Structural Validation of the Self-Compassion Scale with Adolescents and Young People in Central Uganda

Background: Addressing mental health risks is part of a multi-factored approach to addressing HIV vulnerabilities among youth in sub-Saharan Africa (SSA). Evidence shows that self-compassion (SC) is an important youth mental health protective factor. SC consists of 3 main elements: self-kindness, common humanity, and mindfulness that collectively create a self-compassionate frame of mind (Neff, 2003). Research suggests that SC is crucial for promoting positive youth development. Yet, limited research has explored SC among youth in SSA. There is a critical need to validate the Self-Compassion Scale (SCS) in SSA to inform intervention development aimed at enhancing positive youth development and reducing HIV vulnerability.

Methods: We conducted a multi-site cross-sectional survey with a probability sample of high school students (N=406) aged 15 to 24 years in central Uganda. Participants completed sociodemographic surveys, and all six components of the SCS. The SCS is a 26-item measure of overall self-compassion, suitable for youth ages 14 and up. The SCS, using a 5-point Likert scale, has demonstrated strong convergent and discriminant validity, good test-retest reliability, and excellent reliability (Cronbach’s alpha = .93) (Neff, Rude, & Kirkpatrick, 2007). Exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) were conducted to test for validity and reliability of the SCS in the Ugandan youth population.

Results: EFA resulted in all commonalities with higher than .469 loadings. The Kaiser-Meyer-Olkin measure of sampling adequacy was .95 and Bartlett’s Test of sphericity was significant χ2 (325) =10704.05, p<.001. Given these criteria, we found four rotated matrices as opposed to six constructs specified by Neff (2003), with overlaps in construct specifications, e.g., self-judgement and isolation factors loaded together; self-kindness and common humanity factors loaded together with a couple of mindfulness items (“when something upsets me I try to keep my emotions in balance”). Others, such as self-judgement and overidentification factors loaded separately. Resulting factors had a minimum of .4 or above factor loadings with 66% variance explained. CFA was also used to test a single factor and a two-factor model considering the positive and negative items. All the path coefficients between the individual items and the latent variable are positive and significant, except for one. Positive scale items were negatively associated with negative scale items. The resulting 2-factor model had acceptable goodness-of-fit indices [RMSEA<.05; CFI>.95; TLI>.95; SRMR<.08].

Conclusion and Implications: The two-factor model of the SCS was valid among youth in Uganda, making the SCS a potentially useful measure for mental health interventions in SSA. Though the SCS was valid among Ugandan adolescents, identification of cultural points of compassion to consider the sociometric components of the SCS is needed. Among those living in societies that have high social and economic inequities (e.g., Uganda), developing SC is critically needed as part of a multi-factored approach to enhancing mental health and reducing HIV vulnerability. Scale adaptations and practice implications will be discussed.