Self-Compassion Mediates the Relationships Between University Students’ Mindfulness, Dysfunctional Attitudes, and Various Distress and Well-Being Indicators

: Objective : The current study intended to model the relationships between mindfulness, dysfunctional attitudes, and self-compassion in predicting university students’ various indicators of emotional distress and well-being of university students. We aimed to examine the mediative role of self-compassion and the mediative roles of self-coldness and self-warmth in these relationships. Methods : Applying a correlational design, validated instruments were used to measure mindfulness, dysfunctional attitudes, self-compassion (and its two main components: self-warmth and self-coldness), various indicators of emotional distress (i.e., negative affect, depression, anxiety, stress), and well-being (i.e., soothing positive affect, activating positive affect, satisfaction with life). The final sample consisted of N = 176 university students. The proposed models were tested by SEM (structural equation modeling) using SPSS AMOS 20. Results : The main results obtained were consistent with the hypotheses. As expected, higher levels of mindfulness led to higher well-being and lower emotional distress through increased self-compassion, and higher levels of dysfunctional attitudes led to lower well-being and higher emotional distress through decreased self-compassion. However, the results also indicated that self-coldness was more important than self-warmth in all of these relationships.


Introduction
Undergraduate students experience reduced levels of well-being and increased psychological distress compared to the general community (Bore et al., 2016;Larcombe et al., 2016;Regehr et al., 2013); therefore, it is essential to design effective interventions for them to reduce their distress and improve their well-being.
In general, studies have not consistently shown significant differences in the effectiveness of self-compassion interventions compared to other active approaches such as mindfulness or cognitive restructuring practices (Arimitsu & Hofmann, 2017;Mak et al., 2018;Preuss et al., 2021).However, some findings suggest that self-compassion interventions may be superior in terms of their effectiveness (Javidi et al., 2021) or their acceptability and applicability, particularly for students (Ca ndea & Szenta gotai-Ta tar, 2018), therefore, their adaptation and application in academic settings may be recommended.
Although multiple forms of interventions can be effective, targeting the mentioned constructs, it is advisable to explore the relationships between self-compassion, mindfulness, and various cognitive vulnerabilities (such as dysfunctional attitudes) in predicting various indicators of distress and well-being among university students.This exploration can help identify the most opportune points for intervention from a theoretical perspective.Although many studies have separately examined the relationship between self-compassion and mindfulness, as well as between selfcompassion and dysfunctional attitudes, using different models (Ferrari et al., 2018;Li et al., 2022;Liu et al., 2022;Makadi & Koszycki, 2020;Phillips et al., 2018;Podina et al., 2015;Sedighimornani et al., 2019;Wong & Mak, 2013;Xavier et al., 2023), few studies have simultaneously explored these relationships (i.e., the relationships between cognitive vulnerabilities, mindfulness and self-compassion) in predicting mental health.
In some models examining the relationships between self-compassion, cognitive vulnerabilities and mental health, self-compassion has been proposed as a moderator between cognitive vulnerabilities and distress.There is a growing body of studies on the buffering effects of self-compassion on the relationship between different cognitive vulnerabilities and different indicators of distress and well-being (Ferrari et al., 2018;Fonseca & Canavarro, 2017;Li et al., 2022;Phillips et al., 2018;Podina et al., 2015;Wong & Mak, 2013).The results of these studies are promising, suggesting that self-compassion may reduce the detrimental effects of cognitive vulnerabilities (e.g., maladaptive perfectionism, irrational beliefs, implicit dysfunctional attitudes).
However, other studies (Hassani et al., 2021;Liu et al., 2022;Xavier et al., 2023), found that self-compassion mediated rather than moderated this relationship (i.e., cognitive vulnerabilities such as dysfunctional attitudes led to lower levels of selfcompassion, which resulted in higher levels of distress).For example, in a longitudinal study, Liu et al. (2022) found that one of the most common dysfunctional attitudes, negative perfectionism, was a risk factor for depression through the negative component of self-compassion (i.e., self-coldness), and positive perfectionism was a protective factor against depression via the positive component of self-compassion (i.e., self-warmth).
Studies that examine the relationship between mindfulness and self-compassion have shown that self-compassion mediates the relationship between mindfulness and various indicators of distress and well-being, such as social anxiety (Makadi & Koszycki, 2020), shame (Sedighimornani et al., 2019), recovery from mental disorders (Mak et al., 2021), and subjective well-being (Yang et al., 2022).However, Mak et al. (2021) found that only self-warmth mediates the relationship between mindfulness and personal recovery (self-coldness did not).
Thus, it appears that different components of self-compassion (i.e., self-warmth and self-coldness) may play different roles in these associations between dysfunctional attitudes and clinical outcomes and between mindfulness and clinical outcomes (Liu et al., 2022;Mak et al., 2021).The results of meta-analyses (Chio et al., 2021;Muris & Petrocchi, 2016) have also highlighted the importance of separating self-compassionate responses and uncompassionate responses toward the self, the importance of distinguishing self-warmth from self-coldness.Uncompassionate responses (overidentification, isolation, self-judgment) are more strongly related to distress indicators than compassionate responses, but compassionate responses (mindfulness, common humanity, self-kindness) may be more important for well-being than selfcoldness.
The results of mindfulness interventions have also highlighted that they are beneficial not only through increasing mindfulness, but also through increasing selfcompassion (i.e., self-compassion is an important mechanism mediating the effects of the interventions) (Baer, 2003;Duarte & Pinto-Gouveia, 2017;Evans, et al., 2018;Keng, et al., 2016).Bergen-Cico and Cheon (2014) investigated the sequence of changes in meditation practices and found that an increase in mindfulness leads to an increase in selfcompassion, concluding that in line with Neff's theory of self-compassion (2003b), mindfulness precedes self-compassion, therefore mindfulness skills are important for being able to cultivate self-compassion.
To the best of the author's knowledge, few studies have explored these relationships together (i.e., the relationships between cognitive vulnerabilities, mindfulness, and self-compassion).Thimm (2017), for example, examined the relationship between early maladaptive schemas (another well-established cognitive vulnerability), self-compassion, mindfulness, and psychological distress.Their results showed that self-compassion and mindfulness mediated (but did not moderate) the relationship between early maladaptive schemas and psychological distress, thus both mindfulness and self-compassion were found to be mediators.
Based on previous results, in this study, we propose and test a model (Figure 1) for the relationships between dysfunctional attitudes, mindfulness, and self-compassion in predicting various psychological distress (i.e., negative affect, depression, anxiety, stress) and well-being indicators (i.e., soothing positive affect, activating positive affect, satisfaction with life) among university students.This investigation can help identify the most opportune points for intervention to reduce their distress and improve their wellbeing.We hypothesized that mindfulness and dysfunctional attitudes influence university students' various psychological distress indicators (i.e., negative affect, depression, anxiety, and stress) and various well-being indicators (i.e., soothing positive affect, activating positive affect, and satisfaction with life) through self-compassion.Thus, we considered that mindfulness increases well-being and decreases distress by improving self-compassion, while dysfunctional attitudes lead to higher levels of psychological distress and lower well-being by reducing self-compassion.
Based on the results presented on different components of self-compassion (Chio et al., 2021;Liu et al., 2022;Mak et al., 2021;Muris & Petrocchi, 2016) and based on Gilbert's theory of compassion and emotion regulation systems (2009a, 2009b, 2014), another goal was to investigate which predictor variables (i.e., mindfulness and dysfunctional attitudes) affect clinical variables through which component of selfcompassion (i.e., self-warmth or self-coldness).We also tested the second model presented in Figure 2 to achieve this goal.
For the second model, we hypothesized that mindfulness would affect clinical outcomes among students, especially through self-warmth, as well as dysfunctional attitudes, especially through self-coldness.We also hypothesized that in predicting various indicators of distress (i.e., negative affectivity, depression, anxiety, stress), selfcoldness would have greater predictive power than self-warmth, while in predicting various indicators of well-being (i.e., soothing positive affect, activating positive affect, satisfaction with life), self-warmth would have greater predictive power than selfcoldness.

Participants
Taking into account that the first model proposed requires the estimation of 10 distinct parameters and the second requires the estimation of 13 distinct parameters, we needed at least 130 participants to test the models (Collier, 2020).In the end, 181 participants (university students) completed the questionnaires and, after preliminary analyses, five participants were excluded.Further analyses were based on the data of the remaining 176 participants.The majority of the participants were female (n = 140; 79,5%), from Romania (n = 130; 73,9%).Half of the participants studied psychology (n = 87; 49,4%), and half studied in other fields of study (n = 89; 50,6%).The mean age of the participants was 26,06 years (SD = 11,17).

Instruments a) Mindfulness
Mindfulness was assessed using the Five Facet Mindfulness Questionnaire (FFMQ; Baer, et al., 2006), a widely used instrument to measure different components of mindfulness (i.e., observation, description, action with awareness, non-judgment of inner experience, and non-reactivity).This scale consists of 39 items (e.g., "I can perceive emotions without reacting to them"; "I am aware of bodily sensations when I take a bath"), which participants rate on a five-point Likert scale (1 -never; 5 -always true).For this study, the global mindfulness score (i.e., the mean scores of these subscales) was used.
Higher mindfulness scores indicated higher levels of trait mindfulness.For the global mindfulness indicator, the scale had acceptable internal consistency (α = .776).

b) Self-Compassion
Self-compassion was assessed using the Self-Compassion Scale -Short Form (SCS-SF; Raes et al., 2011), a 12-item version of the original Self-Compassion Scale (SCS; Neff, 2003a).The SCS-SF measures each of the components of self-compassion (i.e., selfkindness, self-judgment, common humanity, isolation, mindfulness, over-identification) with two items (e.g., "I try to see my failings as part of the human condition").Responses are to be given on a five-point Likert scale.The six items that measure the negative dimensions of self-compassion are reverse coded.Scores for self-compassion were calculated by averaging the scores on items measuring self-compassionate behaviors (i.e., self-kindness, common humanity, mindfulness) and reverse coded scores on items measuring uncompassionate behaviors towards the self (i.e., self-judgment, isolation, over-identification).The scale showed good internal consistency for self-compassion as a global indicator (α = .822)and for self-coldness (α = .816),but the self-warmth subscale had questionable internal consistency (α = .628).

Depression, Anxiety, and Stress
Levels of depression, anxiety, and stress were assessed using the 21-item version of the Depression Anxiety Stress Scale (DASS-21; Lovibond & Lovibond, 1995).
Participants rated from 0 (does not apply to me at all) to 3 (applies to me very much, or most of the time) how often they usually experience each symptom.The variable scores were calculated by summing the seven items measuring stress, the seven items measuring depression, and the seven items measuring anxiety.The score for each subscale was then multiplied by 2. Higher scores indicate higher levels of depression, anxiety, and stress.

c) Negative Affect
To measure the subjective dimension of distress (i.e., negative affect), we used the abbreviated Hungarian version of the Emotional Distress Profile (Profilul Distresului Emot ional -PDE; Opris & Macavei, 2005).The scale was originally developed and validated in Romania and has good psychometric properties and excellent internal consistency (α = .94)as a complex indicator of emotional distress.
The original scale consists of 26 adjectives describing negative affects, such as "sad," and "depressed".In our study, we used 12 items that had adequate face validity according to the translation.On a five-point Likert scale, participants were asked to rate the extent to which the given affective items were typical of their experiences in the past two weeks.The scale also had excellent internal consistency (α = .924)in measuring the negative affect in the present sample.

Soothing and Activating Positive Affects
Different types of positive affect (i.e., soothing and activating positive affects) were assessed using the Types of Positive Affect Scale (Gilbert et al., 2008), which consists of 18 items and measures three different types of positive affect (i.e., soothing-, relaxing-, and activating positive affect), rated by participants between (1 -Not characteristic of me) and (5 -Very characteristic of me).The scale measures how frequently participants experience these feelings.The variable scores were calculated by summing the dedicated items.For this study, soothing-and activating positive affect were measured.The subscale measuring soothing positive affect showed acceptable (α = .757)and the subscale measuring activating positive affect showed good internal consistency (α = .886).The results are similar to those of the original English instrument, in that the internal consistency of the activating positive affect subscale was higher (α = .83)than that of the subscale measuring safeness/contentment positive affect (α = .73).

d) Dysfunctional Attitudes
The Dysfunctional Attitude Scale (DAS; Weissman & Beck, 1980) is a self-report scale designed to measure the presence and intensity of dysfunctional attitudes.The Hungarian version of the DAS (Kopp M., 1994) consists of 35 items (i.e., five items for each of the seven types of dysfunctional attitudes: need for approval, need for love, need for achievement, perfectionism, entitlement, omnipotence, and autonomy), rated on a 5point Likert scale (-2 = strongly disagree; 2 = strongly agree).For example, one item for measuring the need for approval is: "I need other people's approval in order to be happy".
For this study, a global indicator of dysfunctional attitudes was calculated by summing the scores of the individual items.The scale had excellent internal consistency in measuring dysfunctional attitudes (α = .912).

e) Life Satisfaction
The five-item Hungarian version of the Satisfaction with Life Scale (SWLS; Diener et al., 1985;Martos et al., 2014) measured the agreement with statements (e.g."In most ways my life is close to ideal"), ranging from 1 (strongly disagree) to 7 (strongly agree).
Variable scores were calculated by summing the items.The SWLS has shown high internal consistency, test-retest reliability, and validity (Diener et al., 1985).The current study also confirmed good internal consistency (α = .825).

Procedure and Design
This study used a correlational design and measured two predictors (i.e., dysfunctional attitudes and mindfulness), three mediators (self-compassion and separately self-coldness and self-warmth), and seven outcome variables.Of the seven outcome variables, four were used to operationalize emotional distress (i.e., negative affect, depression, anxiety, and stress) and three were used to operationalize psychological well-being (i.e., soothing positive affect, activating positive affect, and satisfaction with life).
The study was conducted in accordance with the Code of Ethics of the American Psychological Association.Following informed consent, students completed the questionnaire using an online platform (Google Forms).The study was advertised in Introduction to Psychology classes.Students were not rewarded for participating in the study.

Data Analyses and Assessment of Model Fit
SPSS 20 software was used for the preliminary analyses.Pearson correlations were performed to examine the associations between dysfunctional attitudes, mindfulness, self-compassion, and clinical outcomes.
The proposed model was tested using SEM (Structural Equation Modeling) in SPSS AMOS 20 software and Maximum Likelihood (ML) estimation was chosen.We used the bootstrap method to test for direct and indirect effects, generating 5000 samples (95% confidence interval).Effects were considered significant if the confidence intervals of the bootstrap analysis did not include zero (Hayes, 2018;Preacher & Hayes, 2004).
Model fit was assessed using the ratio of the chi-square statistic (CMIN) to the degrees of freedom (DF), standardized root mean square residual (SRMR), comparative fit index (CFI) and general fit index (GFI).
For the chi-square statistic and the degrees of freedom ratio, critical values between 2 and 5 have been recommended as cut-off values (Hu & Bentler, 1999).The CFI values should not be lower than .90,but for a good fit, the CFI values should be above .95(Hu & Bentler, 1999).For the GFI, .95indicates a good fit, while values higher than .90indicate an acceptable fit.For the SRMR, a value less than .08 is considered acceptable and less than .05 is considered a good fit (Schermelleh-Engel, Moosbrugger, & Mu ller, 2003).

Preliminary Analyses
The descriptive statistics of the measured variables (means, standard deviations, skewness, kurtosis, and Cronbach's α values) are presented in Table 1.Data were normally distributed and multivariate normality was tested and confirmed in all cases.First-order correlations between variables are presented in Table 2.

Structural Equation Modeling
The First Model -Self-

Compassion Mediates the Effects
To analyze the fit of the first model to the data, in which self-compassion was proposed as a mediator of the relationships between mindfulness and clinical outcomes, as well as between dysfunctional attitudes and clinical outcomes, a series of (seven) structural equation modeling tests were conducted for different clinical outcomes (four for emotional distress and three for well-being).In all cases, we found that selfcompassion was a significant mediator (i.e., the indirect effects of mindfulness and dysfunctional attitudes were significant in all cases).Consistent with our hypotheses, mindfulness, and dysfunctional attitudes affected all psychological distress indicators (i.e., negative affect, depression, anxiety, and stress) and all well-being indicators (i.e., soothing positive affect, activating positive affect, and satisfaction with life) through selfcompassion.In other words, higher levels of mindfulness led to improved well-being and reduced distress through improvements in self-compassion, and higher levels of dysfunctional attitudes led to higher levels of psychological distress and lower levels of well-being through reductions in self-compassion.Figure 3 shows the first model for negative affect with standardized regression weights.Both mindfulness (ß = .44,p < .01)and dysfunctional attitudes (ß = -.37,p < .01)predicted self-compassion (see Table 3).Self-compassion had a direct effect on all clinical outcomes (Table 2).Self-compassion was a negative predictor of distress indicators: negative affect (ß = -.546,p < .01),depression (ß = -.562,p < .01),anxiety (ß = -.399,p < .01),and stress (ß = -.566,p < .01);and was a positive predictor of well-being: soothing positive affect (ß = .446,p < .01),activating positive affect (ß = .332,p < .01),and satisfaction with life (ß = .384,p < .01).Indirect effects for mindfulness and dysfunctional attitudes were also significant in all cases.We also examined the model fit of the first model for all clinical outcomes (Table 4).The fit indices (GFI and CFI) indicated a good model fit for most outcomes, and the standardized root mean square residual (SRMR) also indicated a low error rate.However, for depression, CFI and SRMR were outside the acceptable range.

The Second Model -Self-Coldness Mediates the Effects
To analyze the fit of the second model to the data, in which self-coldness and selfwarmth were separately proposed as mediators of the relationships between mindfulness and clinical outcomes, and between dysfunctional attitudes and clinical outcomes, we also performed a series of structural equation modelings for different clinical outcomes.The indirect effects of mindfulness and dysfunctional attitudes were significant in all cases (Table 5).Consistent with our hypothesis, dysfunctional attitudes had a direct effect only on selfcoldness (ß = .504,p < .01),but not on self-warmth (ß = -.082,p > .05).However, contrary to our hypothesis, mindfulness was a significant predictor for both, a positive predictor of self-warmth (ß = .399,p < .01),and a negative predictor of self-coldness (ß = -.383,p < .01).The direct effects of mindfulness and dysfunctional attitudes on self-warmth and on self-coldness are shown in Figure 4 and Table 6.Testing the effects of self-warmth and self-coldness separately on the different outcomes, the results showed that self-coldness was more relevant in all cases (not only for indicators of emotional distress).The direct effects of self-warmth on the different indicators of well-being and distress are presented in Table 6.Self-warmth had no significant effect on the distress indicators: neither negative affect (ß = -.018,p > .05),nor depression (ß = .000,p > .05),nor anxiety (ß = .020,p > .05),nor stress (ß = -.089,p > .05).
We found a small effect of self-warmth on soothing positive affect (ß = .161,p < .05)and activating positive affect (ß = .176,p < .05),but based on the bootstrapping method the effect on soothing positive affect was not significant.Self-warmth also did not affect satisfaction with life (ß = .152,p > .05).
Assessing the model fit of the second model for all clinical outcomes (Table 7), we found that the fit indicators (GFI and CFI) indicated a good fit to the model for all outcomes (including depression), and the standardized root mean square residual (SRMR) also indicated a low level of error.

DISCUSSION
The present study aimed to test the fit of two proposed models of the relationship between dysfunctional attitudes, mindfulness, self-compassion (its two components), and various indicators of university students' distress and well-being.In the first model, the mediative role of self-compassion was tested for the relationship between mindfulness and various clinical outcomes (i.e., various indicators of emotional distress: negative affect, depression, stress, and anxiety; and various indicators of well-being: soothing positive affect, activating positive affect, and life satisfaction), and for the relationship between dysfunctional attitudes and these clinical outcomes.In the second model, we separately examined the mediation role of self-warmth (i.e., mean scores for the positive components of self-compassion: mindfulness, common humanity, and selfkindness) and self-coldness (i.e., mean scores for the negative components of selfcompassion: over-identification, isolation, and self-judgment) in these relationships.
Reinforcing previous results (Hassani et al., 2021;Mak et al., 2021;Makadi & Koszycki, 2020;Liu et al., 2022;Sedighimornani et al., 2019;Xavier et al., 2023;Yang et al., 2022), our hypotheses for the first model were confirmed, which means that mindfulness led to an increase in students' well-being and a decrease in their emotional distress through self-compassion, and dysfunctional attitudes led to a decrease in wellbeing and an increase in emotional distress through self-compassion.In most cases, the first model provided an excellent fit to the data, with depression being an exception.When we investigated which of the relationships not represented in the model might increase the model fit, we found that dysfunctional attitudes not only lead to depression through self-compassion but also directly affect it.This is not surprising given that the Dysfunctional Attitudes Scale (Weissman & Beck, 1978) was developed specifically to measure dysfunctional attitudes in depression.
Based on Gilbert's theory (2009aGilbert's theory ( , 2009bGilbert's theory ( , 2014) ) and findings on the relative importance of self-coldness and self-warmth in predicting indicators of distress and wellbeing (Chio et al., 2021;Liu et al., 2022;Mak et al., 2021;Muris & Petrocchi, 2016), in the second model we tested the mediative role of these two components of self-compassion (i.e., self-coldness and self-warmth) separately.This model showed a good fit for the data for all clinical outcomes (including depression).
Our hypothesis that dysfunctional attitudes would influence clinical outcomes, especially through self-coldness, based on the results of Liu et al. (2022), was confirmed.
We found a direct effect of dysfunctional attitudes only on self-coldness (not on selfwarmth) and significant indirect effects on all outcomes.This implies that dysfunctional attitudes lead to a decrease in students' well-being and an increase in their emotional distress through self-coldness.
Our unexpected findings that mindfulness had an indirect effect on clinical outcomes (including well-being indicators), in many cases only through self-coldness, can be explained by the fact that self-warmth has no direct effect on these clinical indicators (mindfulness had a direct effect on both self-warmth and self-coldness).The exception was activating positive affect, which means that, for activating positive affectivity, the influence of mindfulness was mediated by both components of self-compassion, however, for the other six outcomes (including distress indicators and other well-being indicators), only self-coldness mediated the effect of mindfulness.Indeed, in the findings of Mak et al.(2021),only self-warmth mediated the relationship between mindfulness and personal recovery from mental problems, but mindfulness had a direct effect on both self-coldness and self-warmth, as in our study.
Based on the results of meta-analyses (Chio et al., 2021;Muris & Petrocchi, 2016), we expected that self-coldness would have greater predictive power than self-warmth in predicting various indicators of distress (i.e., negative affectivity, depression, anxiety, stress), and that self-warmth would have greater predictive power than self-coldness in predicting various indicators of well-being (i.e., soothing positive affect, activating positive affect, satisfaction with life).However, our results showed that when we control for covariance between the two components when testing the relations in a single model, self-coldness also has greater significance in indicators of well-being.If we were to base our conclusions only on the correlation test (Table 1), we would draw completely different conclusions.Although in our model, self-warmth did not affect any of the distress indicators (i.e., negative affect, depression, anxiety, stress), the correlations between them were significant in all cases.Furthermore, based on correlation analyses alone, the relationships between self-warmth and well-being indicators were similar to the relationships between self-coldness and well-being indicators.
This makes sense given the results of the current meta-analysis (Chio et al., 2021), which found that although self-kindness (r = .39)was more strongly associated with wellbeing than self-judgment (r = -.29), and mindfulness (r = .39)was more strongly associated with well-being than over-identification (r = -.32), common humanity (r = .29)had a significantly weaker relationship with well-being (both eudaimonic and hedonic well-being) than isolation (r = -.36).The effect size of the relationship between selfwarmth and the well-being (r = .38)and effect size of the relationship between selfcoldness (r = -.36) and well-being were significantly different, but the difference was small.Based on these results, we conclude that it is very important to examine the relative importance of self-coldness and self-warmth in models in which the individual effects of each component can be tested while controlling for the effects of the other.
Overall, these results highlight the importance of self-compassion (especially selfcoldness) regarding students' mental health.Based on these findings, the implementation of self-compassion interventions in academic context would be a major step towards supporting students' well-being and reducing their distress, however, randomized controlled trials are needed to test the effectiveness of these interventions.The most wellestablished programs designed to cultivate self-compassion include the Mindful Self-Compassion Program (MSC), developed and tested by Neff and Germer (2012) based on Neff's (2003aNeff's ( , 2003bNeff's ( , 2023aNeff's ( , 2023b) ) (2009a, 2009b, 2009c, 2023) biopsychosocial and evolutionary approach to selfcompassion, the Compassion Focused Therapy (CFT) model.Both are group-based resource-building training programs spanning eight weeks, and both incorporate a blend of written exercises, imaginative practices, meditation, and body-based activities.There is increasing evidence to support the effectiveness of these programs (e.g., Germer & Neff, 2019;Irons & Heriot-Maitland, 2021;Matos et al., 2017;Neff & Germer, 2012), even for university students (e.g., Beaumont et al., 2021;Smeets et al., 2014).
Although generally, studies have not revealed significant differences in the effectiveness of self-compassion interventions when compared with alternative approaches, such as cognitive restructuring (see, e.g., Arimitsu & Hofmann, 2017), some findings indicate the potential superiority of self-compassion interventions, based on their effectiveness (Javidi et al., 2021), or their acceptability and applicability for students (Ca ndea & Szenta gotai-Ta tar, 2018).Therefore, the adaptation of these interventions is highly recommended.
Despite the significance of these results, our study has some limitations.Firstly, although these models have good to excellent fit to the data, the model fit may be overestimated due to low degrees of freedom (df = 2; Collier, 2020).Secondly, our model could be further elaborated by taking into account six different components of selfcompassion (i.e., mindfulness, common humanity, self-kindness, over-identification, isolation, self-judgment), instead of only addressing the negative (self-coldness) and the positive (self-warmth) components.Further research has been proposed which focuses on these specific components, especially based on the results cited by Chio et al. (2021).
Considering the psychometric properties of the six subscales, the original Self-Compassion Scale (Neff, 2003a) is recommended for this purpose due to the low internal consistency of the scale's abbreviated form (Raes et al., 2011).
For further research, it is recommended to consider the different facets of mindfulness (i.e., observation, description, acting with awareness, non-judgment of inner experience, and non-reactivity) and the different types of dysfunctional attitudes in these relationships (i.e., need for approval, need for love, need for achievement, perfectionism, entitlement, omnipotence, and autonomy).The findings of previous studies suggest that different aspects of mindfulness may have different degrees of influence on student distress and well-being (Bodenlos et al., 2015), and that mindfulness interventions may improve these with varying levels of effectiveness (Quaglia et al. & Brown, 2016).Thus, it may be important to consider these separately in these models.
The correlational design also restricts our understanding of these relationships.The results of this study also highlight the importance of examining the effects of self-coldness and self-warmth on emotional distress and well-being in complex models using structural equation modeling (not just correlations) in which the individual effects of each component can be tested while controlling for the effect of the other.

Figure 1 .
Figure 1.The Proposed Model

Figure 2 .
Figure 2. The Proposed Model Accounting for Different Components of Self-Compassion

Figure. 3
Figure. 3 Predicting negative affect mediated by self-compassion

Figure 4 .
Figure 4. Predicting negative affect through self-warmth and self-coldness Without a longitudinal perspective, potentially reversed pathways or alternative explanations remain unaddressed.Therefore, it is recommended to further investigate self-compassion as a mediator in these relationships in a longitudinal design, in particular, based on the results ofMaxwell et al. (2011), which show that cross-sectional analyses can indicate the existence of a significant indirect effect, even when the true longitudinal indirect effect is zero.Randomized controlled trials are also recommended to test the effectiveness of the most well-established programs designed to cultivate selfcompassion for improving students' mental health.butnone of the studies identified examined self-compassion interventions, therefore, the introduction and investigation of these types of interventions is highly recommended during tertiary education.
Standards a) Conflict of Interest -No conflict of interest is associated with this publication.b) Ethical Standards -The study was undertaken in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.c) Informed Consent -All participants gave their informed consent prior to their inclusion in the study.d) Funding -No funding was received to assist with the preparation of this manuscript.

Table 3 .
Standardized direct and indirect effects for the first model

Table 4 .
The model fit of the proposed model for different clinical outcomes

Table 5 .
Standardized indirect effects for the second model Notes: ** The regression is significant at the .01level; CI Confidence Interval; LL Lower Limit; UL Upper Limit

Table 6 .
Standardized direct effects for the second model Notes: ** The regression is significant at the .01level; * The regression is significant at the .05level; CI Confidence Interval; LL Lower Limit; UL Upper Limit

Table 7 .
The model fit of the second model with different components of self-compassion for conceptualization of self-compassion, and the Compassionate Mind Training (CMT; Irons & Heriot-Maitland, 2021), based on Gilbert's