Psychometric validation of the brief loneliness scale: a two-factor model in university students

Studies on loneliness among university students have intensified significantly in recent years. In the contemporary context of accelerated digital transformation, the university experience is increasingly mediated by artificial intelligence systems, introducing new relational and psychosocial dynamics that are reshaping the experience of loneliness. However, in Latin America, there is still a shortage of brief instruments with comprehensive psychometric validation. Importantly, this study advances the field by integrating two major psychometric traditions in loneliness assessment: the unidimensional UCLA framework and the bidimensional approach proposed by De Jong Gierveld, combining parsimony with a more differentiated conceptualization of the construct. The study aimed to psychometrically validate an abbreviated version of the Loneliness Scale among university students, examining its factor structure, reliability, and convergent, discriminant, and concurrent validity, as well as its invariance by gender. A quantitative approach was adopted with a robust sample of university students ( n = 2,025) from a public university in Peru. In addition to the loneliness scale, other scales were applied to verify results and validity. The other constructs evaluated were resilience, perceived stress, self-esteem, academic self-efficacy, and well-being. A clear two-factor structure (emotional loneliness and social loneliness) without cross-loadings was confirmed, with excellent fit (CFI = .968; RMSEA = .030). Overall reliability was high ( Ω = .908). Adequate convergent and discriminant validity were evident, along with consistent associations with self-esteem, well-being, self-efficacy, and stress. The scale showed factorial invariance by gender, allowing for valid comparisons between men and women. The ROC curve analysis showed that the area under the curve (AUC = .998) indicates an almost perfect discriminative ability to classify low and high levels of loneliness in the sample, further supported by substantial agreement (K = .784). The final scale consisted of 10 items assessing structural soundness, classificatory precision, and intergroup stability, constituting a psychometrically robust and efficient tool for detecting and intervening in loneliness in university settings. This study contributes to the theoretical and methodological advancement of standardized measurement in higher education contexts.