Abstract
Loneliness is a well-established risk factor for physical and mental health in later life. This work responds to recent calls to examine the dimensionality of the UCLA Loneliness Scale among older adults. In a preregistered secondary analysis using three measurement occasions from the publicly available National Social Life, Health, and Aging Project (NSHAP), we applied taxometric analyses (i.e., mean above minus below a cut, maximum eigenvalue, latent mode factor analysis, and comparison curve fit index) to examine the latent structure of the 3-item UCLA Loneliness Scale. Results supported the preregistered hypothesis, showing the mean comparison curve fit index (CCFI) was below 0.45 on each measurement occasion (CCFI range: 0.23 – 0.35). Therefore, loneliness, as measured in this sample, likely follows a continuous rather than categorical distribution, with no clear cut point. These findings inform statistical decision-making regarding loneliness classification and use of this scale in this sample.