The Role of Mindfulness, Self‐Compassion, and Emotion Regulation in Eating Disorder Symptoms Among College Students

Today's college students experience greater stress than did past generations (Brunner, Wallace, Reymann, Sellers, & McCabe, 2014; Chugani, 2017; Morgan, 2017). In addition to adjusting to the academic stress of college and the changes associated with emerging adulthood, the current generation of students is characterized by emotional struggles related to an increased drive for achievement and the need to conform (Brunner et al., 2014). The emotional stress related to the college experience can negatively affect health‐related practices (Thomas & Borrayo, 2016) and increase the risk of psychological disorders (Brunner et al., 2014). Mental health issues such as depression, suicidal ideation, anxiety, sleep problems, and disturbed eating behaviors are prevalent among college students (Brunner et al., 2014; Lenz, Taylor, Fleming, & Serman, 2014; Morgan, 2017). Disturbed eating practices, such as dietary restraint and disinhibited eating, which can precipitate eating disorders, are often associated with poor psychological health (Lenz et al., 2014).

Eating disorders often emerge, resurface, or worsen in individuals during college (National Eating Disorders Association, 2013). In a study examining eating disorder prevalence over a 13‐year period among undergraduate students at one university, researchers found that rates of disordered eating behavior increased from 23.4% to 32.6% in women and from 7.9% to 25% in men (White, Reynolds‐Malear, & Cordero, 2011). Although there is little information on prevalence among graduate students, Parker, Lyons, and Bonner (2005) found that 20% of their graduate student sample (n = 297; 72.4% female and 27.6% male) was diagnosed with eating disorders. Regardless of the impacted population, eating disorders are severe psychiatric disorders and take the lives of more than 5% of individuals for every decade they continue to be ill (“Disordered Eating in Midlife and Beyond,” 2012). Moreover, many college students may experience clinically deleterious distress yet not meet the full diagnostic criteria for a specific eating disorder (Schwitzer & Choate, 2015).

Although researchers have extensively studied eating disorder symptoms and interventions, much of the research has focused on variables that negatively affect body image and eating attitudes and behaviors. Therefore, more recent work has recommended correlational and causal examination of protective factors that buffer the effects of poor body image and disordered eating (Braun, Park, & Gorin, 2016). Cognitive factors, such as mindfulness, self‐compassion, and emotion regulation, have established associations with the development and course of eating disorders (Dijkstra & Barelds, 2011; Ferreira, Pinto‐Gouveia, & Duarte, 2013; Harrison, Sullivan, Tchanturia, & Treasure, 2010). However, it is important to better understand how these constructs may, directly and indirectly, disrupt eating disorder symptoms to aid in more successful psychotherapeutic outcomes for at‐risk individuals (Lenz et al., 2014).

Tested Model

This study sought to empirically examine and integrate theoretical frameworks of mindfulness, self‐compassion, emotion regulation, and eating disorders, as outlined by Germer (2009); Lavender, Gratz, and Tull (2011); and MacBeth and Gumley (2012). Germer proposed that a reciprocal relationship exists between self‐compassion and mindfulness, such that self‐compassion requires mindful awareness of personal suffering combined with an attitude of nonjudgment and acceptance. The work of Lavender et al. demonstrated a significant relationship between the construct of mindfulness and eating pathology and showed that the facets of acting with awareness, nonreactivity to inner experiences, and nonjudgment of inner experiences were significantly negatively correlated with eating pathology. Last, MacBeth and Gumley conducted a meta‐analysis and found a large effect size for the relationship between self‐compassion and psychopathology, suggesting that high levels of self‐compassion were associated with lower levels of symptomatology.

Independently, these frameworks examine individual relationships between constructs. Therefore, a structural model was created to examine the relationships among variables and to test the direct, indirect, and mediating effects of these variables on the trajectory of disordered eating symptoms. The combined model posits that mindfulness and self‐compassion maintain a reciprocal relationship. The model also suggests that mindfulness and self‐compassion are adaptive emotion regulation strategies that subsequently have an impact on eating disorder symptoms, both distinctly and as forms of emotion regulation. It was hypothesized that if there was a high level of mindfulness, there would be an increased level of self‐compassion, a decreased level of emotion regulation difficulties, and subsequently less disordered eating pathology. Similarly, it was hypothesized that if there was a high level of self‐compassion, there would be an increased level of mindfulness, a decreased level of emotion regulation difficulties, and subsequently fewer eating disorder symptoms.

Eating Disorders

The Diagnostic and Statistical Manual of Mental Disorders (5th ed.; American Psychiatric Association, 2013) organizes eating disorders into three primary types: anorexia nervosa (AN), bulimia nervosa (BN), and binge‐eating disorder (BED). These disorders are characterized by disturbed eating attitudes and behaviors. Eating disordered attitudes include thoughts and affective states related to concerns about weight, self‐control, and acceptance from self and others (Fairburn & Brownell, 2013). Such attitudes have demonstrated associations between eating cognitions, eating behaviors, and psychological distress (Masuda, Price, & Latzman, 2012). Body image and disordered eating cognitions and behaviors are affected by various neurobiological and psychosocial factors; however, there is also evidence to suggest that these attitudes and behaviors are affected by cognitive factors, including mindfulness (Lavender et al., 2011), self‐compassion (Gilbert, 2009), and emotion regulation (Aldao & Nolen‐Hoeksema, 2010).

Mindfulness

Mindfulness refers to an awareness of moment‐by‐moment experiences that arises from purposeful and nonjudgmental present‐moment attention (Kabat‐Zinn, 2003). Mindfulness practices originated in Eastern spiritual traditions, and it was believed that regular engagement in meditation would develop mindfulness and result in increased levels of calmness, awareness, insight, wisdom, and compassion (Goldstein, 2002). In the last several decades, mindfulness practices have been modified and incorporated into medical and mental health interventions (Baer, Smith, Hopkins, Krietemeyer, & Toney, 2006).

There is consensus that body dissatisfaction is a product of unrealistic and unattainable ideals of what is attractive (Dijkstra & Barelds, 2011). Mindfulness does not focus on reaching a particular goal or appearance ideal but promotes willingness to accept the present state of one's appearance (Kabat‐Zinn, 1990). It fosters the ability to contact and remain in a present‐moment experience, which is contrary to body avoidance. It also encourages compassionate rather than judgmental attention, which is the opposite of the self‐critical process involved in eating disorders.

Dichotomous thinking and rumination have also been linked to the maintenance of eating disorders (Fairburn, Cooper, & Shafran, 2003; Park, Dunn, & Barnard, 2012). Dichotomous thinking is a form of cognitive rigidity that increases the appeal of “forbidden” food (Mann & Ward, 2001), enhances obsessiveness about food, and produces feelings of guilt if such food is consumed (Dewberry & Ussher, 2001). Rumination is a cognitive feature that includes recurring thoughts or worries about issues such as food, eating, body image, and self‐esteem and may reinforce disordered eating behaviors (Schwitzer & Choate, 2015). Mindfulness cultivates a productive mindset that counteracts dichotomous thinking, rumination, and experiential avoidance (Cowdrey & Park, 2012) because it is characterized by nonjudgmental observation of experiences. Therefore, it may encourage awareness of critical thoughts without overidentification with them (Alberts, Thewissen, & Raes, 2012). Prior research demonstrates that there is an inverse relationship between mindfulness and disordered eating cognitions (Lavender et al., 2011).

Self‐Compassion

Self‐compassion refers to the demonstration of kindness to oneself, the observation of one's own imperfection as an inevitable aspect of the human condition, and a mindful awareness of one's distressing thoughts and emotions (Neff, 2003). Those individuals who exhibit greater self‐compassion report higher levels of happiness, relational connectedness, and overall satisfaction with life, as well as lower levels of shame, fear of failure, depression, and anxiety (Neff, Rude, & Kirkpatrick, 2007). Likewise, positive relationships have been found between self‐compassion and optimism, agreeableness, extroversion, social connectedness, emotional intelligence, and personal initiative and negative relationships to neuroticism, rumination, self‐criticism, and thought suppression (Neff, 2003; Neff et al., 2007).

Lower levels of self‐compassion have been found to be related to increased body dissatisfaction, drive for thinness, and eating disorder symptoms (Ferreira et al., 2013). The nature of self‐judgment can evoke distress equivalent to or beyond that elicited by the particular circumstance (Germer, 2009). Disordered eating behaviors can function as a means of managing internal and external threats by avoiding criticism or rejection due to body shape and weight; however, such avoidance of experiences can exacerbate negative emotional states (Germer, 2009).

Because self‐compassion entails meeting one's thoughts, emotions, and experiences with a sense of empathy, kindness, and shared humanity (Neff, 2003), it can be seen as an adaptive way to regulate negative affect and potential threat (Gilbert, 2009). It promotes safety and encouragement to strive toward greater well‐being (Neff, 2004). When such a posture is taken with views about the body, individuals can experience greater connectedness, in that there is recognition of shared, negative body‐related experiences that require not suppression or change but rather validation and acceptance (Berry, Kowalski, Ferguson, & McHugh, 2010). Prior work (e.g., Albertson, Neff, & Dill‐Shackleford, 2015) has demonstrated a link between engagement in self‐compassion meditation practices and significant reductions in body dissatisfaction, body shame, and contingent self‐worth, as well as significant gains in self‐compassion, mindfulness, and body appreciation.

Emotion Regulation

Emotion regulation consists of those extrinsic and intrinsic processes individuals use to influence their psychological experiences and the ways in which they experience and express emotions (Masuda et al., 2012). Adaptive emotion regulation typically indicates the use of strategies that aid in managing unfavorable emotions in such a way that emotions are prevented from interfering with the attainment of goals (Bridges, Denham, & Ganiban, 2004).

Current etiological models of eating disorders highlight the impact of emotion regulation difficulties as a transdiagnostic factor (Treasure, Corfield, & Cardi, 2012). Such models of eating disorders theorize that eating disordered behaviors, such as binge eating and purging, serve the function of reducing or avoiding negative affective or cognitive experiences. Harrison et al. (2010) found that, in general, individuals with eating disorders reported greater deficits in emotional arousal, awareness, understanding, acceptance of emotions, and the ability to act in desired ways regardless of an emotional state.

The increasing prevalence of eating disorder symptomatology among college students warrants continued efforts to identify protective factors for this vulnerable population. The influence of mindfulness, self‐compassion, and emotion regulation on eating disorders is gaining attention in the literature. The combined impact of these constructs has been researched by Germer (2009), Lavender et al. (2011), and MacBeth and Gumley (2012), who have proposed a reciprocal relationship between self‐compassion and mindfulness and have suggested that both of these variables serve as adaptive emotion regulation strategies to mitigate disordered eating attitudes and behaviors. The factor that is believed to influence eating disorder symptoms is emotion regulation, which is affected by both mindfulness and self‐compassion. Thus, the purpose of this study was to test a model of potential protective factors against the development and maintenance of eating disorders among college students.

Method

Participants

Participants included 100 undergraduate and graduate students ages 18 to 52 years old (M = 23, SD = 5.14) from a large, public university in the southeastern United States. Students were recruited through paper flyers posted around campus and an online posting, as well as from an undergraduate honors course. The largest percentage reported being graduate students (56%), which may be due to recruitment methods as well as reflect a tendency for these students to be more involved in research. The majority identified as White (83%), followed by Asian/Pacific Islander (7%), Hispanic or Latino (3%), and Black or African American (2%), and other (2%). Three participants (3%) identified with two ethnicities (White and Native American or American Indian, Hispanic or Latino and Native American or American Indian, and Asian/Pacific Islander and other). The original demographic form omitted the question about sex; therefore, a follow‐up questionnaire was sent to participants to obtain this information. Participants included 12 men and 50 women; information about the sex of the remaining 38 participants was unavailable. Participants were asked to report their involvement in campus organizations, and the most commonly reported affiliations were with the Honors College (31%), academic and professional organizations (14%), Greek life (12%), and athletics (10%). Ten participants indicated that they were previously diagnosed with an eating disorder and seven participants reported receiving past or current treatment for their disorder(s). Twenty participants endorsed that they were currently receiving mental health treatment for concerns other than an eating disorder.

Measures

Demographics. Participants completed a demographic questionnaire, which asked about age, sex, ethnicity, current level of education, involvement in campus organizations, height and weight (to calculate body mass index), history with an eating disorder, history with treatment for an eating disorder, and history with mental health treatment.

Five Facet Mindfulness Questionnaire (FFMQ). The FFMQ (Baer et al., 2006) is a 39‐item instrument that uses a 5‐point Likert‐type scale. It contains five subscales that assess five facets of mindfulness, including Observing, Describing, Acting With Awareness, Nonjudging of Inner Experience, and Nonreactivity to Inner Experience. The FFMQ has a total score range of 39 to 195; higher scores are indicative of greater levels of mindfulness. In its initial validation with two undergraduate samples, Baer et al. (2006) found adequate to good internal consistency, with Cronbach's alphas for the facets ranging from .75 to .91. This study combined subscale scores to create a single overall score and demonstrated excellent internal reliability (α = .92).

Self‐Compassion Scale (SCS). The SCS (Neff, 2003) is a 26‐item measure that uses a 5‐point Likert‐type scale. It measures three facets of self‐compassion, with each facet containing two subscales: (a) Self‐Kindness versus Self‐Judgment, (b) Common Humanity versus Isolation, and (c) Mindfulness versus Over‐Identification. The SCS has a total score range of 26 to 130; higher scores are reflective of greater levels of self‐compassion. The SCS was initially validated with undergraduate students and was determined to have good psychometric properties (Neff, 2003). All of the two‐factor subscales demonstrated adequate internal consistency (.78 for Self‐Kindness, .77 for Self‐Judgment, .80 for Common Humanity, .79 for Isolation, .75 for Mindfulness, and .75 for Over‐Identification). This study combined subscale scores to create a single overall score and demonstrated excellent internal reliability (α = .96).

Difficulties in Emotion Regulation Scale (DERS). The DERS (Gratz & Roemer, 2004) measures six facets of emotion regulation difficulties, including nonacceptance of emotional responses, difficulties engaging in goal‐directed behavior, impulse control difficulties, lack of emotional awareness, limited access to emotion regulation strategies, and lack of emotional clarity. The DERS is a 36‐item measure that uses a 5‐point Likert‐type scale. It has a total score range of 36 to 180; higher scores are indicative of greater emotion regulation difficulties. The DERS was determined to have good psychometric properties in its initial validation with undergraduate students (Gratz & Roemer, 2004). Gratz and Roemer (2004) found good internal consistency of the facets of emotion regulation difficulties (with Cronbach's alphas ranging from .80 to .89). This study combined subscale scores to create a single overall score and demonstrated excellent internal reliability (α = .96).

Eating Attitudes Test–26 (EAT‐26). The EAT‐26 (Garner et al., 1982) is a 26‐item instrument that assesses disordered eating attitudes, beliefs, and behaviors. It uses a 6‐point Likert‐type scale and has a total score range of 0 to 78, with scores of 20 or more being reflective of eating disorder concerns. The EAT‐26 is highly correlated with the original EAT measure (r = .98) and was determined to have good psychometric properties in its initial validation with female undergraduate students (Garner et al., 1982). It has demonstrated excellent internal consistency in samples of Black and White adult women, with Cronbach's alphas of .85 and .88, respectively (Kelly et al., 2012). Good discriminant validity has been established in that it can differentiate those with AN, BN, and BED from controls (Williamson, Prather, McKenzie, & Blouin, 1990). This study yielded good internal reliability (α = .87).

Procedure

Prior to recruitment of participants, approval was obtained from the university's institutional review board. Recruitment occurred via the university counseling center, an honors undergraduate course, the university's online news system, and the campus creative arts center. Flyers and postings included a link to the online survey, which included informed consent, all questionnaires, and a counseling referral list. As a means of incentive, students had the opportunity to enter a drawing to win one of 10 $20 Visa gift cards.

Results

All data met the assumptions of univariate and multivariate normality, univariate outliers, linearity, homoscedasticity, multicollinearity, and correct model specification. Recommended sample size for path analysis models using four variables and two regression equations ranges from 80 (Stevens, 2002) to 100 (Loehlin, 2004). The proposed model hypothesized direct relationships between mindfulness, self‐compassion, emotion regulation, and eating disorder symptoms with sex as a covariate. Because sex was not significantly correlated with the measure of eating disorder symptoms, it was excluded from the model. Removal of sex as a covariate produced a just‐identified model. Use of the PROC CALIS modification function in SAS 9.4 resulted in the recommendation that the direct path from mindfulness to eating disorder symptoms be removed.

The revised hypothesized model (see Figure 1) was tested using the PROC CALIS function in SAS 9.4. Goodness‐of‐fit statistics were assessed, and the model demonstrated adequate fit. Information on goodness of fit can be found in Table 1. The initial model produced a nonsignificant probability value, χ2(91) = 0.1279, p = .7206, for the chi‐square test, suggesting that the model fit the data well. The standardized root‐mean‐square residual (SRMR), an absolute measure of fit, was also considered. The SRMR was .0062; a value of zero suggests perfect fit (Hu & Bentler, 1999).

Figure 1

Revised Hypothesized Model With Standardized Beta Coefficients *p < .05. **p < .001.

Table 1

Fit Indices of the Revised Model

Overall model fit was also assessed using the goodness‐of‐fit index (GFI), the adjusted goodness‐of‐fit index (AGFI), and the root‐mean‐square error of approximation (RMSEA) statistics. In this model, the GFI was equal to .9993 and the AGFI was equal to .9932, which suggested that the model fit the data well (O'Rourke & Hatcher, 2013). The RMSEA was equal to 0, with a 90% confidence interval of 0 to .1895, indicating excellent model fit (MacCallum, Browne, & Sugawara, 1996).

Additional measures of fit were evaluated, including the Bentler comparative fit index (CFI) and the Bentler–Bonett nonnormed fit index (NNFI). Scores above 0.90 indicate good model fit. The initial model produced a CFI of 1 and an NNFI of 1.0298.

To determine whether there were better fitting models, a step‐through statistical process was used to test the direct and then the indirect effects of the relationships between variables. However, none of the recommended changes significantly altered model fit and, in fact, they did not align with theoretical assumptions about the relationship between these variables. Demographic variables correlated with the outcome variable were then included in the revised model to determine whether they accounted for additional variance in the model. Specifically, involvement in a religious organization, r(50) = .408, p = .003; involvement in a service organization, r(50) = –.279, p = .050; status as an undergraduate freshman, r(98) = .198, p = .051; history of an eating disorder, r(98) = .412, p ≤ .0001; and history of eating disorder treatment, r(98) = .336, p = .0007, were subsequently included in the model. When each of these were added as covariates, independently and collectively, the model was not significantly strengthened.

In total, 10 models were analyzed. The best fitting model, based on both theoretical and statistical assumptions, was the revised hypothesized model. Goodness‐of‐fit statistics for this model can be found in Table 1.

Discussion

The purpose of this study was to examine potential protective factors against the development and maintenance of eating disorder symptoms among college students. The resulting model provides some support for the integrated theoretical frameworks for mindfulness, self‐compassion, emotion regulation, and eating disorder symptomatology, as outlined by Lavender et al. (2011), Germer (2009), and MacBeth and Gumley (2012). The model suggests that higher levels of mindfulness and self‐compassion decrease difficulties in emotion regulation, which also decreases the occurrence of eating disorder symptoms. As such, emotion regulation partially mediated the relationship between self‐compassion and eating disorder symptoms and fully mediated the relationship between mindfulness and eating disorder symptoms. The model also provided support for the linkage between mindfulness and self‐compassion. These constructs were found to be significantly and positively related.

One distinct difference between the findings of this study and the originally proposed model is the relationship between mindfulness and eating disorder symptoms. Prior research has shown a significant and negative direct relationship between mindfulness and disordered eating (Baer et al., 2006). In this study, the relationship was fully mediated by emotion regulation abilities. This suggests that only certain facets may be directly related to disordered eating, which was previously demonstrated by Lavender et al. (2011).

Another key difference is the proposed significance of the relationship between the variables of emotion regulation difficulties and eating disorder symptoms. There was a positive relationship between these constructs; however, this relationship did not reach statistical significance. Certain facets of emotion regulation may be more predictive of eating disorder symptoms than others, which aligns with some prior research (e.g., Racine & Wildes, 2013).

In addition, the covariate of sex was removed from this model. Although less is known about the prevalence of eating disorders in males (American Psychiatric Association, 2013), sex has been identified as a marker associated with AN, BN, and BED (DeBate, Blunt, & Becker, 2010). However, sex was not significantly related to disordered eating patterns in this study, which may be due to the limited information that was available for this demographic variable.

Implications for College Counseling

One of the greatest clinical implications of this study is its partial support for the combined theoretical frameworks of Lavender et al. (2011), Germer (2009), and MacBeth and Gumley (2012). It clarifies the distinct and collective impact of protective variables on the development and maintenance of eating disorders. Self‐compassion, both directly and indirectly through emotion regulation, was found to be a key variable in the trajectory of eating disorder symptoms. The findings also indicate that mindfulness and self‐compassion can be viewed as adaptive emotion regulation skills. Counselors working with individuals with an eating disorder or those who are considered to be “at risk” should consider the significance of facilitating development of emotion regulation skills through increasing mindfulness and self‐compassion. Self‐compassion exercises can be found online at www.self-compassion.org.

There are various mindfulness‐ and compassion‐based interventions available to aid in the reduction or prevention of disordered eating behaviors. Dialectical behavior therapy (DBT; Linehan, 1993) is a mindfulness and acceptance‐based intervention that emphasizes emotion regulation. In a meta‐analysis, Lenz et al. (2014) found that DBT may be efficacious in reducing the number of disordered eating episodes as well as decreasing comorbid depression. DBT addresses four areas of skill development: mindfulness, distress tolerance, interpersonal effectiveness, and emotion regulation (Chugani, 2017; Lenz et al., 2014). Chugani (2017) offers suggestions for implementing DBT in a college counseling setting.

Other helpful interventions include mindfulness‐based cognitive therapy (Segal, Williams, & Teasdale, 2002), compassion‐focused therapy (Gilbert, 2009), and the Mindful Self‐Compassion program (Neff & Germer, 2013). Mindfulness‐based cognitive therapy includes psychoeducational and experiential domains, such as meditation, the body scan, mindful eating, self‐acceptance, and the paradox of control (Segal et al., 2002). A meta‐analytic review of this program demonstrated positive physical and psychological outcomes in diverse populations as well as increases in self‐compassion (Chiesa & Serretti, 2009).

Compassion‐focused therapy is an approach intended to increase self‐compassion and emotional responsiveness. It involves a variety of exercises, such as visualization, development of self‐kindness through language, and engagement in self‐compassionate behaviors. A pilot study of the treatment showed significant decreases in self‐attacking tendencies, shame, feelings of inferiority, and depression (Gilbert & Procter, 2006). In addition, the Mindful Self‐Compassion program is intended to help individuals increase self‐compassion and mindfulness through use of meditation practices and self‐compassion letter‐writing exercises. A pilot study of the intervention demonstrated significant increases in self‐compassion, mindfulness, happiness, and life satisfaction, as well as significant decreases in anxiety, stress, and depression, all of which were maintained at 6‐month follow‐up.

Implications of this study's findings can also help to inform university efforts, particularly the establishment and/or improvement of eating disorder awareness and prevention programs on campuses. Undergraduate and graduate students are likely to experience emotional stress related to this phase of life and are at an increased risk of developing psychological problems, including eating disorders (DeBate et al., 2010). Developing programs to inform students about risk factors and effective interventions is critical.

The findings of the present analysis must be interpreted within the context of the study's limitations. There is limited external validity, considering that the data came from university students in the northeastern United States. The sample was a sample of convenience and may not accurately represent students in other universities, clinical populations, or more diverse populations. The study also used self‐report measures. Although survey responses were anonymous, it is possible that social desirability and/or level of introspective ability had an impact on responses. Last, there are limits to using path analysis. This form of analysis is often considered to be a form of causal modeling, but causality cannot be inferred. Despite these limitations, the findings offer valuable contributions to the literature.

It is important for the previously discussed limitations to be addressed in prospective research. Future studies should work to address this study's limitations related to samples, instruments, and research design. It is hoped that by taking into account these shortcomings, the model and related theories can be further elucidated.

Conclusion

This study offers insights about factors predictive of the development and maintenance of eating disorders in college students. It contributes to the relatively small, but growing, literature related to these variables' potentially protective impact on disordered eating. The findings suggest that there is a strong connection between mindfulness and self‐compassion, which influences one's emotion regulation abilities and subsequent eating attitudes and behaviors. Because eating disorders are some of the most fatal psychological disorders, it is hoped that continued research in this area will benefit those who experience their damaging consequences.