Original Research Article Assessment 1–14 Exploring the Relationship Between Ó The Author(s) 2023 Self-Compassion and Compassion for Article reuse guidelines: sagepub.com/journals-permissions Others: The Role of Psychological DOI: 10.1177/10731911231203966 journals.sagepub.com/home/asm Distress and Wellbeing Javier Garcı́a-Campayo1,2, Alberto Barceló-Soler1,2,3, David Martı́nez-Rubio4 , Jaime Navarrete5,6, Adrián Pérez-Aranda1,7 , Albert Feliu-Soler6,7, Juan V. Luciano5,6,7, Ruth Baer8, Willem Kuyken8, and Jesus Montero-Marin5,6,8 Abstract We addressed construct validity and explored the relationship between self-compassion and compassion for others using the two main current operationalizations of compassion (Neff’s and the Sussex-Oxford Compassion Scales, SOCSs). Relationships with psychological distress and wellbeing, and potential differences in the association between self-compassion and compassion for others by level of psychological distress and wellbeing were also explored. Participants (n = 811) com- pleted the Spanish adaptations of the Self-Compassion Scale (SCS), the Compassion Scale (CS), the SOCSs (for the self/oth- ers), the Short Warwick–Edinburgh Mental Well-Being Scale (SWEMWBS), and the Depression Anxiety Stress Scales-21 (DASS-21). We fitted bifactor models to estimate the general factor of each construct for the different operationalizations, and calculated correlations between them. Relationships between self-compassion and compassion for others from the same operationalization were intermediate, while those between the same constructs from different operationalizations were large. Both constructs showed positive associations with wellbeing, while only self-compassion was associated with decreased psychological distress. Participants with good mental health showed higher associations between self-compassion and compassion for others than those with poorer mental health. Self-compassion and compassion for others appear to be dimensional constructs that can converge or diverge. When they converge, it is associated with better mental health. Keywords Sussex-Oxford Compassion Scales, Compassion Scale, Self-Compassion Scale, self-compassion, compassion for others, psy- chological distress, wellbeing 1 Introduction 2 Institute for Health Research of Aragon (IIS Aragon), Zaragoza, Spain University of Zaragoza, Spain 3 Compassion is an important human capacity to respond Navarra Medical Research Institute (IdiSNA), Pamplona, Spain 4 to suffering and has been demonstrated to be associated Universidad Europea de Valencia, Spain 5 Teaching, Research & Innovation Unit, Parc Sanitari Sant Joan de Déu, with decreased anxious and depressive symptomatology, Sant Boi de Llobregat, Spain improved coping with pain, and positive outcomes of 6 Consortium for Biomedical Research in Epidemiology & Public Health quality of life and wellbeing (A. López et al., 2018; (CIBER Epidemiology and Public Health—CIBERESP), Madrid, Spain 7 Mongrain et al., 2011; Van Dam et al., 2011). Many psy- Autonomous University of Barcelona, Spain 8 chological interventions either implicitly or explicitly University of Oxford, UK (i.e., compassion-based psychotherapies) cultivate com- Corresponding Authors: passion, and changes in compassion appear to mediate Adrián Pérez-Aranda, Institute for Health Research of Aragon (IIS Aragon), changes in mental health and wellbeing (Galante et al., Avda. San Juan Bosco, 13, 50009 Zaragoza, Spain. 2014; Navarro-Gil et al., 2020). Email: aparanda@iisaragon.es Compassion is central to many spiritual traditions, Jaime Navarrete, Teaching, Research & Innovation Unit, Parc Sanitari Sant including Buddhism, Christianity, Hinduism, Judaism, Joan de Déu, C/ del Dr. Antoni Pujadas, 42, 08830 Sant Boi de Llobregat, and Islam. Although it can take different forms, the Spain. notion that compassion can be trained and cultivated, as Email: jaime.navarrete@sjd.es 2 Assessment 00(0) well as the intention to transcend self-centered concerns and emotionally connecting with their distress; (d) toler- and the invitation to respond in a friendly manner to ating any uncomfortable feelings aroused in response to pain and suffering, is present in all of them (Feldman & the suffering, so that, we remain accepting and open to Kuyken, 2019). Compassion can be traced in our evolu- the person suffering; and (e) being motivated to act to tionary lineage as a strong instinct that builds social alleviate the suffering. groups, and allow to include those individuals who are These five dimensions can be framed theoretically as an vulnerable (Darwin, 1871). It has been described from unfolding process that broadly speaking starts with recog- ethology through different layers, such as the recogni- nition and progresses to action. They draw on contempo- tion of suffering, concern for others, perspective taking, rary commentators (Feldman & Kuyken, 2011; Gilbert, and targeted helping (De Waal, 2009). Compassion and 2014) to suggest that the processes may be similar for com- empathy are often used interchangeably, but they are passion oriented toward others (i.e., compassion for oth- distinct concepts. While empathy is the cognitive ability ers) and for compassion oriented to the self (i.e., self- to understand and share the feelings of others (Davis, compassion). However, not everyone agrees with this. For 1983; Decety & Jackson, 2004; Singer & Lamm, 2009), example, compassion for others has been understood as compassion involves taking action to alleviate their suf- an emotion that arises when witnessing another’s suffering fering, as it is an understanding imbued with intention and that subsequently motivates a desire to help (Goetz (Gilbert, 2010; Goetz et al., 2010; Keltner & Kogan, et al., 2010), while self-compassion has also been concep- 2018). Thus, while empathy can be a precursor to com- tualized as a system of attitudes toward our own suffering passion, compassion requires the additional step of that can present different characteristics, such as harshness being motivated to help. or kindness, a sense of isolation or the identification of suf- In Buddhism, compassion is one of the four qualities fering as a common human experience, and a tendency to of mind and heart that lead to a happy and fulfilling life overidentify oneself with suffering or to maintain a (Nyanaponika & Bodhi, 1999). Compassion is the wish balanced awareness (Neff, 2003). Testing these premises that all beings be free from suffering and the willingness relies on having adequate measures of compassion, both to help alleviate that suffering. Buddhism also empha- for oneself and others. sizes the importance of self-compassion, which is the There is a range of questionnaires designed to mea- practice of treating oneself with kindness, understand- sure self-compassion and compassion for others, but the ing, and acceptance. This value is rooted in the assump- systematic review conducted by Strauss et al. (2016) con- tion that to have true compassion for others, one must cluded that most presented weak psychometric proper- first cultivate compassion for oneself. The practice of ties. One instrument that showed at least fair quality self-compassion involves treating oneself as one would (i.e., 7 points of 14; Strauss et al., 2016) was the Self- treat a good friend, acknowledging one’s own suffering Compassion Scale (SCS; Neff, 2003), a 26-item tool that without judgment, and offering oneself kindness and is widely used and has been translated into several lan- support (Neff, 2003). In modern times, these concepts guages (Neff et al., 2019), including Spanish (Garcia- have been adapted into secular practices, such as Campayo et al., 2014). The SCS includes six subscales mindfulness-based programs and compassion training, (self-kindness, self-judgment, common humanity, isola- which have been shown to improve wellbeing, reduce tion, mindfulness, and over-identification), but an over- stress and anxiety, and increase resilience (Hofmann arching score of self-compassion that represents the et al., 2011; Neff & Germer, 2013). general construct has also been proposed and evaluated Surprisingly, only recently have operational defini- using bifactor exploratory structural equation modeling tions of compassion been developed, and this work is (ESEM) (Neff et al., 2019). The same authors have still evolving. Some definitions emphasize the elements developed the Compassion Scale (CS; Pommier et al., of recognition, emotional connection, and desire to alle- 2020), a 16-item instrument to assess compassion for viate suffering (Jazaieri et al., 2013); others highlight dis- others based on Neff’s theoretical conceptualization of tress tolerance (Gilbert, 2009), or the acknowledgment self-compassion (Neff, 2003). The CS operationalizes of suffering as a universal, human experience (Feldman compassion as experiencing kindness, a sense of com- & Kuyken, 2011; Neff, 2003). Strauss et al. (2016) have mon humanity, mindfulness, and lessened indifference systematically reviewed the literature and developed a toward the suffering of others. This four-dimensional comprehensive theoretical conceptualization of compas- operationalization of compassion for others that inte- sion, considering it as a cognitive, affective, and beha- grates all the negative items into only one factor has vioral process that consists of five main elements with originally presented sound psychometric properties empirical support (Gu et al., 2017): (a) recognizing suf- using bifactor ESEM, but it has not been widely used to fering; (b) understanding the universality of suffering in date, and there is not a validated Spanish version of this human experience; (c) feeling for the person suffering scale. Garcı́a-Campayo et al. 3 In response to the need for theoretically comprehen- self-compassion was lower in people with lower levels of sive and psychometrically robust operational measures education. Durkin et al. (2016) observed no significant of both self-compassion and compassion for others, Gu relationships between compassion for others and self- et al. (2020) developed the Sussex-Oxford Compassion compassion in community nurses. Therefore, the associ- Scales (SOCSs) based on the multidimensional concep- ation between self-compassion and compassion for oth- tualization proposed by Strauss et al. (2016). These ers, remains unclear. scales assess compassion for others (SOCS-O), and com- There is a natural motivation to seek care and con- passion for the self (SOCS-S), both along the five dimen- nect with others that is thought to be an evolved trait sions of compassion mentioned above. It has been that has helped build social groups and promote survival suggested that although a total score can be derived in our evolutionary history. However, when we are from both the SOCS-O and SOCS-S using a five-factor struggling with mental health issues, this natural motiva- hierarchical model that has shown adequate fit in sam- tion to seek care and connect with others may become ples of health care workers and university students from distorted or inhibited (Gilbert et al., 2017). In general, it the United Kingdom, as these scales were originally has been observed that individuals tend to display more designed to be multidimensional, it would be important compassion to others than to themselves (Neff, 2003). to examine its potential one-dimensionality in greater However, in populations with high levels of mental dis- detail, and how their dimensions collectively relate to tress, there may be a wide variation in the ways self- outcomes (Gu et al., 2020). In addition, some psycho- compassion and compassion for others are manifested metric properties, for example, test–retest reliability, (Gilbert et al., 2017). Poor mental health may be related remain unexplored. SOCS-O scores were higher in to distortions of our natural inclination to connect with females, in line with A. López et al. (2018), and in people care-seeking experiences from ourselves and others. For with previous meditation experience; the latter finding example, an individual suffering from trauma might dis- was expected because compassion is a core element of play compassion for others but may be filled with self- many contemplative programs (Jahoda et al., 2017). criticism and unable to receive compassion from others Correlations between the SOCS-O and the SOCS-S were (Montero-Marı́n et al., 2016; Van der Kolk, 2014). It significant and moderate, but this finding needs to be has been proposed that this distortion of our innate further tested and cross-validated in other populations motivational system may be a contributing factor to (Gu et al., 2020). poor mental health outcomes. Therefore, by recognizing Some studies have reported small but significant posi- and addressing this distortion, we may be better able to tive correlations between compassion for the self and promote mental wellbeing and improve our ability to compassion for others (Neff & Pommier, 2013; connect with others in a healthy and supportive way Pommier et al., 2020). It has also been found that acti- (Gilbert, 2022). vating support-giving schemas increased self- In this context, we first adapted to the Spanish popu- compassion (Breines & Chen, 2013), and that similar lation and tested the construct validity of the two main brain regions were activated both when expressing self- current operationalizations of compassion—Neff’s reassurance and compassion for others (Longe et al., model and the SOCSs model (Gu et al., 2020; Neff, 2010). However, some studies have also reported no sig- 2003; Pommier et al., 2020)—to see whether they are nificant associations between self-compassion and com- adequate measures of compassion. We expected the new passion for others. Leary et al. (2007) found that Spanish operationalizations to have a similar structure compassion for others was not significantly different to those observed in their original English validations among individuals who presented low vs. high self- (Gu et al., 2020; Neff, 2003; Pommier et al., 2020). compassion. Specifically, they saw in undergraduate stu- Once we established the measurement models for the dents that self-compassion predicted positive affect different constructs and operationalizations of compas- when participants watched a videotape of their own per- sion, we attempted to determine the extent to which self- formance, but not when they watched others’ tapes, sug- compassion and compassion for others converge. For gesting that self-compassion might be distinct from that, we calculated the associations between the CSs in more general feelings of compassion for others. A. the same operationalization (SOCS-O/SOCS-S; CS/ López et al. (2018) found no significant relationships SCS) and between those compassion scales that measure between self-compassion and compassion for others in a the same construct in the different operationalizations community sample of adults, and reported that sociode- (SOCS-O/CS; SOCS-S/SCS). Our exploratory hypoth- mographic factors, such as gender and education level, esis is that relationships between scales that measure were differently associated with each type of compas- the same compassion construct using different operatio- sion: compassion for others was higher in women and nalizations will be stronger than those that measure in people with lower levels of education, while different compassion constructs through the same 4 Assessment 00(0) operationalization, and therefore reflecting potential dif- and reviewed by a research assistant. The link to the sur- ferent modes (i.e., differences in terms of a higher or vey was accessible from June 2021 to December 2021. A lower aggregation) of the motivational system function- total of 1,401 individuals accessed the link and volunta- ing (Gilbert et al., 2017). rily agreed to participate in the online survey (Figure 1). We additionally assessed the explanatory power of Among them, 1,362 were Spanish native speakers, with both self-compassion and compassion for others in 1,153 having Spanish nationality and 209 being from accounting for variance in psychological distress and Latin America. Finally, 811 Spanish native speakers mental wellbeing, and examined whether the relation- with Spanish nationality (Table 1) completed all SOCS- ship between self-compassion and compassion for others O, SOCS-S, CS, and SCS items and thus configured the is modified by levels of psychological distress and mental analytic complete case sample (pairwise deletion was the wellbeing. Previous research (Neff & Germer, 2013) has technique used to handle missing data). Most partici- shown that self-compassion (in short, as an attempt to pants were female (80.0%), with a mean age of 43.49 soothe distress in oneself) is a stronger predictor of men- (SD = 12.8; range = 18–85), and mainly married tal health than compassion for others. Thus, we (53.0%), with no children (48.3%), a university educa- expected self-compassion to be more correlated than tion (82.3%), and in employment (66.6%). A subsample compassion for others with both psychological distress of 288 (35.5%) individuals completed the SOCS-O and (in a negative way) and wellbeing (in a positive way). SOCS-S 1 week later to evaluate the stability over time Finally, we hypothesized that the relationship between (Supplemental Table S1). The characteristics of Spanish self-compassion and compassion for others will be native speakers who accessed but did not complete the stronger in individuals with lower psychological distress survey can be seen in Supplemental Table S2. and higher mental wellbeing, indicating a more aggre- We carried out post hoc power analyses to evaluate gated self-compassion–compassion for others motiva- the viability of a bifactor exploratory structure (Morin tional structure (Gilbert, 2022). et al., 2016) as a measurement model for the SOCS-O, We carried out all of these explorations using two dif- SOCS-S, CS, and SCS (Supplemental Figure S1). This ferent and concurrent operationalizations of compas- measurement model permits us to represent the general sion, allowing us to determine whether the lack of constructs mentioned underlying the corresponding consistency in previous studies (Gu et al., 2020; Leary scales, allowing the coexistence of the general factors et al., 2007; A. López et al., 2018; Pommier et al., 2020) and their subdomains, and including the possibility that could be due to the use of different operational items may present cross-loadings on the nontarget sub- definitions. domains, thus estimating the true variance of the general dimension (Morin et al., 2016). This approach repre- sents the most comprehensive measurement model that Methods can accurately describe complex psychological charac- A cross-sectional internet survey was disseminated teristics, such as those under study, facilitating their through several websites from the authors’ scientific comparability. Details on the statistical procedures that research webpage. It was carried out on a commercial were used to evaluate the viability of this measurement system (https://es.surveymonkey.com) using a series of model’s application to the CSs are in Supplemental questionnaires that were presented in a random order Table S3. To test this, a sample of n = 811 subjects, with every time they were to be completed to control for a null hypothesis that root mean square error of approx- carry-over effects (e.g., missing data, less accuracy due imation (Root Mean Squared Error of Approximation to fatigue, etc.). [RMSEA]) would be ł .05 if the true value was .08 (close fit) and an alpha equal to .05, produces a power coefficient (1-beta) in all the models of .99 (MacCallum Data Collection, Participants, and Ethics et al., 1996). The recruitment of participants was carried out through The study protocol was approved by the Ethical the social networks of the research group. Individuals Committee of the regional health authority of Aragon were invited to participate in a study on ‘‘general aspects (CEICA, PI21/197; 05-May-2021). All participants sub- of compassion.’’ The information that appeared in the mitted a written informed consent form online attesting study announcement was the main objective, the link to to their willingness to participate. the online questionnaire, and an email address where participants could contact in case of any doubt about the study. To control for the risk of responses generated Measures by bots, and so on, the IP address and completion time Sociodemographic Characteristics. We collected informa- of each participant were recorded by the online platform tion about self-identified gender (male and female), age, Garcı́a-Campayo et al. 5 Figure 1. Study Flowchart. marital status (single, married/civil partner, separated/ presented adequate fit indices for both the SOCS-S divorced, and widowed), number of children, education (comparative fit index [CFI] = .995, Tucker–Lewis level (no schooling, primary school, secondary school, index [TLI] = .988, RMSEA = .054; 90% confidence and university), and work status (student, unemployed, interval [CI] = .043–.065, weighted root mean square employed, self-employed, homemaker, and retired/ residual [WRMR] = .418) and SOCS-O (Comparative pensioner). fit index [CFI] = .995, Tucker-Lewis index [TLI] = .988, RMSEA = .054; 90% Confidence Interval [CI] = .043–065, Weighted Root Mean square Residual Compassion for the Self/Compassion for Others [WRMR] = .418). The internal consistency of the gen- Sussex-Oxford Compassion Scales. Permission from the eral factor (omega hierarchical, vh) was good (SOCS-O: original authors was obtained for translating/validating vh = .86, SOCS-S: vh = .87). The descriptive statistics the Spanish version of the SOCSs (Gu et al., 2020). A of the SOCSs, factor loadings, and other reliability description of the adaptation process and the SOCSs indices can be found in Supplemental Tables S7–S9. An Spanish versions are shown in Supplemental Tables S4– exploration of the fit of the other potential measurement S6. The SOCS-S and SOCS-O are two 20-item self- models can be seen in Supplemental Table S10. report measures that contain five subscales (four items for each): recognizing suffering, understanding the uni- versality of suffering, feeling for the person suffering, Self-Compassion Scale. The SCS (Neff, 2003) is a 26- tolerating uncomfortable feelings, and acting or being item self-report measure of the following six constituent motivated to act to alleviate suffering. Items are scored elements of self-compassion (Neff, 2016): self-kindness on a 5-point Likert-type scale, ranging from 1 (‘‘not at (five items), common humanity (four items), mindful- all true’’) to 5 (‘‘always true’’). The higher the score is, ness (four items), self-judgment (five reversed scored the greater the degree of self-compassion/compassion items), isolation (four reversed scored items), and over- for others. The bifactor exploratory measurement model identification (four reverse scored items). The scale has a 6 Assessment 00(0) Table 1. Sociodemographic Data of the Study Participants (‘‘almost always’’) that assesses compassion for others. (n = 811). It is based on the SCS model (Neff, 2003) and consid- ers four constituent elements: kindness (four items), Gender (women): n (%) 649 (80.0) common humanity (four items), mindfulness (four Age (in years): M (SD) 43.49 (12.8) Marital status: n (%) items), and indifference (four reverse scored items). Single 282 (34.8) Higher scores indicate higher compassion for others. Married/civil partner 430 (53.0) The bifactor exploratory measurement model has been Separated/divorced 81 (10.0) proposed as the best representation structure of the CS Widowed 18 (2.2) (Pommier et al., 2020). It presented adequate fit Children: n (%) 0 392 (48.3) indices in this study (CFI = .99, TLI = .98, RMSEA 1 139 (17.2) = .04; 90% CI = .03–.05, WRMR = .52), and the 2 230 (28.4) internal consistency of the general factor was good (vh 3 40 (4.9) = .78). The descriptive statistics, factor loadings, and 4 10 (1.2) other reliability indices of the CS can be found in Level of education: n (%) No schooling 2 (0.2) Supplemental Tables S16 and S17. An exploration of Primary school 46 (5.7) the fit of other potential measurement models can be Secondary school 96 (11.8) seen in Supplemental Table S14. University 667 (82.3) Work status: n (%) Psychological Distress and Mental Wellbeing Student 108 (13.3) Unemployed 44 (5.4) Depression Anxiety Stress Scale-21. The Depression Employed 540 (66.6) Anxiety Stress Scale-21 (DASS-21; Lovibond & Self-employed 32 (4.0) Lovibond, 1995) is a 21-item self-report measure com- Homemaker 9 (1.1) posed of three seven-item subscales that assess depres- On a sick leave 23 (2.8) sion (e.g., ‘‘I felt that I had nothing to look forward to’’), Retired/pensioner 43 (5.3) Unable to work 12 (1.5) anxiety (e.g., ‘‘I felt I was close to panic’’), and stress (e.g., ‘‘I tended to overreact to situations’’). It has a 4- Note. n = frequencies. % = percentages. point Likert-type response format from 0 (‘‘did not apply to me at all’’) to 3 (‘‘applied to me very much or 5-point Likert-type response format ranging from 1 most of the time’’). A total score is computed by sum- (‘‘almost never’’) to 5 (‘‘almost always’’), with higher ming all the items, with a higher score indicating higher scores indicating higher self-compassion levels. The vali- psychological distress (range: 0–63). The validated dated Spanish version of the SCS was used (Garcia- Spanish version of the DASS-21 was used (Bados et al., Campayo et al., 2014). The Spanish SCS version is avail- 2005), which showed good internal consistency in the able in Supplemental Table S11. A previous study pro- present study (composite reliability, v = .94). posed the bifactor exploratory measurement model as the best representation structure for the SCS (Neff et al., Short Warwick–Edinburgh Mental Well-Being Scale. The 2019). The descriptive statistics of the SCS and factor Short Warwick–Edinburgh Mental Well-Being Scale loadings using the bifactor exploratory measurement (SWEMWBS; Stewart-Brown et al., 2009) is a seven-item model can be found in Supplemental Table S12. This measure of mental wellbeing (e.g., ‘‘I’ve been dealing model presented good fit indices for the SCS (CFI = with problems well’’), with a 5-point Likert-type scale .989, TLI = .979, RMSEA = .051; 90% CI = .044– from 1 (‘‘none of the time’’) to 5 (‘‘all of the time’’). The .059, WRMR = .446). The internal consistency of the time frame of the instrument was the last 2 weeks. An general factor was good (vh = .92). Other reliability overall score for the SWEMWBS can be calculated by indices can be found in Supplemental Table S13. An summing the scores of each item (ranging from 7 to 35). exploration of the fit of other potential measurement The items are worded positively, and thus a higher score models can be seen in Supplemental Table S14. indicates higher mental wellbeing levels. Items from the Spanish version of the SWEMWBS (M. A. López et al., Compassion Scale. Permission from the original authors 2013) were used, which showed adequate internal consis- was obtained for translating/validating the Spanish tency values in the present study (v = .84). version of the CS (Pommier et al., 2020). A description of the adaptation process and the CS Spanish version are available in Supplemental Tables S4 and S15. The Data Analysis CS is a 16-item Likert-type scale with a 5-point Factor structure and reliability of the Spanish versions response format ranging from 1 (‘‘almost never’’) to 5 of the SOCSs, SCS, and CS were assessed first Garcı́a-Campayo et al. 7 (Supplemental Figure S1 and Table S3). Then, we com- Table 2. Partial Correlations Between the Two Compassion puted the correlation analyses to explore relationships Operationalizations Controlling for the Sociodemographic between the general factors that measure the different Characteristics of Participants (n = 811). constructs (i.e., self-compassion vs. compassion for oth- SOCS-O CS SOCS-S ers) from the same operationalization, and the relation- ships between those general factors that measure the Measure R p r p r p same construct from the different operationalizations CS .64 \ .001 (Neff’s model vs. SOCS’s model). For that, we used total SOCS-S .29 \ .001 .32 \ .001 factor scores that were calculated by means of bifactor SCS .19 \ .001 .27 \ .001 .78 \ .001 exploratory measurement models (Supple- mental Figure S1) and Pearson’s correlation coefficients, Note. Partial correlation coefficients were calculated by controlling for as well as partial correlations after controlling for the the following sociodemographic characteristics: sex, age, marital status, number of children, education level, and work status. SOCS-O = Sussex- general sociodemographic characteristics (e.g., gender, Oxford Compassion Scale-Others; CS = Compassion Scale (others); age, marital status, number of children, education level, SOCS-S = Sussex-Oxford Compassion Scale-Self; SCS = Self- and work status). Total factor scores were used because Compassion Scale; r = partial correlation coefficient; p = p value. they are better proxies of complex latent variables than raw sum scores (T. D. Brown, 2015). The strengths of the associations were interpreted as follows (Cohen, All the tests were two-sided and were performed with 1988): r = .10 to .29 (small); r = .30 to .49 (medium); a significance level of a \ .05. Nevertheless, given the r = .50 to 1 (large). exploratory nature of this study, we did not correct for To assess the explanatory power of both self- multiple testing but balanced statistical significance con- compassion and compassion for others in accounting sidering the amplitude of the findings (i.e., effect size), for variance in psychological distress and mental well- and regarded them as tentative and hypothesis generat- being, we first calculated Pearson’s correlation coeffi- ing (Feise, 2002). Data analysis was conducted using cients between total factor scores for self-compassion SPSS v26 and Mplus v8.4. and compassion for others, using the different operatio- nalizations, on one hand, and for psychological distress and wellbeing, on the other hand. Second, we calculated Results partial correlations and multivariable linear regressions The raw correlations between the different constructs predicting psychological distress and mental wellbeing (i.e., self-compassion and compassion for others) from using both constructs and Neff’s and SOCS operationa- the same operationalization (i.e., Neff’s and SOCSs) lizations separately controlling for the sociodemo- were small (range: r = .14–.29), while the raw correla- graphic variables. tions between the same constructs from the different We explored potential differences in the association operationalizations were large (range: r = .65–.79), sug- between self-compassion and compassion for others by gesting more convergence in terms of operationaliza- level of psychological distress and wellbeing. For that, tions than regarding constructs (Supplemental Figure the sample was split according to a 16 total score in the S2). The same pattern of relationships was observed DASS-21, that is, the cut-off score for screening anxiety after controlling for the sociodemographic characteris- or major depression (Chin et al., 2019), and a 28 total tics of the participants (Table 2). score in the SWEMWBS, which has been proposed to The raw relationships between compassion for others identify participants with high levels of wellbeing (Ng (using both Neff’s and SOCS operationalizations), and et al., 2017). We described the distribution of self- psychological distress and wellbeing were small although compassion and compassion for others by subgroup significant and in the expected direction (range r = .09– using means (SDs), and calculated effect sizes using .26 in absolute value). The raw relationships between Cohen’s d and 95% confidence intervals. Pearson’s cor- self-compassion (using both Neff’s and SOCS operatio- relations between self-compassion and compassion for nalizations), and psychological distress and wellbeing others were calculated for each subgroup independently were medium to large, and they also were positive and (partial correlations adjusting for the sociodemographic significant (range r = .48 and .59 in absolute value) features of participants also were estimated). Then, (Supplemental Figure S3). However, after controlling a z-value was calculated using the Fisher r-to-z transfor- for the sociodemographic characteristics of participants, mation (Fisher, 1925), and the significance of the the relationships between compassion for others, using difference between correlation coefficients was both Neff’s and SOCS operationalizations, and psycho- computed (Steiger, 1980). logical distress were very small, and they did not reach 8 Assessment 00(0) Table 3. Partial Correlations of Self-Compassion/Compassion self-compassion and compassion for others were small for Others and Psychological Distress/Mental Wellbeing to intermediate in the subgroup of high wellbeing (range Controlling for the Sociodemographic Characteristics of r = .23–.33), and they were significantly larger than in Participants (n = 811). the subgroup of low wellbeing, in which they were null (range r = –.08 to .09) (Supplemental Table S22). An DASS-21 SWEMWBS almost identical pattern was found after controlling for r p R p the sociodemographic features (Table 4). Compassion for others SOCS-O –.01 .909 .10 .005 Discussion CS –.07 .057 .17 \ .001 Compassion for the self This research is innovative in several ways. First, it SOCS-S –.43 \ .001 .54 \ .001 addresses the construct validity and reliability of the two SCS –.52 \ .001 .54 \ .001 main current operationalizations of compassion for the Note. Partial correlation coefficients were calculated by controlling for Spanish population, which has not been done before for the following sociodemographic characteristics: sex, age, marital status, some of them (e.g., SOCS-S, SOCS-O, CS). Second, the number of children, education level, and work status. DASS-21 = study examines the extent to which self-compassion and Depression Anxiety Stress Scale-21; SWEMWBS = Short Warwick– compassion for others converge, and this had not been Edinburgh Mental Well-Being Scale; SOCS-O = Sussex-Oxford explored in previous research using two different opera- Compassion Scale-Others; CS = Compassion Scale (others); SOCS-S = Sussex-Oxford Compassion Scale-Self; SCS = Self-Compassion Scale; r = tionalizations concurrently. Third, the research also Partial correlation coefficient; p = p value. explores the associations between self-compassion and compassion for others in relation to psychological dis- tress and mental wellbeing, and specifically investigates statistical significance (Table 3). Details of a hierarchical whether the relationship between self-compassion and multivariable linear regression analysis predicting psy- compassion for others is influenced by levels of psycho- chological distress and wellbeing using both constructs logical distress and mental wellbeing. This aspect has and Neff’s and SOCS operationalizations separately not been adequately studied in previous research. The after controlling for the sociodemographic variables can use of two different operationalizations of compassion be found in Tables S18–S21. concurrently allows for a comparison of results and an The distribution of the compassion measures by sub- evaluation of the consistency of previous studies. group can be seen in Supplemental Table S22. The high Overall, this research provides new insights into the con- wellbeing subgroup (n = 354), compared with the low struct of compassion, its measurement, and its relation- wellbeing subgroup (n = 457), showed higher scores in ship with mental health outcomes. all the compassion measures, with small-to-medium We explored the relationship between self-compassion effects for the SOCS-O (d = 0.37), medium effects for and compassion for others using two different operatio- the CS (d = 0.53), and large effects for the SOCS-S and nalizations of compassion: Neff’s and the SOCSs models SCS (d = 1.13 and d = 1.15, respectively). The low psy- (Gu et al., 2020; Neff, 2003; Pommier et al., 2020). But chological distress subgroup (n = 538), compared with first, we reproduced all the factorial structures that were the high psychological distress subgroup (n = 273), originally proposed (i.e., one-factor model, correlated showed higher scores in all the compassion measures but factors model, hierarchical model), and also evaluated SOCS-O, with very small effects for the SOCS-O (d = the bifactor model following a recent proposal for the 0.08), small effects for the CS (d = 0.28), and large Neff’s operationalization (Neff et al., 2019), and effects for the SOCS-S and SCS (d = 0.87, and d = extended it to the SOCSs to evaluate its potential unidi- 1.03, respectively). Differences between subgroups were mensionality in more detail. Strong general factors of significantly higher in self-compassion that in compas- self-compassion and compassion for others in both oper- sion for others, using both Neff’s and SOCS operationa- ationalizations were observed, suggesting a possible uni- lizations. The raw correlations between self-compassion dimensional (although multifaceted) nature of these and compassion for others were small to intermediate in constructs, allowing the use and comparison of overall the subgroup of low psychological distress (range r = scores under similar metric conditions. Nevertheless, it .20–.33), and they were significantly larger than in the has been proposed that the extent to which the positive subgroup of high psychological distress, in which they and negative items of the SCS contribute to explain self- were null to small (range r = .02–.14) (Supplemental compassion, and the possibility of different self- Table S22). The same pattern of relationships was found compassion facets being formed by positive and negative after controlling for the sociodemographic characteris- items as reflecting a broader construct, might differ tics (Table 4). The raw correlations between across cultural backgrounds (Montero-Marin et al., Garcı́a-Campayo et al. 9 Table 4. Partial Correlations (Controlling for the Sociodemographic Characteristics of Participants) Between Self-Compassion and Compassion for Others Per Psychological Distress and Wellbeing Subgroup Status, and Differences Between Correlations. CS SOCS-S SCS r p r p r p DASS-21 < 16 (n = 538) SOCS-O .66 .000 .35 .000 .25 .000 CS .38 .000 .30 .000 SOCS-S .76 .000 DASS-21 . 16 (n = 273) SOCS-O .60 .000 .16 .009 .04 .490 CS .12 .051 .09 .176 SOCS-S .70 .000 z p z p z p DASS-21 < 16 vs. DASS-21 . 16 SOCS-Others 1.33 .182 2.73 .006 2.88 .004 CS 3.74 .000 2.93 .003 SOCS-Self 1.72 .085 CS SOCS-S SCS r p r p r p SWEMWBS ø 28 (n = 354) SOCS-O .66 .000 .34 .000 .25 .000 CS .34 .000 .30 .000 SOCS-S .76 .000 SWEMWBS \ 28 (n = 457) SOCS-O .58 .000 .11 .018 -.04 .438 CS .11 .016 .04 .469 SOCS-S .69 .000 z p z p z p SWEMWBS ø 28 vs. SWEMWBS \ 28 SOCS-Others 1.83 .067 3.43 .000 3.03 .002 CS 3.43 .000 3.79 .000 SOCS-Self 2.08 .037 Note. Partial correlation coefficients were calculated by controlling for the following sociodemographic characteristics: sex, age, marital status, number of children, education level, and work status. DASS-21 = Depression Anxiety Stress Scale-21; SWEMWBS = Short Warwick–Edinburgh Mental Well- Being Scale; SOCS-O = Sussex-Oxford Compassion Scale-Others; CS = Compassion Scale (others); SOCS-S = Sussex-Oxford Compassion Scale-Self; SCS = Self-Compassion Scale; r = partial correlation coefficient; z = Steiger’s z for significance of the difference between correlations. 2018). It has been also referred that the use of negative with the previous measures of compassion used. Our items (as it occurs in the SCS, and CS) is not in agree- findings considering the general factors of self- ment with the protective nature of the compassion con- compassion and compassion for others derived from struct, and that a clear understanding of compassion both Neff’s and SOCSs’ operationalizations support the would need to separate the influence of negative compo- hypothesis that self-compassion and compassion for nents to avoid its potential tautological influence (Muris others are different yet related constructs, since the rela- et al., 2016). This was one of the premises of the SOCSs tionships between the scales measuring the same con- operationalizations, which only use positive items. struct (i.e., SOCS-O/CS; SOCS-S/SCS, with large Previous studies have reported significant associa- effects) were larger than the associations found between tions between self-compassion and compassion for oth- the scales of each operationalization (i.e., SOCS-O/ ers, although modest in most cases (Ari et al., 2022; Gu SOCS-S; CS/SCS, with small effects), even after control- et al., 2020; Neff & Pommier, 2013; Pommier et al., ling for general sociodemographic characteristics. 2020; Tendhar et al., 2022), but some studies have found Self-compassion was strongly correlated with both psy- small and no significant relationships (Durkin et al., chological distress and wellbeing, and presented high 2016; Leary et al., 2007; A. López et al., 2018). These explanatory power on these variables. This finding is in inconsistent findings may be at least in part due to issues accordance with previous evidence, since self-compassion 10 Assessment 00(0) has been consistently linked to less psychological distress to display less self-compassion than compassion for others and higher wellbeing (Marsh et al., 2018). The meta- (Neff, 2003), and we observed this was specially so for analysis conducted by Zessin et al. (2015) observed a those with worse mental health. It has been observed that strong association between them, particularly in the cases psychopathology is usually accompanied by low levels of of cognitive and psychological wellbeing. On the other self-compassion (Athanasakou et al., 2020), while self- hand, in our study, compassion for others was not associ- compassion might have a central role in recovery (Waite ated with psychological distress after controlling for the et al., 2015). The observed lack of associations between sociodemographic characteristics of the sample, and self-compassion and compassion for others in individuals although significant correlations with wellbeing were with poor mental health may have psychological and phy- observed, these associations were to a much lesser extent siological underpinnings (Hermanto & Zuroff, 2016), sug- compared with self-compassion. Compassion for others gesting there are probably other relevant variables has not been as widely studied as self-compassion, and to involved in this relationship, such as the potential exis- our knowledge, no meta-analyses have been conducted on tence of a psychiatric disorder, and its nature. The ability its relationships with mental health-related outcomes, such to experience empathy and the tendency to trust or dis- as psychological distress or wellbeing. However, most trust others are key aspects of some disorders (e.g., psy- studies seem to point out that compassion for others is not chosis, autism, antisocial personality disorder), while these significantly correlated with these variables (Beaumont are not necessarily present in other psychopathologies, or et al., 2016; A. López et al., 2018), or not as much as self- in people with psychological distress but no psychopathol- compassion (Tendhar et al., 2022). A cross-cultural study ogy. Such tendency to experience empathy and trust/dis- comparing Japanese and U.S. samples found that self- trust others have been related to the ability to experience compassion was related to positive and negative affect, compassion for others (P. Brown et al., 2020; Singer & social anxiety and wellbeing in both countries, while com- Klimecki, 2014). Thus, it could be hypothesized that while passion for others was associated with only positive affect self-compassion would generally be low in most people and social anxiety symptoms (Arimitsu et al., 2019). A with poor mental health (irrespective of the presence of a recent study aiming to reduce stress and promote well- disorder, and its nature), their clinical profile could play a being in health care workers using a mindfulness-based role in the levels of compassion for others, but this program observed a potential mechanistic role of self- hypothesis should be tested in further studies. compassion in both outcomes, but this was not observed The present work presents three notable strengths: it in the case of compassion for others (Strauss et al., 2021). was conducted on a large general sample of Spanish par- All these results indicate that, as our results suggest and as ticipants; it provides a validated version of the SOCSs hypothesized, self-compassion is more strongly related to and CS to be used in Spanish-speaking clinical/research negative and also positive indicators of mental health than settings; and it evaluated the potential unidimensionality compassion for others. of the constructs under study that came from the differ- We examined whether the relationship between self- ent operationalizations of compassion (i.e., Neff’s and compassion and compassion for others can differ among the SOCSs models) in great detail. Nevertheless, some different levels of psychological distress and wellbeing. As limitations need to be acknowledged. We did not admin- expected, the correlations between compassion for the self ister the translated version of our questionnaire to a and for others were significant and positive, with inter- bilingual sample, alongside the English version. Doing mediate effects in the subgroups of participants with low so would have enhanced confidence that the two mea- psychological distress and/or high wellbeing, but they were sures function similarly. The online assessment that was significantly lower, with small or even absent effects, in conducted for this study could have implied a self- the participants with high psychological distress and/or selection bias (i.e., people without digital literacy, or no low wellbeing. This result indicates that mentally healthy access to the internet could have been excluded) individuals tend to present similar tendencies to apply (Wright, 2005). As the use of Virtual Private Networks compassionate attitudes toward themselves and others, (VPNs) is becoming relatively common, IP addresses indicating a more aggregated motivational structure may not accurately reflect a participant’s legitimacy to (Gilbert, 2022). According to Gilbert’s (1989, 2009, 2022; participate or location, and multiple individuals from Gilbert et al., 2017) social mentality theory, self- the same household (who might be using the same VPN) compassionate attitudes are enabled by having received could appear to have the same IP address, leading to compassion from others through secure attachment rela- inaccuracies in data collection. The cross-sectional tionships, which is in turn associated with reduced stress nature of our study hinders the extraction of conclusions and higher positive affect, including positive psychological based on causality; although the impact of self- wellbeing (Di Bello et al., 2020; Stellar et al., 2015), as compassion on wellbeing and psychological distress can reported in the present study. In general, individuals tend be assumed based on previous randomized controlled Garcı́a-Campayo et al. 11 trials that have proven the effects of compassion-based CRediT Statement interventions on improving these outcomes in different JGC: conceptualization, funding acquisition, supervision and samples (Kirby et al., 2017), the potential causal connec- writing—review and editing. ABS: data curation, visualization tion between presenting low levels of self-compassion and writing—review and editing. DMR: resources, validation and experiencing mental disorders is still to be deter- and writing—review and editing. JN: formal analysis, metho- mined. Finally, participants were not evaluated using a dology and writing—review and editing. APA: data curation, clinical interview, systematic observation, or experimen- visualization and writing—original draft. AFS: supervision tal tasks, and our results are solely based on self- and writing—review and editing. JVL: supervision and writing—review and editing. RB: supervision and writing— reported measures. Future studies should include other review and editing. WK: supervision and writing—review and type of assessments to both contrast the results of our editing. JMM: conceptualization, formal analysis, funding study (i.e., absence of associations between self- acquisition, investigation, methodology, project administra- compassion and compassion for others in people with tion, validation and writing—original draft and review and poor mental health) and to explore the hypothesis that editing. has been raised (i.e., such a relationship being larger in some types of disorders than in others). Data Availability New research confirming the intermediate associa- tion between self-compassion and compassion for oth- This study was not preregistered. The data that support the findings of this study are available from the corresponding ers in the healthy population and the absence of authors upon reasonable request. relationships in specific subgroups suffering from psy- chopathology seems warranted. Future research should investigate whether the potential decoupling Declaration of Conflicting Interests process between self-compassion and compassion for The author(s) declared no potential conflicts of interest with others in psychiatric patients can be reverted, if this is respect to the research, authorship, and/or publication of this possible using different types of interventions (e.g., article. psychotherapies, and drugs), and the extent to which this could be a potential cause or an epiphenomenon Funding of clinical improvements. The author(s) disclosed receipt of the following financial sup- port for the research, authorship, and/or publication of this article: This work was supported by the mental health research Acknowledgments group, (B17_23R) that is part of the Department of The authors thank the University of Zaragoza and the men- Innovation, Research and University at the Government of tal health research group (B17_23R) that is part of the Aragón (Spain). This research was funded in whole, or in part, Department of Innovation, Research and University at the by the Wellcome Trust (WT104908/Z/14/Z and WT107496/Z/ Government of Aragón (Spain), the Institute for Health 15/Z) and supported by the National Institute for Health Research Aragon (IIS Aragon), and the Research Network Research Clinical Research Network (NIHR CRN). The views on Chronicity, Primary Care, and Health Promotion expressed are those of the authors and not necessarily those of (RICAPPS), which received a research grant from the Carlos the U.K. National Health Service, the NIHR, or the U.K. III Institute of Health, Ministry of Science and Innovation Department of Health. The authors were not precluded from (Spain), awarded on the call for the creation of Health accessing data in the study, and they accept responsibility to Outcomes-Oriented Cooperative Research Networks submit it for publication. (RICORS), with reference RD21/0016/0005, as well as Feder Funds ‘‘Another way to make Europe.’’ JMM has a ‘‘Miguel Servet’’ research contract from the ISCIII (CP21/00080). Ethical Approval JMM is grateful to the CIBER of Epidemiology and Public The authors assert that all procedures contributing to this Health (CIBERESP CB22/02/00052; ISCIII) for its support. work comply with the ethical standards of the relevant APA is the holder of a ‘‘Sara Borrell’’ research contract from national and institutional committees on human experimenta- the Institute of Health Carlos III (ISCIII, Spanish Ministry tion and with the Helsinki Declaration of 1964 and its subse- of Science and Innovation, CD20/00181). JN has a research quent amendments. contract from the Institute of Health Carlos III (ISCIII; ICI20/00080). AF-S acknowledges funding from the Serra Húnter program (UAB-LE-8015). ORCID iDs David Martı́nez-Rubio https://orcid.org/0000-0002-9346- 3869 Authors’ Note Adrián Pérez-Aranda https://orcid.org/0000-0002-1602- JVL and JM-M share co-senior authorship. 3413 12 Assessment 00(0) Supplemental Material Reviews, 3(2), 71–100. https://doi.org/10.1177/153458230 Supplemental material for this article is available online. 4267187 Di Bello, M., Carnevali, L., Petrocchi, N., Thayer, J. F., Gil- bert, P., & Ottaviani, C. (2020). The compassionate vagus: References A meta-analysis on the connection between compassion Ari, E., Cesur-Soysal, G., Basran, J., & Gilbert, P. (2022). The and heart rate variability. Neuroscience & Biobehavioral compassionate engagement and action scales for self and Reviews, 116, 21–30. https://doi.org/10.1016/J.NEU- others: Turkish adaptation, validity, and reliability study. BIOREV.2020.06.016 Frontiers in Psychology, 13, Article 780077. https://doi.org/ Durkin, M., Beaumont, E., Martin, C. J. H., & Carson, J. 10.3389/fpsyg.2022.780077 (2016). A pilot study exploring the relationship between Arimitsu, K., Hitokoto, H., Kind, S., & Hofmann, S. G. self-com- passion, self-judgement, self-kindness, compas- (2019). Differences in compassion, well-being, and sion, professional quality of life and wellbeing among UK social anxiety between Japan and the USA. Mindfulness, community nurses. Nurse Education Today, 46, 109–114. 10(5), 854–862. https://doi.org/10.1007/s12671-018- https://doi.org/10.1016/j.nedt.2016.08.030 1045-6 Feise, R. J. (2002). Do multiple outcome measures require Athanasakou, D., Karakasidou, E., Pezirkianidis, C., Lakioti, p-value adjustment? BMC Medical Research Methodol- A., & Stalikas, A. (2020). Self-compassion in clinical sam- ogy, 2(1), Article 8. https://doi.org/10.1186/1471- ples: A systematic literature review. Psychology, 11(2), 2288-2-8 217–244. https://doi.org/10.4236/psych.2020.112015 Feldman, C., & Kuyken, W. (2011). Compassion in the land- Bados, A., Solanas, A., & Andrés, R. (2005). Psychometric scape of suffering. Contemporary Buddhism, 12(1), 143–155. properties of the Spanish version of Depression, Anxiety https://doi.org/10.1080/14639947.2011.564831 and Stress Scales (DASS). Psicothema, 17, 679–683. Feldman, C., & Kuyken, W. (2019). Mindfulness: Ancient wis- Beaumont, E., Durkin, M., Hollins Martin, C. J., & Carson, J. dom meets modern psychology. Guilford Press. (2016). Compassion for others, self-compassion, quality of Fisher, R. A. (1925). Statistical methods for research workers. life and mental well-being measures and their association Oliver and Boyd. with compassion fatigue and burnout in student midwives: Galante, J., Galante, I., Bekkers, M. J., & Gallacher, J. (2014). A quantitative survey. Midwifery, 34, 239–244. https://doi. Effect of kindness-based meditation on health and well- org/10.1016/j.midw.2015.11.002 being: A systematic review and meta-analysis. Journal of Breines, J. G., & Chen, S. (2013). Activating the inner caregiver: Consulting and Clinical Psychology, 82(6), 1101–1114. The role of support-giving schemas in increasing state self- https://doi.org/10.1037/a0037249 compassion. Journal of Experimental Social Psychology, Garcia-Campayo, J., Navarro-Gil, M., Andrés, E., Montero- 49(1), 58–64. https://doi.org/10.1016/j.jesp.2012.07.015 Marin, J., López-Artal, L., & Demarzo, M. M. P. (2014). Brown, P., Waite, F., Rovira, A., Nickless, A., & Freeman, D. Validation of the Spanish versions of the long (26 items) (2020). Virtual reality clinical-experimental tests of compas- and short (12 items) forms of the Self-Compassion Scale sion treatment techniques to reduce paranoia. Scientific (SCS). Health and Quality of Life Outcomes, 12(1), Article Reports, 10(1), 8547. https://doi.org/10.1038/s41598-020- 4. https://doi.org/10.1186/1477-7525-12-4 64957-7 Gilbert, P. (1989). Human nature and suffering. Routledge. Brown, T. D. (2015). Confirmatory factor analysis for applied Gilbert, P. (2009). The compassionate mind: A new approach to research (2nd ed.). Guilford Press. https://doi.org/10.1198/ the challenge of life. Constable & Robinson. tas.2008.s98 Gilbert, P. (2010). Compassion-focused therapy: The CBT dis- Chin, E. G., Buchanan, E. M., Ebesutani, C., & Young, J. tinctive features series. Routledge. (2019). Depression, anxiety, and stress: How should clini- Gilbert, P. (2014). The origins and nature of compassion cians interpret the total and subscale scores of the 21-item focused therapy. British Journal of Clinical Psychology, depression, anxiety, and stress scales? Psychological 53(1), 6–41. https://doi.org/10.1111/bjc.12043 Reports, 122(4), 1550–1575. https://doi.org/10.1177/003329 Gilbert, P. (2022). An evolutionary and compassion-based 4118783508 approach to yearning and grief. In D. L. Harris & A. H. Y. Cohen, J. (1988). Statistical power analysis for the behavioral Ho (Eds.), Compassion-based approaches in loss and grief sciences. Lawrence Erlbaum. (1st ed. pp. 1–10). Routledge. Darwin, C. (1871). The descent of man, and selection in relation Gilbert, P., Catarino, F., Duarte, C., Matos, M., Kolts, R., to sex. John Murray. Stubbs, J., Ceresatto, L., Duarte, J., Pinto-Gouveia, J., & Davis, M. H. (1983). Measuring individual differences in Basran, J. (2017). The development of compassionate empathy: Evidence for a multidimensional approach. Jour- engagement and action scales for self and others. Journal of nal of Personality and Social Psychology, 44(1), 113–126. Compassionate Health Care, 4, Article 4. https://doi.org/ https://doi.org/10.1037/0022-3514.44.1.113 10.1186/s40639-017-0033-3 De Waal, F. B. M. (2009). The age of empathy: Nature’s lessons Goetz, J. L., Keltner, D., & Simon-Thomas, E. (2010). Com- for a kinder society. Souvenir Press. passion: An evolutionary analysis and empirical review. Decety, J., & Jackson, P. L. (2004). The functional architecture Psychological Bulletin, 136(3), 351–374. https://doi.org/ of human empathy. Behavioral and Cognitive Neuroscience 10.1037/a0018807 Garcı́a-Campayo et al. 13 Gu, J., Baer, R., Cavanagh, K., Kuyken, W., & Strauss, C. Lovibond, S., & Lovibond, P. (1995). Manual for the Depres- (2020). Development and psychometric properties of the sion Anxiety Stress Scales. Psychology Foundation of Sussex-Oxford Compassion Scales (SOCS). Assessment, Australia. 27(1), 3–20. https://doi.org/10.1177/1073191119860911 MacCallum, R. C., Browne, M. W., & Sugawara, H. M. Gu, J., Cavanagh, K., Baer, R., & Strauss, C. (2017). An (1996). Power analysis and determination of sample size for empirical examination of the factor structure of compas- covariance structure modeling. Psychological Methods, sion. PLOS ONE, 12(2), Article e0172471. https://doi.org/ 1(2), 130–149. https://doi.org/10.1037/1082-989X.1.2.130 10.1371/journal.pone.0172471 Marsh, I. C., Chan, S. W. Y., & MacBeth, A. (2018). Self- Hermanto, N., & Zuroff, D. C. (2016). The social mentality compassion and psychological distress in adolescents—A theory of self-compassion and self-reassurance: The interac- meta-analysis. Mindfulness, 9(4), 1011–1027. https://doi. tive effect of care-seeking and caregiving. The Journal of org/10.1007/S12671-017-0850-7 Social Psychology, 156(5), 523–535. https://doi.org/ Mongrain, M., Chin, J. M., & Shapira, L. B. (2011). Practicing 10.1080/00224545.2015.1135779 compassion increases happiness and self-esteem. Journal of Hofmann, S. G., Grossman, P., & Hinton, D. E. (2011). Happiness Studies, 12(6), 963–981. https://doi.org/10.1007/ Loving-kindness and compassion meditation: Potential for s10902-010-9239-1 psychological interventions. Clinical Psychology Review, Montero-Marı́n, J., Gaete, J., Demarzo, M., Rodero, B., 31(7), 1126–1132. https://doi.org/10.1016/j.cpr.2011.07.003 Lopez, L. C., & Garcı́a-Campayo, J. (2016). Self-criticism: Jahoda, A., Stenfert Kroese, B., Pert, C., Jahoda, A., Stenfert A measure of uncompassionate behaviors toward the self, Kroese, B., & Pert, C. (2017). Mindfulness and third wave based on the negative components of the Self-Compassion therapies. In Cognitive behaviour therapy for people with Scale. Frontiers in Psychology, 7, Article 1281. https://doi. intellectual disabilities (pp. 181–212). Palgrave Macmillan. org/10.3389/fpsyg.2016.01281 https://doi.org/10.1057/978-1-137-47854-2_9 Montero-Marin, J., Kuyken, W., Crane, C., Gu, J., Baer, R., Jazaieri, H., Jinpa, G. T., McGonigal, K., Rosenberg, E. L., Al-Awamleh, A. A., Akutsu, S., Araya-Véliz, C., Ghorbani, Finkelstein, J., Simon-Thomas, E., & Goldin, P. R. (2013). N., Chen, Z. J., Kim, M. S., Mantzios, M., Rolim Dos San- Enhancing compassion: A randomized controlled trial of a tos, D. N., Serramo López, L. C., Teleb, A. A., Watson, P. compassion cultivation training program. Journal of Happi- J., Yamaguchi, A., Yang, E., & Garcı́a-Campayo, J. (2018). ness Studies, 14(4), 1113–1126. https://doi.org/10.1007/ Self-compassion and cultural values: A cross-cultural study s10902-012-9373-z of self-compassion using a Multitrait-Multimethod Keltner, D., & Kogan, A. (2018). Compassionate love: A (MTMM) analytical procedure. Frontiers in Psychology, 9, framework for research. Emotion Review, 10(2), 112–121. Article 2638. https://doi.org/10.3389/fpsyg.2018.02638 https://doi.org/10.1177/1754073917749619 Morin, A. J. S., Katrin Arens, A., & Marsh, H. W. (2016). A Kirby, J. N., Tellegen, C. L., & Steindl, S. R. (2017). A meta- bifactor exploratory structural equation modeling frame- analysis of compassion-based interventions: Current state work for the identification of distinct sources of construct- of knowledge and future directions. Behavior Therapy, 48, relevant psychometric multidimensionality. Structural 778–792. https://doi.org/10.1016/j.beth.2017.06.003 Equation Modeling, 23(1), 116–139. https://doi.org/10.1080/ Leary, M. R., Tate, E. B., Adams, C. E., Allen, A. B., & Han- 10705511.2014.961800 cock, J. (2007). Self-compassion and reactions to unplea- Muris, P., Otgaar, H., & Petrocchi, N. (2016). Protection as sant self-relevant events: The implications of treating the mirror image of psychopathology: Further critical notes oneself kindly. Journal of Personality and Social Psychol- on the Self-Compassion Scale. Mindfulness, 7(3), 787–790. ogy, 92(5), 887–904. https://doi.org/10.1037/0022-3514.92. https://doi.org/10.1007/s12671-016-0509-9 5.887 Navarro-Gil, M., Lopez-del-Hoyo, Y., Modrego-Alarcón, M., Longe, O., Maratos, F. A., Gilbert, P., Evans, G., Volker, F., Montero-Marin, J., Van Gordon, W., Shonin, E., & Rockliff, H., & Rippon, G. (2010). Having a word with Garcia-Campayo, J. (2020). Effects of Attachment- yourself: Neural correlates of self-criticism and self- Based Compassion Therapy (ABCT) on self-compassion reassurance. NeuroImage, 49(2), 1849–1856. https:// and attachment style in healthy people. Mindfulness, doi.org/10.1016/j.neuroimage.2009.09.019 11(1), 51–62. https://doi.org/10.1007/s12671-018-0896-1 López, A., Sanderman, R., Ranchor, A. V., & Schroevers, M. Neff, K. D. (2003). The development and validation of a scale J. (2018). Compassion for others and self-compassion: Lev- to measure self-compassion. Self and Identity, 2, 223–250. els, correlates, and relationship with psychological well- https://doi.org/10.1080/15298860390209035 being. Mindfulness, 9(1), 325–331. https://doi.org/10.1007/ Neff, K. D. (2016). The Self-Compassion Scale is a valid and s12671-017-0777-z theoretically coherent measure of self-compassion: Erra- López, M. A., Gabilondo, A., Codony, M., Garcı́a-Forero, C., tum. Mindfulness, 7(4), 1009. https://doi.org/10.1007/ Vilagut, G., Castellvı́, P., & Alonso, J. (2013). Adaptation s12671-016-0560-6 into Spanish of the Warwick-Edinburgh Mental Well-being Neff, K. D., & Germer, C. K. (2013). A pilot study and rando- Scale (WEMWBS) and preliminary validation in a student mized controlled trial of the mindful self-compassion pro- sample. Quality of Life Research, 22(5), 1099–1104. https:// gram. Journal of Clinical Psychology, 69(1), 28–44. https:// doi.org/10.1007/s11136-012-0238-z doi.org/10.1002/jclp.21923 14 Assessment 00(0) Neff, K. D., & Pommier, E. (2013). The relationship between (WEMWBS): A Rasch analysis using data from the Scot- self-compassion and other-focused concern among college tish Health Education Population Survey. Health and Qual- undergraduates, community adults, and practicing medita- ity of Life Outcomes, 7(1), Article 15. https://doi.org/10. tors. Self and Identity, 12(2), 160–176. https://doi.org/10. 1186/1477-7525-7-15 1080/15298868.2011.649546 Strauss, C., Gu, J., Montero-Marin, J., Whittington, A., Chap- Neff, K. D., Tóth-Király, I., Yarnell, L. M., Arimitsu, K., Cas- man, C., & Kuyken, W. (2021). Reducing stress and pro- tilho, P., Ghorbani, N., & Mantzios, M. (2019). Examining moting well-being in healthcare workers using mindfulness- the factor structure of the Self-Compassion Scale in 20 based cognitive therapy for life. International Journal of diverse samples: Support for use of a total score and six Clinical and Health Psychology, 21(2), 100227. https://doi. subscale scores. Psychological Assessment, 31(1), 27–45. org/10.1016/j.ijchp.2021.100227 https://doi.org/10.1037/pas0000629 Strauss, C., Lever Taylor, B., Gu, J., Kuyken, W., Baer, R., Ng Fat, L., Scholes, S., Boniface, S., Mindell, J., & Stewart- Jones, F., & Cavanagh, K. (2016). What is compassion and Brown, S. (2017). Evaluating and establishing national how can we measure it? A review of definitions and mea- norms for mental wellbeing using the short Warwick– sures. Clinical Psychology Review, 47, 15–27. https://doi. Edinburgh Mental Well-being Scale (SWEMWBS): org/10.1016/j.cpr.2016.05.004 Findings from the Health Survey for England. Quality of Tendhar, T., Marcotte, M. A., Saikia, M. J., & de Mesquita, P. Life Research, 26(5), 1129–1144. https://doi.org/10.1007/ B. (2022). Relationship of compassion for self and others to s11136-016-1454-8 sense of well-being of college students. Journal of American Nyanaponika, T., & Bodhi, B. (1999). The four sublime states: College Health. Advance online publication. https://doi. Contemplations on love, compassion, sympathetic joy, and org/10.1080/07448481.2022.2074276 equanimity. Wisdom Publications. Van Dam, N. T., Sheppard, S. C., Forsyth, J. P., & Earleywine, Pommier, E., Neff, K. D., & Tóth-Király, I. (2020). The devel- M. (2011). Self-compassion is a better predictor than mindful- opment and validation of the Compassion Scale. Assess- ness of symptom severity and quality of life in mixed anxiety ment, 27(1), 21–39. https://doi.org/10.1177/10731911 and depression. Journal of Anxiety Disorders, 25(1), 123–130. 19874108 https://doi.org/10.1016/j.janxdis.2010.08.011 Singer, T., & Klimecki, O. M. (2014). Empathy and compas- Van der Kolk, B. (2014). The body keeps the score: Brain, mind, sion. Current Biology, 24(18), R875–R878. https://doi.org/ and body in the healing of trauma. Penguin Books. 10.1016/J.CUB.2014.06.054 Waite, F., Knight, M. T., & Lee, D. (2015). Self-compassion Singer, T., & Lamm, C. (2009). The social neuroscience of and self-criticism in recovery in psychosis: An interpretative empathy. Annals of the New York Academy of Sciences, phenomenological analysis study. Journal of Clinical Psy- 1156, 81–96. https://doi.org/10.1111/j.1749-6632.2009. chology, 71(12), 1201–1217. https://doi.org/10.1002/jclp. 04418.x 22211 Steiger, J. H. (1980). Tests for comparing elements of a correla- Wright, K. B. (2005). Researching internet-based populations: tion matrix. Psychological Bulletin, 87(2), 245–251. https:// Advantages and disadvantages of online survey research, doi.org/10.1037/0033-2909.87.2.245 online questionnaire authoring software packages, and web Stellar, J. E., Cohen, A., Oveis, C., & Keltner, D. (2015). survey services. Journal of Computer-Mediated Communica- Affective and physiological responses to the suffering of tion, 10(3), JCMC1034. https://doi.org/10.1111/j.1083- others: Compassion and vagal activity. Journal of Personal- 6101.2005.tb00259.x ity and Social Psychology, 108(4), 572–585. https://doi.org/ Zessin, U., Dickhäuser, O., & Garbade, S. (2015). The rela- 10.1037/pspi0000010 tionship between self-compassion and well-being: A meta- Stewart-Brown, S., Tennant, A., Tennant, R., Platt, S., Parkin- analysis. Applied Psychology: Health and Well-Being, 7(3), son, J., & Weich, S. (2009). Internal construct validity of 340–364. https://doi.org/10.1111/aphw.12051 the Warwick-Edinburgh Mental Well-Being Scale