Loneliness Test — The Real Scales and What They Measure

If you are reading this because things feel unbearable right now, please talk to a person rather than a questionnaire. In the United States, call or text 988 (Suicide & Crisis Lifeline). In the UK and Ireland, call Samaritans free on 116 123. In Australia, call Lifeline on 13 11 14. Anywhere else, findahelpline.com lists verified services by country. They are free, and you do not have to be in crisis to call.

A loneliness score cannot tell you anything a conversation with a real person cannot tell you better.

Loneliness is subjective. It is the distress of a gap between the connection you have and the connection you want. Because it is a feeling, the only way to measure it is to ask — which is why every loneliness test is a questionnaire, and why the questions chosen determine the answer you get.

Four instruments do most of the work in the research literature. They differ in length, in what they separate out, and in how their results are read:

• The UCLA Loneliness Scale, Version 3 — 20 items, the reference instrument in the field.
• The three-item short form — built so loneliness could be measured in large population surveys and telephone interviews.
• The De Jong Gierveld scale — six items, and the only one of the four that separates emotional loneliness from social loneliness.
• The ONS/GSS harmonised measure — four questions, used across UK government statistics.

None of them is a diagnostic test. Loneliness is not a disorder in any diagnostic manual, and no score on any of these scales is a clinical finding. What they do well is give a consistent number you can compare across people, across countries, or across time in one person's life.

The UCLA scale is the instrument most other work is measured against. Version 3 was published by Daniel Russell in the Journal of Personality Assessment in 1996, refining earlier versions from 1978 and 1980.

How it works. Twenty statements about companionship, belonging and being understood, each rated from 1 (never) to 4 (often). The total runs from 20 to 80, and a higher score means more loneliness.

How well it holds up. Russell reported internal consistency (coefficient alpha) between 0.89 and 0.94 across samples, and a test-retest correlation of 0.73 over a full year — unusually good stability for a mood-adjacent measure.

The thing people get wrong. The UCLA scale has no official cut-off. There is no score at which you become lonely. Studies choose a threshold to suit their design: Cigna's 2018 US Loneliness Index, for instance, surveyed 20,096 adults with the 20-item scale and classified everyone scoring 43 or above as lonely. The national average came out at 44 — which is why that study is reported as finding roughly half of Americans lonely. Move the cut-off two points and the headline moves by millions of people. The number is real; so is the arbitrariness of the line.

Mary Elizabeth Hughes, Linda Waite, Louise Hawkley and John Cacioppo published a three-item version in Research on Aging in 2004, designed for large population studies where twenty questions is twenty too many.

How it works. Three questions — how often you feel you lack companionship, how often you feel left out, and how often you feel isolated from others — each answered hardly ever, some of the time, or often, scored 1 to 3. The total runs from 3 to 9.

How it is read. Many surveys treat a total of 6 or more as "lonely". That is a convention adopted for convenience, not a validated clinical boundary, and different studies use it differently.

This short form is the reason so much national data on loneliness exists at all. It fits into a phone interview or a general health survey without displacing anything else, which is how loneliness ended up being tracked at population scale in the first place.

Jenny de Jong Gierveld's scale, published in a validated six-item form with Theo van Tilburg in Research on Aging in 2006 and tested on data from 9,448 respondents, does something the UCLA instruments do not: it splits loneliness into two components.

• Emotional loneliness — the absence of one close, intimate attachment. Three items measure it.
• Social loneliness — the absence of a wider network of people around you. Three items measure that.

The distinction is not academic. Someone recently widowed after fifty years of marriage may have a full diary, many friends and no confidant; someone who has just moved cities may have a devoted partner and know nobody else. Both can produce the same overall loneliness score for entirely opposite reasons, and what would help each of them is different. A single total hides that; this scale does not.

If you have seen a British loneliness statistic, it almost certainly came from this measure. The UK Government Statistical Service harmonised measure has four questions: one direct — "How often do you feel lonely?" — plus three indirect items on lacking companionship, feeling left out and feeling isolated from others, which are the same three used in the UCLA short form.

Published UK headline figures use the direct question at the strictest cut. In the Community Life Survey for 2023/24, covering October 2023 to March 2024, 7% of adults in England — around 3.1 million people — said they felt lonely often or always. The 2024/25 survey found the same 7%.

That 7% is often set against American figures of 33% or 50% as if one country were seven times lonelier than another. It is not. The English figure counts only "often or always" on one direct question; the US figures come from a two-item measure at a looser threshold and from a 20-item scale with a chosen cut-off. The instrument explains almost all of the gap. We set out the full comparison, with every instrument and threshold named, in our loneliness statistics for 2026.

It is not a diagnosis. Loneliness is not a mental disorder and does not appear as one in the diagnostic manuals. No score on any scale is a clinical finding. What is diagnosable, and treatable, is depression — which overlaps with loneliness, is not the same thing, and is worth raising with a doctor if the feeling is persistent, or if it comes with hopelessness, changes to sleep or appetite, or loss of interest in things you used to enjoy.

Thresholds are choices. Every cut-off you will read about was chosen by researchers for a particular study population. Scoring one point below a line does not mean you are fine; one point above does not mean you are ill.

It is a snapshot. These instruments measure how you feel now, or over a recent window. Loneliness after a bereavement, a move or a break-up is an ordinary human response to a real event and often eases. Loneliness that has persisted for months behaves differently, and is the kind the health research is mostly about.

It says nothing about your worth or your character. Loneliness is common — a quarter of the world's adults reported feeling very or fairly lonely in the Meta-Gallup survey — and it is a signal about circumstances, not a verdict on the person.

These get used interchangeably and should not be.

Loneliness is a feeling: the gap between the connection you have and the connection you want. It is measured by asking.

Social isolation is objective: few contacts, little interaction. It is measured by counting. People can be isolated and perfectly content.

Living alone is a household arrangement, nothing more. Most people who live alone are not lonely, and being alone is not a problem to be solved.

The research keeps them separate because the useful response differs. Isolation responds to more contact. Loneliness responds to contact that means something — more of the wrong kind can make it worse. Living alone raises one narrow practical question that has nothing to do with either feeling: if something went wrong today, how long before anyone knew? For more on the health research behind these distinctions, see the health effects of isolation and what the mortality research actually shows.

Nothing on this page, and nothing on any website, substitutes for the two things the evidence keeps pointing at: contact that means something, and professional help when the weight of it becomes too much.

• Talk to your doctor. Persistent loneliness often travels with depression, anxiety and poor sleep, all of which respond to treatment.
• Pick one small, repeatable point of contact rather than a resolution to be more social. A standing weekly call beats an ambitious plan that collapses.
• Do something alongside other people — a class, a volunteering shift, a walking group. Shared activity generates conversation more reliably than events designed to generate conversation.
• Use a helpline when you need one. They are not only for emergencies: 988 in the US, Samaritans on 116 123 in the UK and Ireland, Lifeline on 13 11 14 in Australia, and findahelpline.com elsewhere.

If you are asking these questions on behalf of an older parent rather than yourself, our elderly loneliness assessment covers what to look for and what to say.

We publish a loneliness self-check. It draws on the themes of the UCLA scale, but it is an informal awareness tool, not a validated instrument: it has not been psychometrically tested, it has no research cut-off behind it, and its score carries no clinical meaning. If you want a validated measure, the UCLA scale and the De Jong Gierveld scale above are the ones the literature uses.

We also make a daily check-in app, and it is worth being exact about what it has to do with loneliness: almost nothing. It is not company. It is not care. It does not treat loneliness and we will not claim it does. It answers one narrow question for people who live alone — if something goes wrong today, how long before someone knows? You confirm you are okay once a day; if you don't, the people you chose are told, with your location.

The daily check-in is free (Try It, $0) on iPhone and Android, anywhere in the world. Automatic alerts to your contacts start on Protect Me at $29.99 a year, with location on the move on Protect Me On The Move at $39.99 a year, and there is a one-time Stay Connected lifetime upgrade at $4.99. Check-first phone fall detection is iPhone only, on Protect Me and above, stays off until you switch it on, and only runs while the app is open; it alerts your own contacts, never calls emergency services, and is best-effort rather than a medical device. The app interface is currently English only.

That is a safety mechanism. It is worth having for the same reason a smoke alarm is worth having, and for no larger reason than that.