Without a reliable and valid instrument of compassion and self-compassion, it is hard to conduct quality research in compassion even though it is a rapidly growing area of interest in many scientific fields. However, recently the Sussex-Oxford Compassion Scales (SOCS; Gu et al., 2020) were developed, with two parallel versions for compassion (SCOS-O) and self-compassion (SCOS-S) containing 20 items each. Because the SOCS scales are newly developed measures of compassion, this is the first study to be conducted beyond the original research (Gu et al., 2020) that validates their factor structure. Our sample consisted of 1080 respondents; 19.35% were males and 80.65% were females. Mean age was 29.29 (SD = 11.15). Convenience sampling via social networks was used. To analyse the data, we reproduced all the confirmatory models in Gu et al. (2020) for both scales: the one-factor model, five-factor model, and five-factor hierarchical model. Our research results showed that for the SCOS-O, the multidimensional definition of compassion for others fits the factor structure of the scale, but this does not apply to the SCOS-S. Self-compassion has two dominant factors over and above five specific factors: Rational Compassion (containing two specific factors: Recognising Suffering and Understanding the Universality of Suffering) and Emotional/Behavioural Compassion (containing three specific factors: Feeling for the Person Suffering, Tolerating Uncomfortable Feelings, and Acting or Being Motivated to Act to Alleviate Suffering). Therefore, the total self-compassion score is unsuitable for use because the scale lacks essential unidimensionality; however, the compassion for others total score can be used safely.
