Evaluating the Multifactor Structure of the Long and Short Versions of the Self-Compassion Scale in a Clinical Sample

Abstract

Objective: There has been a growing interest in the concept of self-compassion in Eastern psychology. The aim of the present study was to explore the dimensionality of the widely used Self-Compassion Scale (SCS), in a clinical and non-clinical sample.Method: Several Confirmatory Factor Analysis (CFA) were computed in a mixed clinical (n = 316) and in a non-clinical sample (n = 1128) from the Portuguese population. Also, differences were tested between the groups in the SCS six factors.Results: The CFA supported both a six-factor model and a hierarchical model in both samples. Also, the SCS showed good psychometric properties, with good internal consistency, test-retest reliability and convergent validity. Our study further suggests that individuals with several psychopathological disorders showed significantly lower selfcompassionate abilities.Conclusions: The SCS is thus a reliable instrument to assess self-compassion, allowing a rich phenomenological analysis of this construct, which is useful for research and in particular for clinical practice.

Construction and Factorial Validation of the Chinese Version of the Self-Compassion Scale for Gifted Students

“…The assumption of normality of the items and the existence of outliers was assessed. When conducting a CFA, one should never be governed by the fit indices of the model alone (Castilho, Pinto-Gouveia, & Duarte, 2015). There are other factors to consider such as the factor loadings and the discriminant validity.…”

“…We calculated the SELF-COMPASSION SCALE FOR GIFTED STUDENTS 638 average variance extracted (AVE), or average shared variance, of each latent construct to assess convergent validity. Convergent validity is the extent to which indicators (items) of a specific construct share a high proportion of variance in common and is indicated by values of AVE 50 (Castilho et al, 2015). We compared the value of the AVE with the square multiple correlation between constructs to assess discriminant validity.…”

“…We compared the value of the AVE with the square multiple correlation between constructs to assess discriminant validity. To assume that all variables are orthogonal of one another, the value of AVE should be greater than the square multiple correlation between the respective variables (Castilho et al, 2015). A final step in CFA is to consider changes to a specific model that has poor fit indices, that is, model representation, and several different procedures can be used to assist in such modifications.…”

“…A final step in CFA is to consider changes to a specific model that has poor fit indices, that is, model representation, and several different procedures can be used to assist in such modifications. In the present study, we used the inclusion of additional parameters, specifically modification indices (MI; Castilho et al, 2015). The representation of the models was thus based on MI ( > 11; p < 0.001) and theoretical assumptions.…”

“…To assess scale reliability, we used both Cronbach's alpha and composite reliability, which provide a much less biased estimate of reliability than alpha and are more appropriate for multi-dimentional scales (Castilho et al, 2015). The statistical procedures were computed using PASW Statistics (version 18) and IBM…”