Pursuing a college education in late adolescence and early adulthood is a critical life stage marked by major physical and psychological changes. This stage is frequently associated with elevated stress levels among students. Reference Barbayannis, Bandari, Zheng, Baquerizo, Pecor and Ming1 Importantly, student stress is multifactorial, Reference Karyotaki, Cuijpers, Albor, Alonso, Auerbach and Bantjes2 shaped by a range of influences, including rigorous academic demands, intensive study requirements, time management difficulties, peer competition, financial constraints, familial expectations and the need to adapt to unfamiliar environments. Reference Pascoe, Hetrick and Parker3,Reference Freire, Ferradás, Regueiro, Rodríguez, Valle and Núñez4
The cumulative effect of these stressors can significantly undermine students’ motivation and academic performance, thereby increasing drop-out rates. Reference Pascoe, Hetrick and Parker3 Furthermore, empirical studies have consistently demonstrated that such stressors are strong predictors of mental disorders, including depression, anxiety and substance use disorders, throughout the college years. Reference Barbayannis, Bandari, Zheng, Baquerizo, Pecor and Ming1
In response to these challenges, there is growing scholarly interest in mindfulness-based interventions as a means of mitigating stress and promoting psychological well-being among college students. These approaches offer promising avenues for enhancing resilience and supporting mental health in academic settings. Reference Kral, Davis, Korponay, Hirshberg, Hoel and Tello5 For instance, a meta-analysis of 21 neuroimaging studies involving 300 participants showed 132 morphological changes following mindfulness exercises. Reference Fox, Nijeboer, Dixon, Floman, Ellamil and Rumak6 In particular, there were changes in key areas for metacognition, extra- and intra-corporeal awareness, memory consolidation, reflection and reconsolidation, self-regulation and emotion regulation, as well as intra- and inter-hemispheric communication.
The influence of self-compassion on psychological distress
Mindfulness serves as a foundational element for cultivating a compassion-focused approach and is considered a prerequisite for the development of self-compassion. Self-compassion is a multidimensional construct encompassing cognitive, emotional and behavioural components. It involves three key elements: self-kindness as opposed to self-judgement, recognition of shared human experiences rather than feelings of isolation, and mindfulness rather than over-identification with negative emotions. Reference Davoudi Moghaddam, Saiedmanesh and Choobforoushzadeh7 Moreover, self-compassion contributes to the formation of positive intrapersonal and interpersonal relationships, offering a supportive and protective framework that enhances the development, persistence, stability and advancement of perceived mental health benefits. Reference Tiwari, Pandey, Rai, Pandey, Verma and Parihar8 Over the past few decades, a growing body of research has consistently identified self-compassion as a significant protective factor and a critical psychological variable influencing students’ mental health. Reference Kotera, Andrzejewski, Dosedlova, Taylor, Edwards and Blackmore9
Similarly, other studies have shown the positive relationship and influence of the two variables of mindfulness and self-compassion on mental health.
Mindfulness-based self-compassion
Given the close relationship between mindfulness and self-compassion, the concept of mindfulness-based self-compassion has emerged as a distinct framework. This approach encourages individuals to engage with painful emotions and experiences without avoidance or suppression. Instead, it promotes a conscious and accepting awareness of one’s difficulties, fostering a compassionate stance towards oneself. Within this framework, individuals are encouraged to observe their personal shortcomings and challenges with clarity and kindness, thereby cultivating a balanced and supportive mental perspective. Reference Davoudi Moghaddam, Saiedmanesh and Choobforoushzadeh7 For example, there is evidence of a significant positive relationship between mindfulness, self-compassion and mental health in students, establishing these as strong predictors of psychological well-being. Reference Chaudhuri, Dema, Wangmo and Gautam10 In another study, mindfulness and self-compassion collectively predicted 70% of the changes related to negative emotions and depressive symptoms in the general population. Reference López, Sanderman and Schroevers11
Research has consistently demonstrated that conscious self-compassion is a significant positive contributor to mental health and overall quality of life, particularly among individuals facing critical or challenging life circumstances. For example, the literature suggests that conscious self-compassion coupled with fostering emotion regulation empowers individuals to attain higher levels of mental well-being and an enhanced quality of life. Reference Ten Hoopen, de Nijs, Duvekot, Greaves-Lord, Hillegers and Brouwer12
Cognitive regulation of emotion is a specific form of self-regulation that refers to the capacity to monitor, evaluate, understand and modify emotional reactions in a manner that is beneficial for normal functioning. It involves a process whereby a person autonomously organises their emotions in order to be in the present consciously or unconsciously while adjusting responses according to environmental demands. Reference Gratz and Roemer13
Maladaptive emotion regulation strategies are related to negative emotions, negative self-belief and emotional events (such as blaming oneself or others, rumination (mental) and catastrophising), and focusing on adaptive emotion regulation strategies (such as acceptance, refocusing, reappraisal and taking a broader view) improves people’s understanding of emotion management. Reference Dadfarnia, Hadianfard, Rahimi and Aflakseir14
In summary, research into college students’ mental health indicates its critical impact on students’ academic performance. There is therefore a need to know the factors involved and appropriate interventions. In this regard, self-compassion, mindfulness and cognitive emotion regulation may all be relevant factors.
Method
We investigated the relationship between university students’ mindfulness and self-compassion and their mental health using structural modelling. We also studied whether the cognitive regulation of emotion (adaptive and maladaptive) mediated this relationship.
Research design
This study adopts an applied research approach using a descriptive correlational method. The target population consisted of all young adults enrolled in full-time study at Azad University in Tehran (Iran) during the academic year 2022–2023, selected through available sampling techniques. Out of the 350 distributed questionnaires, responses were obtained from 246 students (70%); structural equation modelling necessitates a sample size of at least 200 individuals. Inclusion criteria encompassed individuals aged over 19 years, willing to provide informed consent, and free from psychiatric or medical conditions requiring medication at the start of the study.
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2013. All procedures involving human subjects/patients were approved by the human ethics committee of Allameh Tabataba’i University in Tehran (approval number 225/247231). Prior to participation, all individuals were presented with a consent form containing comprehensive explanations regarding the study’s objectives and procedures. Participants were assured of the confidentiality of their personal information and were informed of their right to withdraw from the research at any point.
Measures
Mindful Attention Awareness Scale
The Mindful Attention Awareness Scale (MAAS) is a 15-item questionnaire that measures awareness and attention to current events and experiences in daily life. Reference Brown and Ryan15 The items measure the construct of mindfulness on a 6-point Likert scale (from 1 for ‘almost always’ to 6 for ‘almost never’). This scale provides an overall score for mindfulness ranging from 15 to 90, where a higher scores indicates greater mindfulness. Cronbach’s alpha coefficients for the scale’s internal homogeneity are between 0.80 and 0.87, and the test–retest coefficient has been reported to be constant over 1 month. This scale shows strong psychometric properties and has been validated in academic and community samples.
Correlational, quasi-experimental and experimental studies have shown that the MAAS taps unique qualities of mindfulness that are associated with, and predictive of, a variety of self-regulation and well-being constructs. Reference Brown and Ryan15 The psychometric validity of the Farsi version of the MAAS in a sample of Iranian students showed acceptable levels of discriminant validity. Reference Nooripour, Ghanbari, Hoseinian, Vakili and Dobkins16 In the present study, the Cronbach’s alpha was 0.73.
Self-Compassion Scale
The Self-Compassion Scale contains 26 items to measure the following six characteristics: self-kindness (5 items); self-judgement (5 items); common humanity (4 items) versus isolation (4 items) and mindfulness (4 items) versus over-identification (4 items). Reference Neff17 Items are scored on a 5-point Likert scale, from 1 (‘almost never’) to 5 (‘almost always’), where a higher score indicates greater self-compassion. The overall validity of the scale was confirmed by a Cronbach’s alpha of 0.92. The subscales also show good internal consistency (0.75–0.81). In the psychometric analysis of the Farsi version of this scale for students, the range of Cronbach’s alpha for the subscales was 0.68–0.77, the Cronbach’s alpha for the whole scale was 0.90, the range of test–retest coefficients was 0.71–0.56 and the range of correlation between items was 0.54–0.78, which indicated the favourable reliability of the scale. Reference Hasani and Pasdar18 Confirmatory factor analysis indicated an acceptable fit of the items. In addition, the correlation coefficients between subscales ranged from 0.32 to 0.65, and the scale showed good content, divergent, convergent and criterion validity. Reference Hasani and Pasdar18 In the present study, Cronbach’s alpha was 0.78 for the total score on this scale and 0.70–0.84 for the subscales.
Cognitive Emotion Regulation Questionnaire
The Cognitive Emotion Regulation Questionnaire (CERQ) evaluates the way people think after experiencing life-threatening or stressful events. Reference Garnefski, Kraaij and Spinhoven19 The questionnaire has 36 items, which are rated on a 5-point Likert scale, from 1 (‘never’) to 5 (‘always’). The items conceptually form 9 different subscales that each represent specific styles of cognitive emotion regulation. Negative styles include blaming oneself or others, rumination (mental) and catastrophising; positive styles include acceptance, refocusing, reappraisal and taking a broader more positive view. The positive and negative styles subscales and the questionnaire itself had good reliability (Cronbach’s alpha of 0.91, 0.87 and 0.93 respectively). Reference Garnefski, Kraaij and Spinhoven19 Validation of the Farsi form of this questionnaire reported acceptable Cronbach’s alpha and test–retest coefficients. Reference Samani and Sadeghi20 In the present study, Cronbach’s alpha for the total scores on the positive and negative styles subscales were 0.79 and 0.84 respectively.
General Health Questionnaire
The General Health Questionnaire (GHQ) has 60 questions, although there are also shortened forms of 30, 28 and 12 questions. The 28-question form was used in the present study, with scoring on a 4-point Likert scale (from 0 to 3), with a maximum potential score of 84. In all options, as well as the total score, low scores indicate good mental health and high scores indicate a problem in the person’s mental health. Many studies have confirmed the validity and reliability of this questionnaire. A study of the psychometric properties of the Farsi version of this questionnaire for students demonstrated acceptable validity and reliability. Reference Taghavi21 In the present study, a Cronbach’s alpha of 0.79 confirmed the reliability of this questionnaire.
Statistical analysis
To check the integrity of our data analysis, we conducted normality tests such as the Shapiro–Wilk and Kolmogorov–Smirnov tests on our data (p > 0.05 indicating a normal distribution). These tests are essential as they help us understand the distribution of our data and make informed decisions about the statistical techniques to apply. The Shapiro–Wilk test is particularly useful for smaller samples, Reference Sayar and Anilan22 whereas the Kolmogorov–Smirnov test is more suitable for larger samples. Reference Govindasamy, Isa and Mohamed23 Both tests allowed us to reject the null hypothesis of normality, indicating that our data deviate significantly from a normal distribution, which would necessitate different analytical approaches (Supplementary Table 1, available at https://doi.org/10.1192/bji.2025.10074). We therefore used variance-based partial least squares structural equation modelling (PLS-SEM), which does not require the normality assumption. Reference Hair, Sarstedt, Ringle, Noor, Ma and Olmos24
The first step in analysis is to assess the internal consistency of all constructs and subconstructs (items). The loading of the items on their respective constructs assesses the reliability of the items: Items with factor loadings below 0.50 were removed from the analysis. The internal consistency of constructs is measured using the composite reliability and Cronbach’s alpha. The minimum acceptable level of these indicators is 0.6 and values >0.95 are a sign of redundancy of items.
Discriminant validity was initially verified by examining the cross-loading matrix so that no item loaded more highly on another construct than on the construct it is meant to measure. Next, the average variance shared between a construct and its measures was assessed using the average variance extracted (AVE). A construct should share more variance with its measures than it shares with other constructs in the model. Reference Hair, Hult, Ringle and Hair25 Discriminant validity is adequate if the square root of the AVE is significantly greater than the correlation coefficient. Finally, PLS-SEM was used to examine the direct and indirect relationships between participants’ mindfulness, self-compassion, adaptive and maladaptive cognitive emotion regulation strategies and their mental health. The analysis was conducted with a 95% confidence level and evaluated how constructs influenced each other, focusing on the potential mediating roles of self-compassion and cognitive emotion regulation.
Results
Demographic characteristics
Of the 246 participating young adults in full-time study, 158 (64.2%) identified as female, 76 (30.9%) as male and 12 (4.9%) did not specify their gender. In terms of age, 18 students (7.3%) were <20 years old, 50 (20.3%) were 21–25 years old, 57 (23.2%) were 26–30 years old, 28 (11.4%) were 31–35 years old, 32 (13%) were 36–40 years old, 52 (21.1%) were >40 years old and 9 (3.7%) did not give their age. The mean age of the total sample was 30.91 years (s.d. = 7.05); for females the mean was 32.18 years (s.d. = 6.65) and for males 28.93 years (s.d. = 7.70). Seventy per cent (n = 269) were enrolled on a Bachelor’s degree programme, and the remainder were on Diploma, Masters or Doctorate programmes.
Data characteristics
First, we assessed the internal consistency of all constructs and subconstructs (items) using the composite reliability and Cronbach’s alpha. Supplementary Table 2 shows that the level of reliability for some constructs did not satisfy acceptable thresholds (values in the range 0.6–0.95 are considered acceptable). We therefore conducted a model modification process and checked for convergent validity (Table 1). This was determined from the reliability of individual items and the value of the AVE. Reference Hair, Hult, Ringle and Hair25 Factor loadings for all indicators were above 0.6, which was well above the minimum threshold of 0.5 (Table 1). Reference Hair, Risher and Sarstedt26 The AVE for all constructs was above 50%, which indicates that the second target, of assessing convergent validity, was obtained. As the last step in determining the quality of the measurement model, we assessed the discriminant (also called the divergent) validity of the constructs. The objective of the discriminant validity assessment was to confirm that a reflective construct has stronger connections with its own indicators rather than with those of any other construct in the PLS path model. We used the Fornell–Larcker criterion, in which the square root of the AVE for each construct was compared with the correlation of that construct with other constructs of the model. Reference Hair, Hult, Ringle and Hair25 Table 2 shows that this criterion was achieved for all constructs.
Measurement model test results (modified model)
AVE, average variance extracted; CER, cognitive emotion regulation; PosReap, positive reappraisal; PosRefc, positive refocusing; PutPers, putting into pespective; RefPL, refocus on planning; Cat, catastrophising; OthBl, blaming others; Rumn, rumination; SlfBlm, self-blame; GH, general health; SomSymp, somatic symptoms; Sc, subscale; AnxIns, anxiety and insomnia; Dep, depression; MA, mindful awareness; SC, self-compassion; SlfJug, self-judgement; OvrIdfc, over-identification.
Discriminant validity of the model’s constructs
CER, cognitive emotion regulation. Bold indicates statistically significant correlations.
Table 3 shows the results of the structural model analysis, which examined the relationships between the structures of the research model. For each of the relationships, the value of the path coefficient is presented as an indicator of the intensity of the effect: if the t-value is >1.96, the relationship is significant. In such a situation, the significance level will be <5%. The standard deviations in this table also show how, in the bootstrapping steps, the alternative data scattered for the current findings. The lower the value of this index, the higher the confidence in the corresponding finding.
Results of path coefficient testing
CER, cognitive emotion regulation.
The PLS-SEM findings (Fig. 1) revealed significant positive direct effects of mindfulness on both self-compassion (t = 4.96, p < 0.05, r = 0.332) and mental health (t = 5.969, p < 0.05, r = 0.385), indicating that higher levels of mindfulness are associated with increased self-compassion and improved mental health. Similarly, self-compassion demonstrated positive direct effects on adaptive cognitive emotion regulation (t = 5.101, p < 0.05, r = 0.341) and mental health (t = 3.515, p < 0.05, r = 0.219), suggesting that individuals with greater self-compassion tend to use more adaptive emotion regulation strategies and experience better mental health outcomes. In contrast, self-compassion exhibited a negative direct effect on maladaptive cognitive emotion regulation (t = 2.860, p < 0.05, r = −0.231), implying that higher self-compassion is linked to less use of maladaptive emotion regulation.
The partial least squares model. Mindful awareness is exogeneous. CER, cognitive emotion regulation; MA, mindful awareness; SC, self-compassion; Sc, subscale; GH, general health; PosReap, positive reappraisal; PosRefc, positive refocusing; PutPers, putting into perspective; RefPL, refocus on planning; SomSymp, somatic symptoms; AnxIns, anxiety and insomnia; Dep, depression; MalAdaptiv, maladaptive; Cat, catastrophising; OthBl, blaming others; Rumn, rumination; SlfBlm, self-blame; SlfJug, self-judgement; OvrIdfc, over-identification.
Adaptive cognitive emotion regulation positively predicted mental health (t = 2.382, p < 0.05, r = 0.153), highlighting the importance of adaptive coping mechanisms for psychological well-being. Mindfulness also showed a negative direct effect on maladaptive cognitive emotion regulation (t = 2.449, p < 0.05, r = −0.229), suggesting that more mindful individuals are less likely to rely on maladaptive emotion regulation strategies. Given these direct effects, post hoc analysis suggested a potential serial mediation pathway where mindfulness influences mental health indirectly through self-compassion and adaptive cognitive emotion regulation. However, the analysis did not support the hypothesised relationships between maladaptive cognitive emotion regulation and mental health (t = 0.599, p > 0.05) or between mindfulness and adaptive cognitive emotion regulation (t = 1.568, p > 0.05).
Discussion
Main findings
Although the relationship between mindfulness and self-compassion has been studied in the past, Reference Neff, Dahm, Ostafin, Robinson and Meier27,Reference Bluth and Blanton28 the present study adds to the literature by establishing the mediating role of cognitive emotion regulation on mental health in Iranian university students. In this regard, our study shows that there were positive and significant relationships between participants’ mindfulness, self-compassion and adaptive cognitive emotion regulation and their mental health. Also, there were negative and significant relationships between mindfulness and self-compassion and maladaptive cognitive emotion regulation. However, there was no significant relationship between maladaptive cognitive emotion regulation and mental health. By contrast, adaptive cognitive emotion regulation did mediate the relationship between self-compassion and mental health (p < 0.05).
Interpretation of our findings and comparison with existing literature
Our results are consistent with studies that were restricted to individual or constituent parts of the pathway, such as the effect of mindfulness on reducing maladaptive cognitive emotion regulation. Reference Al-Refae, Al-Refae, Munroe, Sardella and Ferrari29–Reference Keng, Smoski, Robins, Ekblad and Brantley33 Other examples include the use of self-compassion in enhancing adaptive cognitive emotion regulation, Reference Al-Refae, Al-Refae, Munroe, Sardella and Ferrari29,Reference Doorley, Kashdan, Weppner and Glass34 as well as the positive influence of adaptive cognitive emotion regulation on students’ mental health. Reference Al-Refae, Al-Refae, Munroe, Sardella and Ferrari29,Reference Tran, Vo-Thanh, Soliman, Khoury and Chau35 Similarly, others have shown a positive relationship between mindfulness and mental health, Reference Yuan, Sun, Zhao, Liu and Liang36 as well as self-compassion and mental health. Reference Kotera, Andrzejewski, Dosedlova, Taylor, Edwards and Blackmore9,Reference Min, Jianchao and Mengyuan37 Finally, as in this research, a review of five studies showed that emotion regulation significantly mediated the relationship between self-compassion and mental health. Reference Inwood and Ferrari38
Unlike some work, Reference Abdi, Babapoor and Fathi39,Reference Habibzadeh, Beyki and Porzoor40 we did not find a relationship between mindfulness and adaptive cognitive emotion regulation. This disparity could stem from gender distribution differences across studies. In our research, females outnumbered males by nearly two to one (64.2% females, 30.9% males, 4.9% unknown), whereas the two other studies had a more balanced gender distribution. Reference Abdi, Babapoor and Fathi39,Reference Habibzadeh, Beyki and Porzoor40 Given findings that females employed adaptive emotion regulation more frequently than males but still reported worse general health, Reference Abdi, Babapoor and Fathi39 gender composition may have influenced the outcomes of the present study. This is consistent with other work suggesting gender differences in emotion regulation strategies. Reference Nolen-Hoeksema and Aldao41,Reference Goubet and Chrysikou42 It is also possible that mindfulness may influence adaptive emotion regulation indirectly, through increased self-compassion, rather than directly. Reference Nolen-Hoeksema and Aldao41,Reference Goubet and Chrysikou42 Additionally, previous studies employed Pearson correlation and regression analyses, whereas our study used PLS regression. Similarly, we found no significant relationship between maladaptive cognitive emotion regulation and mental health. One explanation is that there is, in fact, a bidirectional relationship between the two, with the former being both a predictor and outcome of psychological distress. Reference Dawel, Shou and Gulliver43 As a result, maladaptive strategies may not always show a clear causal path to poor mental health, especially in non-clinical populations. Reference Dawel, Shou and Gulliver43
Mindfulness
As mentioned above, our study found that high mindfulness levels can reduce maladaptive emotion regulation strategies. Mindfulness involves two key components: presence (conscious attention to the present moment) and non-judgement (a non-evaluative stance towards mental events). Reference Iani, Lauriola and Chiesa44 These components help individuals remain connected to their experiences without avoidance. Presence does not eliminate planning for the future, and non-judgement does not mean accepting everything. Instead, mindfulness fosters flexibility and adaptability in response to emotions. This interactive relationship, known as the dialectic of mindfulness, helps individuals maintain calm and emotional stability, preventing impulsive negative reactions and reducing the use of maladaptive strategies. Reference Wheeler, Arnkoff and Glass45
Emotion regulation systems
Neurophysiological research identifies three main emotion regulation systems: threat, stimulus and relief. Reference Depue and Morrone-Strupinsky46 The threat system triggers anxiety and anger, prompting protective behaviours. Reference Depue and Morrone-Strupinsky46 The stimulus system is goal-oriented, eliciting pleasure-driven behaviours. The soothing system focuses on safety, reducing distress through nurturing and affection. Reference Gilbert47 Maladaptive cognitive emotion regulation strategies are linked to the threat and stimulus systems, whereas adaptive strategies are associated with the soothing system. Balanced emotion regulation systems are crucial for mental health, but imbalance can lead to distress and psychopathology, often due to an overactive threat system and reduced soothing system activity, including less self-compassion. Reference Inwood and Ferrari38
High threat system activation can increase the prevalence of mental illness. However, this can be attenuated by self-compassion, which involves kindness to oneself, awareness of painful thoughts and recognising shared human struggles. Self-compassion fosters cognitive and behavioural flexibility, non-judgement, forgiveness, love, self-respect and a positive self-image. It helps manage emotions adaptively, avoiding denial, isolation or aggression, thereby improving mental health. Cognitive emotion regulation strategies can also activate positive emotions and mitigate the negative effects of an overstimulated threat system. Reference Savarimuthu, Mariya Joseph, Irulandi and Ibrahim48 These adaptive strategies help balance goal achievement with intellectual, emotional, behavioural and personality integrity. They require rational cognitive activity to evaluate events and solve problems effectively, enhancing logical thinking, reasoning and evaluation.
Limitations
This study has several limitations. First, the sampling was not random and was limited to students at a single university in Tehran. There were insufficient numbers to include differences by gender, field of study or level of education as moderating variables in the models, thereby limiting the generalisability of our findings. Future research should include the consideration of these and other moderating variables, as well as extension to other educational institutions.
Clinical implications
Our findings suggest that interventions to promote mindfulness, self-compassion and the use of positive cognitive emotion regulation may improve the mental health of young adults in full-time study. For instance, a cognitive intervention based on mindfulness and self-compassion was reported to be effective in increasing self-compassion and emotion regulation in a general population sample of people over 18 years of age. Reference Al-Refae, Al-Refae, Munroe, Sardella and Ferrari29
Supplementary material
The supplementary material is available online at https://doi.org/10.1192/bji.2025.10074
Data availability
The data that support the findings of this study are available from the corresponding author (S.K.) on reasonable request. The data are not publicly available owing to restrictions, such as contact information that could compromise the privacy of research participants.
Author contributions
B.A. had the original idea for the study. M.S. conducted the study and collected the data. M.A. carried out the data analysis and wrote the first draft of the paper, which was critically revised for important intellectual content by all other authors. S.K. then wrote the final submitted version of the paper.
Funding
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Declaration of interests
S.K. is a member of the BJPsych International editorial board and did not take part in the review or decision-making process of this paper.
Open access
eLetters
No eLetters have been published for this article.