The Dimensionality of Self-Compassion: A Psychometric Evaluation of the Self-Compassion Scale–Youth in Portuguese Adolescents

Abstract Objectives This study sought to contribute to an ongoing debate over the dimensionality of the Self-Compassion Scales. Specifically, we aimed to (1) test the factorial structure of the Portuguese-language Self-Compassion Scale-Youth (SCS-Y); (2) evaluate whether total scores are sufficiently unidimensional to be interpretable; and (3) assess construct validity. Method The sample comprised 1467 Portuguese 8th graders ( M age = 14.1 years, 48.8% girls). We tested the factorial structure of the SCS-Y using exploratory structural equation modeling (ESEM) and confirmatory factor analysis (CFA). The network alignment between self-compassion subcomponents was tested with exploratory graph analysis (EGA). Scale unidimensionality was evaluated using bifactor indices. Construct validity was evaluated by estimating correlations between SCS-Y scales and indicators of subjective well-being (SWB). Results A correlated six-factor CFA model and a two-bifactor CFA model fit the data well. Bifactor indices indicated that scores tapping into uncompassionate self-responding (i.e., self-criticism) were essentially unidimensional (ECV = 0.63, ω H = 0.78). In contrast, there was greater multidimensionality in compassionate self-responding (i.e., self-compassion; ECV = 0.42, ω H = 0.54), with the mindfulness and common humanity subscales explaining a sizeable amount of variance. The EGA revealed a two-dimensional structure (self-compassion and self-criticism communities). As evidence of construct validity, self-compassion was correlated with positive affect ( r = 0.33) and self-criticism with negative affect ( r = 0.43). Conclusions The Portuguese SCS-Y is a psychometrically adequate tool for assessing the six theorized subcomponents of self-compassion. While self-criticism is a well-defined unidimensional construct, caution is needed when interpreting a total score of compassionate self-responding. Preregistration This study was not preregistered.