Evaluating Loneliness Measurements across the European Union

Loneliness has been associated with several detrimental effects for individuals and societies, making it a priority for monitoring across the European Union. While many loneliness measures currently exist, notable gaps exist regarding knowledge of their psychometric structure, reliability, comparability, and validity, particularly as it pertains to their suitability for EU-wide population surveys. Relying on data from the EU Loneliness Survey covering the 27 EU member states (N=25,646), we examined the factor structure, internal consistency, measurement invariance, and construct validity of the six-item De Jong Gierveld Loneliness Scale (DJGLS-6), the three-item UCLA Loneliness Scale (T-ILS), and a single-item measure of loneliness. Our analyses followed a two-step process, with exploratory analyses conducted in an exploratory fold, followed by pre-registered confirmatory analyses in a confirmatory fold. The DJGLS-6 showed a sufficient fit to a two-factor structure in 11 countries, sufficient internal consistency across all 27 countries, scalar invariance for a cluster of five countries, metric invariance for a cluster of eight countries, no invariance for a cluster of 17 countries, and sufficient construct validity for 25 countries. The T-ILS showed a sufficient fit to a one-factor structure in 23 countries, sufficient internal consistency across all 27 countries, scalar invariance across all 27 countries, and sufficient construct validity for 21 countries. Finally, the single-item measure of loneliness demonstrated sufficient construct validity in 19 countries. Overall, these findings suggest that the T-ILS may currently be the most suitable measure for monitoring loneliness across the European Union.