Assessing self-compassion in older adults: Factorial structure of the self-compassion scale and invariance across sex at birth

Abstract

Objectives Self-compassion is a valuable resource for positive ageing and should be measured in a reliable and valid manner. However, findings regarding the factor structure of the Self-Compassion Scale (SCS) have been inconsistent and have been particularly lacking in older adults. Hence, the present study intended to investigate the factor structure, internal consistency, and measurement invariance across sex at birth of the SCS in adults ≥ 65 years old.

Method The present study used a sample of 418 community-dwelling, Portuguese older adults aged 65–94 years (M = 74.45, SD = 6.94; 59.3% female). Confirmatory factor analyses were conducted to test six different measurement models, and reliability and multigroup analyses were performed for the best fitting model.

Results Although all models initially showed poor adjustment, the correlated 6-factor model had the comparatively better fit. After modifications, this model achieved an acceptable fit (RMSEA = 0.060 and SRMR = 0.062). Internal consistency of this model was adequate (Cronbach's alpha values ranging from 0.67 to 0.86) and strong invariance across sex at birth was demonstrated (i.e., configural, metric, and scalar models showed non-significant Δχ2).

Conclusions Current findings suggest that assessing the self-compassion components is advised in older adults, as this 6-component model was reliable and provided an acceptable fit for both male and female older adults. This 6-component model is a parsimonious, theoretically sound, and statistically valid option to assess self-compassion in this population. It is, however, not an ideal solution, and an acceptable fit was only achieved after modifications in the initial 6-component model.