The Development and Psychometric Evaluation of Two New Scales of Self-Compassion for Preadolescents

The study of self-compassion has flourished in the decades following its inception into the academic literature in 2003. Self-Compassion was pioneered by researcher Kristen Neff (2003a) and defined most simply as “compassion turned inward” (Neff, 2016b, p. 265). Findings from multiple studies have consistently supported Neff’s initial hypothesis that adults and adolescents who are able to relate to themselves with kindness and caring (i.e., are more self-compassionate) tend to display greater resilience and report greater levels of subjective wellbeing (e.g., Barnard & Curry, 2011; Neff, 2009; Neff, Rude, & Kirkpatrick, 2007; Wei et al., 2011). Consequently, interventions aimed at inspiring a healthy and balanced attitude towards oneself, particularly during times of hardship, failure, or distress, are increasingly finding their way into the clinician’s ‘toolbox’ (Barnard & Curry, 2011; Gilbert & Procter, 2006). More recently, investigation has turned towards the existence, and/or relevance, of self-compassion in younger populations—such as those in their middle years of childhood (Stolow et al., 2016; Sutton et al., 2018). This preadolescent period of development offers a unique opportunity for young people to foster positive ways of relating to themselves with love, support and understanding, rather than blame, judgement, or cruelty, especially during a time when the conceptual sense of self is still emerging and malleable (Bosacki, 2016).

Unfortunately, theoretical and methodological issues pertaining to the most appropriate means to conceptualise and measure self-compassion have impeded accurate and consistent measurement of this construct. Prominent researchers in the field are divided (see Muris, Otgaar, & Petrocchi, 2016; Neff, 2016a): One position asserts that self-compassion can, and should, continue to be measured with the current version of the Self-Compassion Scale (SCS; Neff, 2003b), or its abbreviated version, the Self-Compassion Scale-Short Form (SCF-SF; Raes et al., 2011). Meanwhile, an alternate view contests that the use of a total self-compassion score is not warranted and argues for separating the SCS into two separate scales: one measuring the positive, protective features of self-compassion, and the other measuring a theoretically distinct construct, coined by some ‘self-coldness’ (e.g., Gilbert et al., 2011), others ‘self-criticism’ (e.g., López et al., 2015), or ‘self-condemnation’ (e.g., Dundas et al., 2016). Muris and colleagues further assert that self-compassion should only be measured by its positive, protective qualities (Muris, Meesters, et al., 2016; Muris, Otgaar, & Petrocchi, 2016; Muris & Petrocchi, 2017).

Currently, the most widely used measure of self-compassion is the 26-item self-report SCS (Neff, 2003b). Consistent with Neff’s (2003a) seminal definition of self-compassion, half of the scale items are designed to capture the core ‘positive’ components of self-responding, namely self-kindness, common humanity, and mindfulness. The other half of the items are designed to capture the tendency for ‘negative’ self-responding; namely self-judgment, isolation, and over-identification.

Neff’s initial validation study of the SCS (Neff, 2003b) comprised a series of confirmatory factor analyses indicating that a six-factor correlated model and a higher-order model were a reasonable fit to the data. Neff (2003b) justified the use of a total score to represent a single self-compassion construct with six subscales. However, more recent efforts to validate the SCS within different populations have yielded mixed findings. Some researchers have successfully replicated Neff’s (2003b) six-factor correlated structure utilising confirmatory factor analyses in community (e.g., Garcia-Campayo et al., 2014; Mantzios et al., 2015) and clinical samples (e.g., Castilho et al., 2015). Meanwhile, others have been unable to replicate the six-factor factor solution (e.g., Brenner et al., 2017; Costa et al., 2016; López et al., 2015; Petrocchi et al., 2014; Williams et al., 2014).

These discrepant and inconsistent findings have cast doubt on the SCS’s validity and reliability. In addition, as only three published studies have replicated the higher-order model of self-compassion (Castilho et al., 2015; Cunha et al., 2016; Dundas et al., 2016) there is debate as to whether a total self-compassion scale (computed from both the positive and negative scale items) is a theoretically valid measure of the self-compassion construct. Brenner et al. (2017), undertook a detailed examination of the internal structure of the SCS using oblique, higher-order, and bi-factor structural models in a sample of 1,115 college students. They found support for a bi-factor model comprised of two general factors; one comprised of the positive items (‘self-compassion’) and the other the negative items (‘self-coldness’). The authors reported their findings as compelling evidence that the SCS measures the presence of two “theoretically distinct constructs” (p. 696), rather than a single underlying self-compassion factor advocated by Neff and colleagues.

According to the evolutionary theories articulated by Gilbert (2005), compassionate behaviour and uncompassionate behaviour have a different underlying physiology and therefore require separate measurement; Gilbert et al. (2011) stated that self-compassion is better represented by only the three positive constituents (i.e., self-kindness, common humanity, and mindfulness). This stance is congruent with that of Muris and his colleagues, who have strongly argued for a distinction to be made between the positive, protective qualities of self-compassion, and the opposing qualities that they also termed self-coldness (Muris, 2016; Muris, Meesters, et al., 2016; Muris, Otgaar, & Petrocchi, 2016). In a recent meta-analysis, Muris and Petrocchi (2017) presented evidence that the items measuring the ‘negative’ components of compassion (i.e., those items that load on to the self-judgement, isolation, and over-identification subscales) are more strongly associated with psychopathology (.47 < r < .50) than the positive components (-.27 < r < -.34). Muris and Petrocchi (2017) concluded that use of a total SCS score will result in an “inflated relationship with symptoms of psychopathology” (p. 373) and thus warned future researchers against using any of the negative items of the SCS when measuring self-compassion.

In response to the ongoing controversy, Neff and colleagues (2019) published a large-scale study (N = 11,685) to examine the factor structure of the SCS with 10 community, 6 student, 1 mixed community/student, 1 meditator, and 2 clinical samples. Data were analysed using exploratory structural equation modelling (ESEM). For all samples, “excellent fit” (p. 2) was found for an ESEM six-factor correlated model and an ESEM single bifactor model, with 95% of item variance was explained by a general factor. In conclusion, the authors surmised that there was support for “use of a total SCS score, or six subscale scores, but not two separate scores representing compassionate and uncompassionate self-responding” (p. 3). This finding is in direct contrast to the two-factor models (reflecting positive and negative ways of self-responding) indicated by the results of Brenner et al. (2017), Costa et al. (2016), Pfattheicher et al. (2017), and López et al. (2015), respectively.

Theoretical and methodological issues pertaining to the most appropriate measurement of self-compassion in adult populations are mirrored within the literature pertaining to youth. Currently, the only validated adolescent scale of self-compassion is the Shortened Self-Compassion Scale for Adolescents (S-SCS-A; Muris, Meesters, et al., 2016). Consistent with Muris’s (2016) argument to reconceptualise self-compassion as a purely positive, protective construct, this nine-item questionnaire was derived from a selection of items from only the positive subscales of Neff’s (2003b) original SCS. Items were then simplified and modified with the assistance of a panel of three young adolescents and three psychologists. Results from a Principal Components Analysis and Oblique Rotation method findings supported the hypothesised three-factor structure (i.e., self-kindness, common humanity, and mindfulness). All subscales demonstrated adequate internal reliability (all Cronbach’s alphas > .70); concurrent validity was demonstrated via positive correlations with measures of self-esteem and self-efficacy.

At the time of completing this study, only two published studies had attempted to measure self-compassion in children under 12. The earliest, by Stolow et al. (2016) used an adaptation of Neff’s (2003b) original 26-item SCS. One hundred and ninety-three children and adolescents were sampled across three cohorts: Grade 5 (mean age 9.9 years), Grade 8 (mean age 12.7 years), and Grade 11 (mean age 16 years). Utilising Principal Components Analysis with an Oblique Rotation their scale revealed a two-factor solution. Factor 1 was comprised of the ‘negative’ items from the self-judgment, isolation, and over-identification subscales. Factor 2 consisted of items from the ‘positive’ subscales of self-kindness, common humanity, and mindfulness. The two factors did not correlate with one another (r = -.03), indicating that an Orthogonal Rotation may have provided a clearer solution than the Oblique Rotation employed (Field, 2009; Tabachnick & Fidell, 2013). All items (bar item 22, which was subsequently removed from the analysis) had factor loadings exceeding .40. Factor 1 was named ‘Self-Compassion Scale Negative Items’ (SCS-NEG), and Factor 2 ‘Self-Compassion Scale Positive Items’ (SCS-POS). Both factors displayed high internal consistency across the whole sample and for each age grade (all Cronbach’s alphas > .8). This two-factor solution was consistent with the findings of Brenner et al. (2017), Costa et al. (2016), Pfattheicher et al. (2017), and López et al. (2015) in their respective examinations of the factor structure of the SCS with adult and adolescent populations. Stolow et al. (2016) concluded that their results indicate a “need to re-examine the factor structure of the SCS, across age and other demographic variables” (p. 15).

Sutton et al. (2018) chose to administer a modified version of the 12-item SCS-SF (Raes et al., 2011) to a sample of 382 children aged 8–12 years in Canada. According to their methodology, the SCS-SF scale was adapted, “by altering the language to be age-appropriate” (Sutton et al., 2018, p. 6). Data was analysed via confirmatory factor analyses of three competing models. The first was a unidimensional model, where one common factor (self-compassion) was regressed onto the twelve items; it had poor fit. A second model examined the fit of two separate but correlated factors, comprising the negatively worded items and the positively worded respectively; this model had fair fit. Finally, based on the recommendations of Neff (2016a, 2016b), a bi-factor model was also examined. While this model showed adequate fit, examination of the standardised factor loadings revealed that the general self-compassion factor did not account for significant variance among the items. Rather, items loaded more strongly onto their corresponding negative and positive self-responding factors. Therefore, Sutton et al. (2018) concluded that their 12-item scale had a two-factor structure, whereby the negatively worded items and positively worded items formed two discrete subscales. They named these subscales ‘positive self-compassion’ and ‘negative self-compassion’ respectively. Each factor demonstrated acceptable internal consistency (.81 < Cronbach’s alpha < .83). Echoing Stolow et al.’s (2016) conclusions regarding their adapted SCS, these findings add further weight to the literature reporting that the SCS measures two distinct constructs rather than one coherent and overarching self-compassion construct (e.g., Brenner et al., 2017; Costa et al., 2016; Pfattheicher et al., 2017; López et al., 2015).

Unfortunately, limitations identified within both the Stolow et al. (2016) and Sutton et al. (2018) studies cast doubt as to the utility of their respective modified self-compassion scales. Firstly, examination of the methodologies of both studies revealed an apparent lack of rigour in terms of item selection and scale development. Neither study reported to have employed a clear method of selecting, adapting, or rewording their scale items, nor was item comprehension checked within their child samples prior to scale administration. In contrast, Muris, Otgaar, and Petrocchi (2016), outlined a detailed a procedure of item selection, deletion and/or modification in their development of a self-compassion scale specifically for adolescents (i.e., the S-SCS-A).

Secondly, there are questions regarding the generalisability of findings from both the Stolow et al. (2016) and Sutton et al. (2018) studies. For example, the unnamed 26-item scale adopted by Stolow et al. (2016) was administered to children from a broad age range (9-16 years), within a relatively small sample (N = 193). Thus, the applicability of their overall findings to the preadolescent age range is uncertain. Meanwhile, Sutton et al. (2018) sampled a larger number of children within a more specific preadolescent age range (8–12 years); however, they used a shortened 12-item measure, ruling out any potential for more detailed subscale analysis.

Thirdly, neither study attempted to gather data from a parent-perspective, an important oversight given the valuable information parents can provide to supplement children’s self-reports (Snow et al., 2005). Levitt et al. (2007) emphasise “the importance of relying on multiple informants” in the assessment of childhood disorders (p. 176). This is echoed by Achenbach (2006), who argued that even though discrepancies are frequently found between self-reports and reports by others, the “use of multi-source data is essential for clinical assessment” (p. 94).

More recently, Neff et al. (2020) published a new scale designed to measure self-compassion in youth. Over a series of four studies, these authors developed and validated the Self-Compassion Scale-Youth version (SCS-Y) with children attending Grades 6, 7 and 8 of middle school in the USA (mean age = 12.4). Their final 17-item scale contained three items from each of Neff’s original SCS six subscales (with the exception of the over-identification subscale, which had two items), with each item re-worded to be more “developmentally appropriate” (p. 3). Utilising ESEM, Neff et al. (2020) concluded that their findings mirrored those of the adult SCS, supporting the use of the six subscale scores and a single ‘global’ self-compassion score (calculated as a grand mean of the subscales’ scores, after reverse scoring the negative items). This is in direct contrast to previous studies involving the measurement of self-compassion in children where both Stolow et al. (2016) and Sutton et al. (2018) found strong support for a two-factor solution using exploratory and confirmatory factor analyses (CFA), respectively. It is noted that ESEM, whilst an important step in survey development, has fit indices that are somewhat more liberal than CFA (Marsh et al., 2009), which may explain the different findings reported in these studies. Neff et al. (2020) did not attempt to obtain a parent-reported measure of self-compassion, which is a unique contribution of the current study.

Method

This study was granted approval by the Human Ethics Committee of the University of Southern Queensland, approval number H17REA022.

Results

Discussion

The aim of this study was to develop, pilot, and provide preliminary evidence of validation for two new measures of self-compassion for children in their preadolescent years (i.e., aged 9–12 years)—a self-report measure (SCS-P) and a parent-reported measure (SCS-P-PR)—to enable accurate and consistent evaluation of this construct. The ‘tween’ years were chosen in favour of a younger, or broader age-range of children, to capture those within a similar range of cognitive and language abilities. In Australia, this age bracket covers children in their last two years of primary school (i.e., Years 5 and 6), and thus face an impending transition to high school. Therefore, self-compassion may hold particular relevance to this cohort, who are likely to require support to successfully navigate this transition. It was envisioned that development of a parent-perspective scale could supplement the children’s self-report and thus provide a more detailed, and holistic, assessment of the self-compassion construct.

Prior to the development of the scales in this study, only two prior studies had aimed to investigate self-compassion in children under 12 years using different adaptations of Neff’s SCS (Neff, 2003b): Stolow et al. (2016) and Sutton et al. (2018). However, as neither study employed a clear method of selecting, adapting, or rewording their scale items, nor checked for item comprehension within their child samples prior to scale administration, these differing measures were not deemed suitable to explore self-compassion within the preadolescent age group. This study developed two new scales following a similar process to the S-SCS-A (Muris, Meesters, et al., 2016), using Neff’s (2003b) SCS as a template. The SCS-P-PR became the parent-perspective measure, utilising the same order and wording of items to facilitate comparison with the child-reported measure. This process was consistent with the development of other informant-rated scales, such as the PANAS-C-P and SDQ parent form.

Exploratory factor analysis was the chosen method of factor analysis for both the SCS-P and SCS-P-PR, as exploratory methods are necessary when prior research is scarce, and/or when piloting new measures (Patil et al., 2008). As little is known regarding the conceptualisation of self-compassion within cohorts of preadolescent children, and the two measures were being piloted for the first time, an exploratory method of factor analysis was deemed more favourable than confirmatory methods.

Preliminary results were promising and indicative of both measures having sound psychometric properties. From a maximum likelihood method, a two-factor solution emerged as the best fit to the data for both measures. In both cases, the same items loaded onto the same two factors: one representing the ‘positive’ facets of self-compassion (i.e., the items designed to tap into self-kindness, common humanity and mindfulness), and the other representing the ‘negative’ facets of self-compassion (i.e., self-judgement, isolation and over-identification). The two factors of the SCS-P and SCS-P-PR, named Uncompassionate Self-Responding and Compassionate Self-Responding, showed strong internal consistencies (Cronbach’s alphas ranging from .88 to .95). Furthermore, each factor displayed concurrent validity; significant correlations in the expected directions were found between Uncompassionate Self-Responding and Compassionate Self-Responding with indicators of psychosocial wellbeing (i.e., positive and negative affect, satisfaction with life, internalising difficulties and externalising difficulties, prosocial behaviour) and resilience on both the SCS-P and SCS-PR, respectively.

Further preliminary analyses revealed that convergent validity was evident between the SCS-P and the SCS-P-PR. Congruence between the factors elicited on both the self-report and parent-report versions of the scale were confirmed with strong Tucker's phi coefficients (Lorenzo-Seva et al., 2002). Despite only a small sample of matched-pairs (n = 65), significant, moderate correlations (.30 < r < .40) were found between child self-reported and parent-reported Uncompassionate Self-Responding and Compassionate Self-Responding. This is an important preliminary finding. Firstly, it supports prior research that has suggested that self-compassion is a trait observable by others (e.g., Neff & Beretvas, 2013). Secondly, it indicates the viability of collecting potentially valuable supplementary data regarding a child’s self-compassion from an informant. In fact, the method of triangulation of data from multiple informants is frequently recommended by researchers interested in studying psychological phenomena as a method to enhance assessment (Snow et al., 2005). Meta-analyses have revealed that cross-informant agreement in the field of child and adolescent research is approximately r = .30 (Achenbach et al., 1987). While this modest correlation could be interpreted as reflecting low reliability between informants, Achenbach et al. (1987) argued that different informants validly contribute different information: “low correlations between informants may indicate that the target variables differ from one situation to another, rather than that the informants' reports are invalid or unreliable” (p. 213).

As no prior research had assessed the convergent validity between child self-report and parent-rated self-compassionate responding, this study addressed an important gap within the current knowledge. Future research could assess more precisely what the parent-report ‘adds’ to the assessment of self-compassion, beyond that supplied via self-report. Indeed, according to the Self-Other Knowledge Asymmetry Model (Vazire, 2010), there are many personality constructs that others can describe better than the self. Where self-compassion lies on this continuum of self vs other-knowledge remains to be investigated.

The findings of this preliminary validation study suggests that both the SCS-P and SCS-P-PR scales possess sound psychometric properties; if future studies can replicate these findings, both scales can be used by researchers as reliable and valid measures of self-compassion in children aged 9–12 years. The finding that scores on SCS-P and SCS-P-PR are best represented by two distinct factors (Compassionate Self-Responding and Uncompassionate Self-Responding) rather than six subscales or one overarching construct, is contrary Neff’s (2003b) recommended practice of subscale analysis and/or combining the positive and negative items of the SCS into a total self-compassion score. Rather, the finding that items on the SCS-P and SCS-P-PR formed two orthogonal constructs is consistent with a growing body of literature suggesting that the positive and negatively worded items of the SCS form two theoretically distinct factors (e.g., Brenner et al., 2017; Castilho et al., 2015; Costa et al., 2016; López et al., 2015). This finding is also consistent with the results of factor analyses reported by Stolow et al. (2016) and Sutton et al. (2018), who similarly administered adapted versions of the SCS to children under 12 years. It is inconsistent, however, with the findings of Neff et al. (2020), in their recent validation of the SCS-Y. Rather than two distinct constructs, Neff and colleagues found support for a bi-factor solution, whereby self-compassion could be reliably represented as six subscale scores, or via calculation of a single score representative of a “self-compassionate mindset” (p. 11). While Neff et al. (2020) recommend future researchers to adopt a single score approach, they do concede that an alternative is to calculate two composite scores, composed of the three positive and three negative subscales scores respectively, as a “heuristic to understand the mechanisms of action entailed by self-compassion” (p. 11).

There is related argument in the literature (Muris, 2016; Muris, Meesters, et al., 2016; Muris, Otgaar, & Petrocchi, 2016; Muris & Petrocchi, 2017), that posits all negatively worded items should be excluded from measures of self-compassion. According to these researchers, the negative items represent indicators of psychological maladjustment, such as anxiety and depression, and therefore are not relevant to the ‘true’ protective nature of self-compassion. Muris, Meesters, et al. (2016) does not include any of the negatively worded items in their adolescent measure of self-compassion (S-SCS-A). However, the results from this study suggest that it is important to measure the negative aspects of self-compassion in addition to the positive aspects; there is evidence that both tendencies can operate independently within preadolescent children, and are not mutually exclusive (i.e., they are not two ends of the same scale). The findings from this study do, therefore, offer support for Neff’s (2003b, 2016a, 2016b) assertions that both the negative and positively worded items on the SCS tap into important elements relevant to the assessment of an individual’s tendencies to relate to themselves during times of suffering, and thus warrant inclusion on measures of self-compassion. The recognition of both positive, and negative, aspects of self-responding is also consistent with the Buddhist teachings on self-compassion (Kraus & Sears, 2008). However, the current findings indicate that compassionate self-responding and negative-self responding (as measured with the SCS-P and SCS-P-PR, respectively) operate as two theoretically and psychometrically distinct constructs rather than two inherent components of the self-compassion construct.

This study extends the knowledge regarding self-compassion’s relationship with other psychological constructs in children. Firstly, findings from the current study support those of Sutton et al. (2018), who similarly found that both positive and negative aspects of self-compassion were significantly associated with satisfaction with life and positive affect in their sample of children aged 8–12 years. Neff et al. (2020) also found the three ‘positive’ self-compassion subscales to correlate positively with life satisfaction, and the three negative subscales correlated in the negative direction, in their sample of children in Grades 6, 7 and 8 (mean age 12.2). Together, these studies suggest that for preadolescents, positive ways of responding to the self (i.e., with kindness, connection, and acceptance), may lead to more objective evaluations, and a greater appreciation of life’s experiences in general. Conversely, a negative mode of self-responding during times of hardship may entail preadolescents becoming over-identified with negative internal and external experiences. This in turn would reduce positive affect and satisfaction with life in general. It is recommended that future research adopt a qualitative approach to attain a deeper meaning of the nature of theses associations in preadolescents.

This study also revealed self-compassion’s relationship with other aspects of psychosocial wellbeing in preadolescence. While associations between self-compassion and psychosocial wellbeing were expected based on robust findings in adult and adolescent populations (Neff, Kirkpatrick, & Rude, 2007; Neff & McGehee, 2010; Neff, Rude, & Kirkpatrick, 2007), the current study was the first to examine, and reveal, that compassionate self-responding was inversely associated with measures of preadolescent externalising difficulties (i.e., conduct and hyperactivity problems), while uncompassionate self-responding was positively associated with these difficulties. One possible interpretation of this finding is that the adoption of a self-compassionate mode of responding during difficult times reduces the likelihood of getting ‘caught-up’ in uncomfortable thoughts and feelings that may otherwise lead to disruptive, externalised behaviour. It is recommended that future researchers explore the nature of the relationship between self-compassion and psychosocial difficulties in greater detail, as this has the potential to inform programs aimed specifically at reducing these problematic behaviours in preadolescents.

Another key finding elicited from this study was that preadolescents’ compassionate self-responding (as reported on both the SCS-P and SCS-P-PR) was positively related to resilience, while uncompassionate self-responding was associated with resilience in the negative direction. This finding was expected based on previous research with adult and adolescent populations (e.g., Bluth et al., 2018; Neff & McGehee, 2010), and is in line with the recent findings of Neff et al. (2020), who reported correlations with resilience in the same directions for their positive and negative subscale scores. An interpretation of these findings is that responding to the self with compassion during times of hardship can provide both an internal source of resilience (e.g., via improved self-concept; Alvord & Grados, 2005), and an external source of resilience (e.g., via social competence, Alvord & Grados, 2005) for preadolescents. Conversely, engaging in negative ways of self-responding during challenging times will inevitably decrease resiliency resources; this may occur at the individual level (e.g., by over-identifying with negative thoughts or feelings) and/or external levels (e.g., via social isolation, causing detachment from resilience resources that may be present in the family or community). Further research exploring the relationships between styles of self-responding and resilience is required to replicate and validate these interpretations.

Finally, it is noted that gender was not found to be related to preadolescents’ compassionate self-responding or uncompassionate self-responding in this study. Prior research with children has reported mixed results in this regard (Neff et al., 2020; Stolow et al., 2016; Sutton et al., 2018). It may be the case that preadolescent children display reduced gender differences than when compared with their adolescent counterparts (see Bluth & Blanton, 2015), as gender identity is less pronounced during this phase of development. Future research with larger samples is needed to clarify the nature of gender differences in the preadolescent age-range.

This study has several limitations, including modest sample sizes and reliance on convenience sampling methods. Notably, mothers were greatly over-represented in the sample of parents completing the SCS-P-PR; future validation studies will need to focus on collecting father’s perspectives to redress this balance. Another limitation of this study was the employment of a cross-sectional, rather than longitudinal, design. Therefore, causal pathways in the models presented cannot be determined. Longitudinal studies are recommended to future researchers. In addition, while efforts were made to employ a large, diverse sample of Year 5 and Year 6 students from the Darling Downs Region of Queensland, the effects of clustering bias cannot be ruled out and therefore generalisability is limited. Replication of this study’s findings across larger and more diverse samples of children and youth is vital, as self-compassion is likely to be interpreted differently depending on culture, race, ethnicity, and social economic status.

Measurement factors may have also influenced the results of the factor analyses of the SCS-P and SCS-P-PR. While phrasing half of a scale’s items in a positive direction, and half in a negative direction can help control for respondent response bias (Comrey, 1988), it has also been revealed that positively and negatively worded items frequently load onto distinct factors (see Spector et al., 1997). Indeed, research conducted with children from Grades 4–6 (i.e., aged 8–12 years) revealed that children were more likely to agree with an item that was positively-worded than they were to disagree with an item that was negatively worded (Benson & Hocevar, 1985). However, following the evolutionary arguments of Gilbert et al. (2011) it can be contended that it is theoretically logical that the positive and negative aspects of self-responding should be measured separately, as they are related to different affective and physiological systems. Therefore, to rule out measurement artefacts as influencing the current results, future research should apply different methods of balancing item phrasing to correct for this potential form of response bias.

To surmise, the availability of psychometrically sound instruments is essential to enable accurate and consistent measurement of the self-compassion construct in preadolescent populations. This research has provided preliminary evidence supporting the viability of both the SCS-P and SCS-P-PR scales as valuable contributions to the toolbox of measures available for the assessment of self-compassion in preadolescents. As the first known parent-reported measure of self-compassion, the SCS-P-PR can be used to both supplement and enhance the assessment of this construct within the preadolescent age-range. Indeed, the development and validation of the SCS-P and SCS-P-PR marks an important step towards improving the screening and measurement of self-compassion within this cohort. It is noted, however, that validation is an ongoing process. Therefore, further research using the SCS-P and SCS-P-PR across more diverse populations is essential to ensure that the promising results from this preliminary study can be replicated. Furthermore, the SCS-P-PR offers a means to both educate and include parents in the assessment of their child’s style of self-responding and enable the accurate identification of both protective features and vulnerabilities to support the mental health and development of preadolescents. Indeed, future research based on qualitative research methods may provide a deeper understanding of how compassionate self-responding, and uncompassionate self-responding, interact and influence mental health and wellness in Australian youth. Establishing how the different aspects of self-compassion protect and/or promote vulnerability towards psychosocial wellbeing and resilience in preadolescent children is a critical direction for future research.