UCLA Loneliness Scale (Version 3):
Scoring and Interpretation

The UCLA Loneliness Scale (Version 3) is a 20-item self-report assessment of perceived loneliness and social isolation, defined as a subjective shortfall in the quality or quantity of desired interpersonal connections. Loneliness is distinct from objective social-network size: a person may have many contacts yet still perceive inadequate relationships.

Russell (1996) evaluated the revised wording and psychometric performance across college students, nurses, teachers and older adults, and the scale remains widely cited in social, health and aging research. The Version 3 scoring key must not be confused with the original 1978 form or with the widely used three-item survey loneliness index.

• Items: twenty frequency-rated responses, coded from 1 to 4.
• Reverse-coded positions: in the standard Version 3 order, items 1, 5, 6, 9, 10, 15, 16, 19 and 20 (nine items). Item 4 is scored in the loneliness direction and is not reversed.
• Reversal rule: the reversed score for a standard 1 to 4 item equals 5 minus the response.
• Total: the twenty keyed values are summed for a single score from 20 to 80, and higher scores indicate more loneliness.
• Subscales: none in the primary total-score interpretation. Wording-related factors are investigated psychometrically.
• Versions: do not adopt cut-offs or item numbering from a different UCLA version. The revision history runs from the original 1978 form and the Revised UCLA (1980) to Version 3 (1996), alongside multiple short forms.

1. Verify the standard 20-item UCLA Version 3 order and that the numerical responses run from 1 to 4.
2. Reverse positions 1, 5, 6, 9, 10, 15, 16, 19 and 20 as 5 minus the raw response.
3. Add all 20 keyed items to produce ucla_total, which ranges from 20 to 80.
4. The code uses complete-case scoring; the cited original abstract does not establish a universal missing-item prorating rule. Document a prespecified policy for incomplete responses.

* Version 3 order: reverse items 1, 5, 6, 9, 10, 15, 16, 19, 20.
RECODE ucla1 (1=4)(2=3)(3=2)(4=1) INTO ucla1_r.
RECODE ucla5 (1=4)(2=3)(3=2)(4=1) INTO ucla5_r.
RECODE ucla6 (1=4)(2=3)(3=2)(4=1) INTO ucla6_r.
RECODE ucla9 (1=4)(2=3)(3=2)(4=1) INTO ucla9_r.
RECODE ucla10 (1=4)(2=3)(3=2)(4=1) INTO ucla10_r.
RECODE ucla15 (1=4)(2=3)(3=2)(4=1) INTO ucla15_r.
RECODE ucla16 (1=4)(2=3)(3=2)(4=1) INTO ucla16_r.
RECODE ucla19 (1=4)(2=3)(3=2)(4=1) INTO ucla19_r.
RECODE ucla20 (1=4)(2=3)(3=2)(4=1) INTO ucla20_r.
COMPUTE ucla_total=SUM.20(ucla1_r,ucla2,ucla3,ucla4,ucla5_r,ucla6_r,ucla7,ucla8,ucla9_r,ucla10_r,ucla11,ucla12,ucla13,ucla14,ucla15_r,ucla16_r,ucla17,ucla18,ucla19_r,ucla20_r).
EXECUTE.

items <- paste0("ucla",1:20)
x <- data[items]
reverse <- c(1,5,6,9,10,15,16,19,20)  # Version 3 order
x[reverse] <- lapply(x[reverse], function(z) ifelse(is.na(z),NA,5-z))
data$ucla_total <- rowSums(x,na.rm=FALSE)

Higher values reflect more frequently endorsed loneliness after reverse coding the positive social-connection items. Do not present popular unvalidated groupings as universal clinical cut-offs. A descriptive distribution and sample-specific comparisons are generally preferable.

Treat the UCLA-3 total primarily as a continuous outcome in observational studies of social determinants, depressive symptoms, quality of life or perceived support, for example in a Multiple Linear Regression with covariates. The distribution can be skewed, so inspect floor and ceiling frequencies and model residuals, and consider rank-based methods such as Correlation Analysis (Pearson & Spearman) or appropriate generalized models where warranted.

Start reliability with alpha from the keyed items through Reliability Analysis (Cronbach’s Alpha), and preferably add omega. For new language versions, check the factor model and differential wording effects with Confirmatory Factor Analysis (CFA) rather than assuming a simple one-factor model will fit, and test comparability across groups or languages through Measurement Invariance. A universal minimal clinically important difference was not established in the original 1996 publication.