A Protective Factor Against Mental Health Problems in Youths? A Critical Note on the Assessment of Self-Compassion

Without any doubt it was Kristin Neff who gave the research on self-compassion an enormous boost. This American professor in psychology, currently working at the University of Texas at Austin, was the first to systematically investigate self-compassion in relation to people’s mental health problems. Neff produced a considerable number of scientific publications, while also advocating the protective power of this self-related concept in the popular media. According to Neff (2003), self-compassion consists of three key components. The first component is self-kindness and refers to the tendency to be caring to and understanding with oneself when confronted with personal failures. The second component is common humanity, which is concerned with the inclination to recognize that one’s failures and problems are a normal part of human life. The third and final component is mindfulness and has to do with the ability of not becoming too absorbed with personal failures and problems but rather to maintain a balanced view on what goes right and what goes wrong.

Since its introduction, more than a decade ago, self-compassion has been increasingly examined as a protective factor within the context of mental health in adults. Most of the evidence obtained in this research indicates that this construct is indeed inversely related to psychological well-being: individuals with higher levels of self-compassion generally exhibit lower levels of anxiety, depression, stress, and other psychopathological symptoms (e.g., MacBeth and Gumley 2012). Of particular interest is that self-compassion may also provide a lead for clinical interventions: by promoting these self-soothing skills, psychological health can be enhanced. This idea has not only been embraced by Acceptance and Commitment Therapy which currently includes self-compassion as one of its main targets (Hayes et al. 2012), but there are also treatment programs that specifically focus on the enhancement of this self-related construct (e.g., Gilbert 2009; Neff and Gerner 2013; Smeets et al. 2014).

A systematic search carried out in the Web-of-Science database with [SELF-COMPASSION] and [CHILD* or ADO*] as search terms learns that so far nine studies have investigated self-compassion in relation to mental health problems in young people. As can be seen in Table 1, studies were conducted with a total of 3477 young participants (1864 boys and 1613 girls) who were predominantly of an adolescent age (range between 11 and 20 years). Five of the studies were conducted with randomly recruited middle and high school students. The other four investigations made use of at-risk samples of youths who (a) had a history of maltreatment (Tanaka et al. 2011), (b) poorly performed at school and thus were prone to victimization (Jativa and Cerezo 2014), (c) attended a residential program (Barry et al. 2015), or (d) had been exposed to a potentially traumatic event (a forest fire disaster; Zeller et al. 2015). In terms of type of psychopathology, all studies incorporated measures for assessing internalizing symptoms such as anxiety, depression, stress etc., but a number of investigations also included measures of externalizing problems such as aggression and substance use.

The correlations between self-compassion and various symptoms of psychopathology, as documented in these studies, are shown in the right column of Table 1. As can be seen, correlations were all negative, thus indicating that higher levels of self-compassion were associated with lower levels of psychopathology and vice versa. Using Wilson’s (2010) online meta-analysis calculator, correlations were transformed into effect sizes (r), which were averaged for each study and then pooled across all studies. The overall effect size for the relationship between self-compassion and symptoms of psychopathology in young people was large, r = −0.49 (95 % CI −0.67 to −0.32), and comparable to the pooled effect size of r = −0.54 as reported by MacBeth and Gumley (2012) in their meta-analysis of studies conducted with adults. Thus, the results of research carried out in young people correspond nicely with what has been reported in the adult literature, and seem to be in keeping with the notion that self-compassion is a protective factor against psychopathology.

Researchers widely rely on Neff’s (2003) Self-Compassion Scale (SCS) or its abbreviated version, the Self-Compassion Scale-Short Form (SCS-SF; Raes et al. 2011), for measuring this construct. Critique can be raised regarding the validity of these scales because half of the items do not measure the three key components of self-kindness, common humanity, and mindfulness, but rather assess their precise counterparts of self-judgment, isolation, and over-identification. These negative items, which are reversed before adding them to the positive items to produce a total self-compassion score, are problematic as they seem to measure characteristics that are already known to be associated with psychopathology. More specifically, self-judgment shows clear similarity with harsh self-criticism (e.g., Zuroff et al. 1990), isolation shares features with social withdrawal and loneliness (e.g., Rubin and Coplan 2004), whereas over-identification matches with self-absorption and self-focused rumination (e.g., Lyubomirsky and Nolen-Hoeksema 1995), all of which have been demonstrated to be pervasive features of mental health problems, especially those of an internalizing nature. Thus, it can be argued that the widely used SCS and its abbreviated version, the SCS-SF, may not be optimal instruments for measuring the true protective nature of self-compassion, mainly because these scales include negative items that tap toxic mechanisms which may inflate the relationship with psychopathology.

So far, studies on the relation between self-compassion and psychopathology in youths have also largely neglected this assessment issue with the SCS or SCS-SF. Of the eight studies that used the original SCS or SCS-SF (see Table 1), only Barry et al. (2015) reported on the links between separate self-compassion components and psychopathology in their sample of residential adolescent males. Their data confirm that the negative components were more substantially associated with psychopathology indices than the positive components, but the researchers remain silent about the implications of this result. In a recently conducted study, I was able to directly examine the differential associations between the negative and positive components of self-compassion and psychopathology in young people. One-hundred-and-eighty-four high school students aged 12–16 years completed the SCS-SF and Achenbach’s (2009) Youth Self-Report for measuring internalizing and externalizing symptoms. As displayed in Table 2, total self-compassion showed the expected negative correlations with psychopathology, although the link with internalizing was clearly more substantial than that with externalizing (Z = 4.10, p < 0.001). Most importantly, in the case of internalizing, the negative components of self-compassion were stronger correlates of symptoms than the positive components (Z = 4.69, p < 0.001). An additional analysis revealed that a total of 38.2 % of the variance in internalizing symptoms could be explained by self-compassion. However, the contribution of the positive components was only 6.1 %, whereas the remaining 32.1 % was attributable to the reversely scored negative components. In the case of externalizing, the positive and negative components appeared to be equally strong correlates of symptoms, and here the inclusion of the negative components had less influence on the total variance explained (i.e., 2.2 of 6.5 %). In conclusion, these data are the first to actually demonstrate that the negative components in measures such as the SCS and SCS-SF produce inflated correlations with psychopathology. This inflation effect seems to be stronger for internalizing than for externalizing symptoms, but obviously this impression needs further investigation.