study of 14,743 people aged 18 to 81, a direct question and the six-item De Jong Gierveld scale gave a similar overall prevalence but “attribute loneliness to somewhat different people”: women were more often lonely on the direct question, men on the scale, and the link between age and loneliness differed between the two[9]. Who looks lonelier can depend on the question asked (see loneliness across the life course).

Wording, question order, interview mode and language all move the figures

A 2024 systematic review and meta-analysis pooled 101 prevalence rates from 45 studies of community-dwelling people aged over 60 in many countries. Its overall pooled estimate was 31.6%, but the measure used was a statistically significant moderator: prevalence was lowest for single-item questions and highest for the 20-item UCLA scale. How the data were collected mattered too, with prevalence highest in face-to-face interviews and lowest in telephone and computer-assisted telephone interviews[10]. That pooled figure averages very different measures (the hub’s prevalence study sets out the main national estimates, measure by measure).

Question order counts too. The ONS asks for the three-item scale before the direct question, because the scale is “designed to measure loneliness without mentioning the word” and its own testing showed that asking the direct question first “produces different results”[6].

Translation adds another layer. When researchers tested the Spanish version of the six-item Lubben Social Network Scale (described below) with 56 older people in Los Angeles and Mexico City, participants understood family, friends, neighbourhood and “private matters” in ways that led the authors to conclude that the Spanish version “needs to be adapted” for Mexican and Mexican-American older adults[11].

Social isolation is measured by counting ties, and the counts depend on the yardstick too

Measures of social isolation ask about facts rather than feelings: how often people are in contact with family and friends, and whether they take part in groups and civic life. An early example is the index of “social and community ties” that Lisa Berkman and Leonard Syme applied to a 1965 survey of 6,928 adults in Alameda County, California. Over nine years, the most isolated had age-adjusted relative risks of death of 2.3 for men and 2.8 for women compared with the most connected. The link held after accounting for self-reported health, socioeconomic status and health practices: an observational association, not proof of cause[12].

For work with older people, the Lubben Social Network Scale is among the most used. Its abbreviated six-item version (LSNS-6) asks about family and friends, giving a total and two subscales, and its authors proposed cut-points based on whether people have back-up ties to turn to. Tested with older people living at home in three European cities, it classed a fifth of the Hamburg sample as at risk of social isolation, against smaller shares in London and Solothurn[13].

Share of older people living at home classed as at risk of social isolation on the six-item Lubben Social Network Scale, three European samplesLubben and colleagues 2006, community-dwelling older people, abbreviated Lubben Social Network Scale (LSNS-6) with the authors' proposed cut-points: Hamburg 20%; London 15%; Solothurn, Switzerland 11% classed as at risk for social isolation.0%5%10%15%20%25%Hamburg20%London15%Solothurn (Switzerland)11%
Social isolation, not loneliness: the scale counts family and friendship ties, and the cut-points mark people with few back-up ties. Samples of community-dwelling older people in each city; the shares describe these study samples, not whole countries. Source: Lubben et al., The Gerontologist, 2006.

Ageing cohorts often build their own index. The English Longitudinal Study of Ageing assessed isolation in 6,500 people aged 52 and over in 2004–2005 from contact with family and friends and participation in civic organisations, with loneliness measured separately. After adjusting for demographic factors and baseline health, the most isolated fifth had a hazard ratio for death of 1.26, while loneliness was not independently associated with mortality[14]. The lesson for measurement (the mortality study weighs the health findings) is that isolation and loneliness can behave differently in the same data.

Services are asked to prove an effect with tools that were not built for it

The UK's ONS guidance on national indicators, last revised in December 2018, recommends four questions, made up of the three-item scale and the direct question, for people aged 16 and over, with an adapted version for children aged 10 to 15[6]. The direct question has five answer options, from most frequent to never.

Community organisations face a harder task. A 2023 study for the UK Department for Culture, Media and Sport, by the Campaign to End Loneliness, surveyed professionals who organise or deliver loneliness interventions. Of the 48 who answered, 19 (40%) said one or more loneliness measures had been used to evaluate their work; ten of those used a different measure, “mainly unstandardised”[15]. And 29 of 48 respondents (60%) said they had no specific funding for evaluation. Some interviewees reported that the three-item scale “had been validated as a screening measure but was not intended to measure intervention impact” and was not sensitive to change. The report also noted that “many services do not reach enough people” to detect an effect, and that loneliness questions could leave some people feeling stigmatised or distressed[15].

Which loneliness measures were used to evaluate loneliness services: respondents to a 2023 UK government-commissioned surveyMacIntyre and Hewings for DCMS, 2023. Of 48 survey respondents who answered, 19 said one or more loneliness measures had been used. Number naming each measure: ONS direct measure 5; UCLA-3 measure 2; another UCLA measure 1; Campaign to End Loneliness measure 2; a different, mainly unstandardised measure 10.024681012A different measure,mainly unstandardised10ONS direct question5Three-item scale (UCLA-3)2Campaign to End Lonelinessmeasure2Another UCLA measure1
Number of respondents naming each measure, among the 19 of 48 respondents (40%) who said a loneliness measure had been used; the counts add up to 20, so at least one named more than one. A small, self-selected online survey of professionals running loneliness interventions, not a census of services. Source: MacIntyre and Hewings for the Department for Culture, Media and Sport, 2023.

Campaign to End Loneliness guidance for practitioners says measuring impact “is not easy”, warns that people “might find the questions intrusive or distressing”, and advises explaining why questions are asked, letting people stop at any point and having support ready afterwards. It describes asking at the start and end of a service, or at a first meeting and then every two or three months[16].

That design has a trap. A service that takes people because they are lonely can see average scores fall even with no effect, because people are often recruited at a low point (regression to the mean). An uncontrolled before-and-after drop is not evidence that a scheme worked, and a flat result from a handful of participants is not evidence that it failed.

What the evidence asks of us

Name the measure every time. A loneliness figure without its question, answer options and cut-off cannot be compared with anything; anyone quoting one should say which measure, which answers counted, and who was asked and how. Low-level loneliness matters too: the Swedish study found poorer outcomes even among people answering “hardly ever”[2].

Use more than one lens. Scales and direct questions are “not equivalent” and work best together[2], and loneliness and isolation need measuring separately[4][14]. Surveys that include only the direct question risk undercounting people, such as some men, who may be reluctant to use the word[6].

Build better shared tools. WHO’s Commission on Social Connection made “improved measurement (including developing a global Social Connection Index)” one of five areas of its 2025 roadmap[3].

Ask small services for what they can reasonably give. The UK study for government recommended funding to cover evaluation costs and evaluation co-produced with delivery organisations; people it consulted asked funders not to request data that will not be used and to value qualitative evidence[15]. A men’s shed or lunch club cannot run a controlled trial; pooled, well-designed evaluations across many groups can. The people who come through the door deserve to be asked with care, and counted honestly.

Language: this piece says "older people" and treats loneliness as a common human experience that anyone can have, not a personal failing or an illness. It keeps apart three things that are often mixed up: loneliness (how people feel about their relationships), social isolation (how few social contacts they have) and living alone (a household arrangement). Where a study uses another term, such as "seniors" or "older adults", we keep it when quoting that study.

Why we're publishing this

Loneliness figures are used to argue for funding, to set targets and to judge whether services work. The groups that bring older people together are on the receiving end of all three, often without anyone explaining where the numbers come from. Knowing how a figure was produced lets them read claims critically, question a target built on the wrong measure, and make their case in terms that hold up. This study is part of the hub’s Loneliness in Later Life research library.

How to read this data

The percentages here belong to particular studies and samples, not whole countries. The comparisons of direct questions and scales come from single surveys in the US, Norway and Sweden and one meta-analysis; they show that measures disagree, not by how much everywhere. Most surveys cited sample people living in private households, so people in care homes and those who cannot answer for themselves are under-represented. The health associations mentioned are observational, not cause and effect. The UK survey of people running loneliness services was a small online survey (49 respondents) and describes their experience, not all services. We deliberately do not reproduce scale items or scoring rules: the scales were built for research on groups, not for judging one person’s feelings.

Use this data

Free to cite with attribution to Understanding Aging. u3a groups, men's sheds, senior centres, lunch clubs, befriending services, charities, campaigners and journalists are welcome to reuse these figures and charts. For the underlying figures as a spreadsheet, get in touch.

Suggested citation: Understanding Aging (2026). How Loneliness Is Measured, and Why Different Questions Count Different People. https://understandingaging.org/research/measuring-loneliness

Important: This article summarises published research and statistics for general information. It is not medical or care advice and cannot be used to diagnose, treat or predict anything about an individual. If loneliness, low mood or isolation is affecting your health or someone else’s, a doctor or another health professional can help, and local services for older people can often put you in touch with groups and support nearby.

Sources

  1. Shiovitz-Ezra S, Ayalon L. Use of direct versus indirect approaches to measure loneliness in later life. Research on Aging 34(5):572–591, 2012. cris.huji.ac.il/en/publications/use-of-direct-versus-indirect-approaches-to-measure-loneliness-in
  2. Dahlberg L, McKee KJ, Naseer M, Lennartsson C, Agahi N. How shall we measure loneliness among older adults? Comparison of direct single-item and indirect multi-item measures of loneliness in a Swedish national study. Aging & Mental Health, 2026. europepmc.org/article/MED/42954353
  3. World Health Organization. Social connection linked to improved health and reduced risk of early death (news release on the report of the WHO Commission on Social Connection). WHO, 2025. who.int/news/item/30-06-2025-social-connection-linked-to-improved-heath-and-reduced-risk-of-early-death
  4. Hughes ME, Waite LJ, Hawkley LC, Cacioppo JT. A short scale for measuring loneliness in large surveys: results from two population-based studies. Research on Aging, 2004. europepmc.org/article/MED/18504506
  5. Russell DW. UCLA Loneliness Scale (Version 3): reliability, validity, and factor structure. Journal of Personality Assessment, 1996. europepmc.org/article/MED/8576833
  6. Office for National Statistics. Measuring loneliness: guidance for use of the national indicators on surveys. ONS, 2018 (notice added 2025). ons.gov.uk/peoplepopulationandcommunity/wellbeing/methodologies/measuringlonelinessguidanceforuseofthenationalindicatorsonsurveys
  7. Victor CR, Pikhartova J. Lonely places or lonely people? Investigating the relationship between loneliness and place of residence. BMC Public Health, 2020. europepmc.org/article/PMC/PMC7251825
  8. De Jong Gierveld J, Van Tilburg T. The De Jong Gierveld short scales for emotional and social loneliness: tested on data from 7 countries in the UN generations and gender surveys. European Journal of Ageing, 2010. europepmc.org/article/PMC/PMC2921057
  9. Nicolaisen M, Thorsen K. Who are lonely? Loneliness in different age groups (18–81 years old), using two measures of loneliness. International Journal of Aging & Human Development, 2014. europepmc.org/article/MED/25265679
  10. Stegen H, Duppen D, Savieri P, et al. Loneliness prevalence of community-dwelling older adults and the impact of the mode of measurement, data collection, and country: a systematic review and meta-analysis. International Psychogeriatrics, 2024. europepmc.org/article/MED/38525677
  11. Vilar-Compte M, Vargas-Bustamante A, Lubben J. Validation study of the abbreviated version of the Lubben Social Network Scale Spanish translation among Mexican and Mexican-American older adults. Journal of Cross-Cultural Gerontology, 2018. europepmc.org/article/MED/29340902
  12. Berkman LF, Syme SL. Social networks, host resistance, and mortality: a nine-year follow-up study of Alameda County residents. American Journal of Epidemiology, 1979. europepmc.org/article/MED/425958
  13. Lubben J, Blozik E, Gillmann G, et al. Performance of an abbreviated version of the Lubben Social Network Scale among three European community-dwelling older adult populations. The Gerontologist, 2006. europepmc.org/article/MED/16921004
  14. Steptoe A, Shankar A, Demakakos P, Wardle J. Social isolation, loneliness, and all-cause mortality in older men and women. Proceedings of the National Academy of Sciences, 2013. europepmc.org/article/MED/23530191
  15. MacIntyre H, Hewings R (Campaign to End Loneliness). Evaluation of interventions to tackle loneliness. Department for Culture, Media and Sport, 2023. gov.uk/government/publications/exploring-interventions-to-tackle-loneliness/evaluation-of-interventions-to-tackle-loneliness
  16. Campaign to End Loneliness. Using a loneliness scale: benefits, drawbacks and good practice. Campaign to End Loneliness, undated. campaigntoendloneliness.org/?p=5224