UCLA Loneliness Scale: 20 questions, one number
Loneliness is not the same thing as being alone. The UCLA Loneliness Scale was built on that distinction in 1978, and it asks how often you feel out of step with the people around you rather than how many of them there are.
Russell, Peplau and Ferguson wrote the first version. Russell, Peplau and Cutrona revised it in 1980, adding ten positively worded items so the response set would not drift in one direction. Russell's 1996 third version put "How often do you feel" in front of every item and simplified three statements that older respondents found hard to parse. That third version is the standard one, and the one used here.
Answering and scoring
Twenty items, four options each: never, rarely, sometimes, always, scored 1 to 4. Nine items are reverse-scored (1, 5, 6, 9, 10, 15, 16, 19, 20), so the total runs from 20 to 80. Higher means lonelier. There are no subscales. That keeps it simple, and it is also the most criticised thing about the scale: it reports one overall degree and cannot separate social loneliness from emotional loneliness.
| Total | Band |
|---|---|
| 20-34 | Low |
| 35-49 | Moderate |
| 50-64 | Moderately high |
| 65-80 | High |
These are descriptive ranges used in the literature. The scale has never had a clinical cutoff, and it was not designed to produce a diagnosis.
Reliability
Internal consistency was .96 for the 1978 original, .94 for the 1980 revision, and .89 to .94 for the third version. Retest was .73 over one year for the third version and .73 over two months for the original. Moderate stability is what you want here. Loneliness responds to circumstance, and a scale that ignored that would be measuring something else.
A confirmatory factor analysis of the third version fits a global bipolar loneliness factor plus two method factors for the positively and negatively worded items. Method factors are a measurement artefact, and their presence is a reminder that the wording itself carries part of the variance.
Shorter forms exist for surveys where twenty items is too many: an 8-item version with alpha between .74 and .90, a 6-item version, and a 3-item version Hughes and colleagues wrote in 2004 for telephone interviews.
What the numbers look like in different groups
Group averages vary enough that a score means more next to the right reference point. In a meta-analysis of cancer patients, the weighted mean was 38.26 (95% CI 35.51 to 41.00). It was 34.35 (33.60 to 35.17) within a year of diagnosis and 37.38 (35.38 to 39.38) beyond it. Married patients averaged 33.48 (31.36 to 35.60). Unmarried patients averaged 37.82 (35.48 to 40.17).
A study of Chinese nursing home residents found 38.6 (SD 10.4) among frail inpatients. A second study in the same population split at 35.2 (SD 9.1) without suicidal ideation against 51.4 (SD 13.1) with it. That gap is large, and it says something about what the scale is sensitive to rather than about what loneliness causes.
A 1992 estimate put this scale in 80% of empirical loneliness studies at the time. Nothing has displaced it since.
Reading your own score
A high total is not a diagnosis of anything. It is a report of how often you felt a certain way, and it reads better next to your circumstances than against them. If the number is high and has stayed high, or if loneliness is affecting sleep, work or health, that is a reason to talk to someone qualified. This page is not a diagnosis and does not replace one.