An exploratory psychometric network analysis of loneliness scales in a sample of older adults.

To examine the relationships within and between commonly used measures of loneliness to determine the suitability of the measures in older adults. Further, to determine whether certain items in these measures are more psychometrically robust in terms of capturing different types of loneliness across this population. Data were obtained from 350 older adults via completion of an online survey. Four measures of loneliness were completed. These were the University of California Los Angeles Loneliness Scale 4 (Version 3), the de Jong Gierveld Loneliness Scale, the Social and Emotional Loneliness Scale for Adults (Short Version) and a direct measure of loneliness. Analysis via a regularized partial correlation network and via clique percolation revealed that only the SELSA-S encompassed loneliness relating to deficits in social, family and romantic relationships. The remaining measures tapped mostly into social loneliness alone. The direct measure of loneliness had the strongest connection to the UCLA item-4 and the de Jong Gierveld item-1 exhibited the strongest bridge centrality, being a member of the most clusters. The results indicate that should researchers be interested in assessing loneliness resulting from specific relationships, then the SELSA-S would be the most suitable measure. Whereas the other measures are suitable for assessing loneliness more generally. The results further suggest that the de Jong Gierveld item-1 may be a more suitable direct measure of loneliness than that currently employed as it taps into a greater number of relationships.