Evaluating the Factor Structure and Measurement Properties of a Scale to Assess Loneliness Among Older Adults

BACKGROUND. Loneliness is a depressed feeling related to perceived social isolation and a lack of meaningful companionship. It is often an element of impaired quality of life, suffering, and feeling distressed. Older adults may be especially vulnerable to loneliness. For example, older adults more often experience the death of friends and family, and may have greater difficulties mobilizing for social activities due to functional limitations. Because loneliness is an abstract concept, it is important that measures of loneliness are reliable, valid, and easily administered. Indeed, researchers have created several instruments with which to measure loneliness. The Revised University of California Los Angeles (R-UCLA) scale is one promising measure of loneliness. A brief 11-item version of the R-UCLA has been incorporated into the Health and Retirement Study (HRS). To date, no studies have examined the factor structure of the 11-item scale. It is critical to understand the instrument’s factor structure because assessing scale dimensionality is a key component of validity testing.

PURPOSE. To evaluate the measurement properties of the 11-item R-UCLA scale; and, in particular, to examine factors of loneliness through an exploratory factor analysis (EFA) and confirmatory factor analysis (CFA).

METHODS. Data were drawn from the 2008 HRS and its Psychosocial Supplement. The HRS is a nationally representative study of the health and economic status of older adults 50+ in the U.S. The final analytic sample included 4,347 older adults. The HRS used the 11-item R-UCLA (3-point Likert scale; Cronbach’s a=.88) for measuring older adults’ loneliness (e.g., How much of the time do you feel left out?). The sample was randomly split into two sub-samples before analysis. Demographic differences between sub-samples were examined. The EFA was conducted to identify the factor structure of the loneliness scale using the first sub-sample. The CFA was conducted using the second sub-sample.

RESULTS. Measurement properties are as follows: sub-sample 1 (a=.87; M=16.4; SD=4.5; range=22), and sub-sample 2 (a=.88; M=16.6; SD=4.7; range=22). Results of EFA identified two models, a two-factor model and a three-factor model. Factor 1 indicated feeling isolated, Factor 2 indicated social connectedness, and Factor 3 labeled sense of belonging. According to the EFA, the three-factor model demonstrated the best fit to the data. However, the results of CFA showed that the two-factor model (feeling isolated and social connectedness) was the most appropriate to describe the factor structure of the 11-item loneliness scale (χ²=650.63, p<.001; RMSEA=.08; CFI=.98; and TLI=.97). Furthermore, in the three-factor model, there was a very strong correlation (r=.85, p<.001) between social connectedness and sense of belonging, suggesting conceptual redundancy.

IMPLICATIONS. This study contributes to future research by examining the factor structure and instrument properties of the R-UCLA loneliness scale using a large population-based dataset. Furthermore, results show that two factors (feeling isolated and social connectedness) contribute to older adults’ loneliness. Thus, based on results, we should consider these factors in efforts to reduce older adults’ loneliness in social work practice.