Eating disorders in adolescents have a severe impact on psychological and physical health (Smink et al., 2012). However, the effectiveness of current treatments is moderate, with remission rates of only 40% (Lock & Le Grange, 2019). Exploring the role of actual-ideal weight discrepancy, the discrepancy between an individual’s actual and ideal body weight, may prove valuable in understanding eating disorders (Jacobi & Cash, 1994). Examining actual-ideal weight discrepancy is important given weight discrepancy is associated with poorer treatment outcome in eating disorders (Boyd et al., 2018). The association between actual-ideal weight discrepancy and eating disorder symptoms has been demonstrated in both non-clinical and clinical eating disorder samples (Cash & Deagle, 1997; Forston & Stanton, 1992; Strauman et al., 1991). Self-related constructs, which encompass an individual’s self-perceptions, including perfectionism, self-esteem and self-compassion, may offer a valuable lens to understand actual-ideal weight discrepancy (Bardone-Cone et al., 2020).

Low self-esteem and elevated perfectionism are both maintaining processes of eating disorders in the transdiagnostic theory of eating disorders (Fairburn et al., 2003a) which guides Cognitive Behaviour Therapy in adolescents and adults (CBT-E; Dalle Grave & Calugi, 2020; Fairburn, 2008). Self-esteem and perfectionism are not therapeutic targets in the leading treatment for adolescents, family-based treatment (FBT; Lock & Le Grange, 2019). In the pursuit of more effective and personalised treatment for eating disorders, research on maintaining factors is needed (Bardone-Cone et al., 2020). Perfectionism is a risk and maintaining factor for eating disorders and has been associated with greater actual-ideal weight discrepancy in non-clinical adult samples (Bills et al., 2023; Egan et al., 2011; Landa & Bybee, 2007; Limburg et al., 2017; Stackpole et al., 2023; Wade & Tiggemann, 2013). The association between perfectionism and actual-ideal weight discrepancy in clinical samples of individuals with eating disorders has not been explored but is important to inform future research on tailored treatment plans for adolescent eating disorders. Low self-esteem is also a risk factor for eating disorders (Colmsee et al., 2021; Krauss et al., 2023). Actual-ideal weight discrepancy has been associated with low self-esteem in a non-clinical adolescent sample (Argyrides & Sivitanides, 2017). No study has examined whether low self-esteem is associated with higher actual-ideal weight discrepancy and eating disorder symptoms in a clinical sample of adolescents with eating disorders.

Self-compassion, which includes self-kindness, acceptance, and understanding towards one’s failures and suffering (Neff, 2003), has also been associated with eating disorder symptoms in non-clinical and clinical samples (Morgan-Lowes et al., 2023; Turk & Waller, 2020). Lower self-compassion and higher actual-ideal weight discrepancy are both associated with body dissatisfaction in adult non-clinical samples (Jansen et al., 2022). However, research on self-compassion and eating pathology in adolescents with eating disorders is limited with only one existing cross-sectional study. Pullmer et al. (2019) found a negative relationship between self-compassion and eating pathology, and that self-compassion was associated with eating pathology both directly and through the mediator of psychological distress. It was concluded that other variables may account for the association between self-compassion and eating disorders symptoms, and that additional factors should be examined (Pullmer et al., 2019). To date, no study has examined actual-ideal weight discrepancy and self-compassion in a clinical sample of adolescents diagnosed with eating disorders.

Research is needed to examine whether actual-ideal weight discrepancy mediates the relationship between self-related constructs and eating disorder symptoms in adolescents. Understanding this relationship may help to inform future treatment research to alleviate the impact of higher actual-ideal weight discrepancy. The aim of this study was to examine whether actual-ideal weight discrepancy is a potential mediator of the relationship between self-related constructs and eating disorder symptoms in adolescents. Specifically, we aimed to understand if perfectionism, self-esteem, and self-compassion contribute to eating disorder symptoms through the mediating pathway of actual-ideal weight discrepancy.

There are two novel contributions of the current study. First, while previous studies have typically examined self-related constructs individually, our study examines these collectively, providing a comprehensive understanding of their associations with adolescent eating disorders. Second, no study to date has examined the relationships between actual-ideal weight discrepancy and perfectionism, self-esteem, and self-compassion in a clinical sample of adolescents diagnosed with eating disorders. Understanding how these self-related constructs maintain symptoms in a clinical population is essential (Morgan-Lowes et al., 2023). Our examination of the mediating role of actual-ideal weight discrepancy provides new insights into symptom maintenance, addressing a significant gap in the previous literature.

It was hypothesised that higher perfectionism, lower self-esteem, and lower self-compassion would be associated with greater eating disorder symptoms in a sample of adolescents diagnosed with eating disorders. We also hypothesised an indirect effect of perfectionism, self-esteem and self-compassion, through actual-ideal weight discrepancy as a mediator of these associations.

Method

Participants

This study used data collected from participants recruited for the Helping to Outline Paediatric Eating Disorders (HOPE) Project database; a large, ongoing sequential cross-sectional registry which began recruitment in 1996 (Watson et al., 2013). The HOPE Project is based at the Perth Children’s Hospital, Western Australia. This hospital provides the only statewide child and adolescent eating disorder program in Western Australia’s public health (i.e., cost-free) system. Covering a vast expanse of 2.5 million km2 (965,000 miles2) and a total population exceeding 2.8 million, the program is the primary hub for specialised care in the region. The array of services includes inpatient, day program, and outpatient care, all delivered by a multidisciplinary team of professionals, including physicians, psychologists, dietitians, and other specialists. Beyond patient care, the program provides outreach and training for healthcare professionals in the region.

Inclusion criteria for this study were: (a) an eating disorder diagnosis according to the Diagnostic and Statistical Manual of Mental Disorders-5 (DSM-5; (American Psychiatric Association, 2013) (b) identify as female, (c) 12–17 years of age, and (d) first presentation to intake assessment at the program. Only female participants were included as the database contains almost all individuals who identify as female. Participants were excluded if they did not complete the Eating Disorder Inventory–3 Perfectionism subscale (EDI-P) (Garner, 2004), and the question in the Eating Disorder Examination (EDE, Version 12) (Fairburn & Cooper, 1993) pertaining to actual-ideal weight discrepancy. There were 242 participants who completed the EDI-P (Garner, 2004) since its inception in the database, between 2011 and 2021. Of these participants, 45% responded to the ideal weight question in the EDE (Fairburn & Cooper, 1993) with the remainder not answering the question regarding what they would wish their ideal weight to be. Not answering consisted of responses such as “I am not sure” or “I don’t know”. Participants who did not answer this question were excluded from the study by removal from the potential dataset available for the study. The final sample comprised 114 females with a DSM-5 eating disorder diagnosis (see Table 1 for demographics).

Table 1 Demographics of the female sample (n = 114)
Full size table

Measures

Actual-ideal weight discrepancy (Argyrides & Sivitanides, 2017)

Actual-ideal weight discrepancy was operationalised consistent with Argyrides and Sivitanides (2017) by obtaining participants actual weight (in kilograms) at medical examination and by obtaining participants desired weight (in kilograms) from the Eating Disorder Examination (Fairburn & Cooper, 1993) question “What weight would you like to be?”. The difference between their actual and desired weight (measured in kilograms) is referred to as actual-ideal weight discrepancy. Argyrides and Sivitanides (2017) demonstrated that this measure exhibited convergent validity, as it was significantly and positively associated with appearance dissatisfaction, weight-related anxiety, internalization of the thin ideal, and disordered eating symptoms.

Eating disorder inventory – 3 perfectionism subscale (Garner, 2004)

The six-item EDI-P subscale was used to assess perfectionism (Garner, 2004). Items were scored on a 6-point Likert scale ranging from 1 (always) to 6 (never). An example item is “I hate being less than the best at things”. Items are summed and a higher score indicates greater perfectionism. The EDI-P has demonstrated good internal consistency and validity (Clausen et al., 2011; Garner, 2004; Shore & Porter, 1990). The internal consistency in the current study was good (Cronbach’s α = 0.74).

Rosenberg self-esteem scale (Rosenberg, 1965)

The Rosenberg Self-Esteem Scale (RSES) (Rosenberg, 1965) is a 10-item self-report scale, and the total score was used to measure self-esteem. Items are scored on a 4-point Likert score, ranging from 1 (strongly agree) to 4 (strongly disagree). An example item is “At times, I think I am no good at all”. The RSES score ranges from 10 to 40, with a higher score indicating higher self-esteem (Rosenberg, 1965). The RSES has good discriminant (Robins et al., 2001) and convergent validity with other measures of self-esteem (Bosson et al., 2000). The internal consistency in this study was excellent, Cronbach’s α = 0.92.

Self-compassion scale-short form (Raes et al., 2011)

The 12-item Self-Compassion Scale-Short Form (SCS-SF) (Raes et al., 2011) total score was used to assess self-compassion. An example item is “when I’m going through a very hard time, I give myself the caring and tenderness that I need”. Answers are given on a 5-point Likert scale, ranging from 1 (almost never) to 5 (almost always). Higher scores indicate greater self-compassion, and the total score was the sum of the 12 items, after reverse scoring negatively worded items. The SCS-SF has demonstrated good internal consistency (Cronbach’s α = 0.86) and near-perfect convergent validity the SCS long form (r =.97) (Raes et al., 2011). The internal consistency was good in the current study (Cronbach’s α = 0.88).

Child-adapted eating disorder examination (Fairburn & Cooper, 1993)

The EDE was used to measure eating disorder symptoms, using a child informant version which has been adapted in the clinical setting and has good reliability and validity (O’Brien et al., 2016). The child-informant version is similar to the Child EDE (ChEDE) (Bryant-Waugh et al., 1996; Watkins et al., 2005), but the development of our measure and eating disorder service predated the introduction of the ChEDE. Items were summed using manual scoring, with the subscales of restraint, eating concerns, shape concerns and weight concerns being added to form a total score. In this study, internal consistency was good to excellent, α = 0.95 (Total), α = 0.78 (Restraint) α = 0.80 (Eating Concerns) α = 0.91 (Shape Concerns), and α = 0.87 (Weight Concerns).

Procedure

Ethics approval for this study was granted from the Perth Children’s Hospital ethics committee (HREC: 2042/EP). Reciprocal approval was also gained from Curtin University ethics committee HREC (HRE2017-0119). Data collection took place at the standard two-day clinical intake assessment conducted by a multidisciplinary team. Upon referral and admission to the program, HOPE project information sheets were supplied, and child and parental consent were obtained. Participant confidentiality was ensured through the anonymization of the data.

Statistical analysis

Zero-order correlations were used to examine the correlation between all variables. A path analysis was conducted in Mplus V8 (Muthén & Muthén, 2017) with maximum likelihood estimation and 5000 bootstrapped resamples. Mplus was chosen for path analysis due to the advantage of robust handling of missing data. Maximum likelihood estimation was appropriate for the current model considering all model variables were continuous. Perfectionism, self-esteem, and self-compassion served as covarying predictors, with actual-ideal weight discrepancy as the mediator, and eating disorder symptoms as the outcome. Age was entered as a covariate, given its correlation with all predictors. All predictors were standardised before testing the model. Direct and indirect effects were tested using MacKinnon’s (2008) methods, with significance of indirect effects determined by bootstrapped 95% confidence intervals. The bootstrapping method is robust in cases where traditional assumptions about normality may not hold and provides reliable estimates, even with small sample sizes or complex models with multiple paths and variables (Preacher & Hayes, 2004). Although one case was observed to be a multivariate outlier, its inclusion or exclusion did not significantly alter the results. Thus, analyses involved all 114 participants. A negligible (< 1%) amount of data was missing across all model variables. Expectation maximisation was used to impute these missing data. The data were assumed to be missing completely at random (MCAR; Little’s MCAR test, χ2(653) = 828.65, p <.001). The rationale for assuming missing data were MCAR was based on the Little’s MCAR test results suggesting that missing data were likely random and not related to systemic patterns.

Results

Descriptive statistics and intercorrelations

Table 2 presents means, standard deviations, and zero-order correlations between variables. All predictors were significantly correlated with eating disorder symptoms and actual-ideal weight discrepancy in the hypothesised directions, with the exception of perfectionism, which was not significantly correlated with actual-ideal weight discrepancy.

Table 2 Correlations between and means and standard deviations of study variables note
Full size table

Relationships between self-related constructs, actual-ideal weight discrepancy, and eating disorder symptoms

Figure 1 represents the path model depicting the paths from the predictors (perfectionism, self-esteem, and self-compassion) to eating disorders symptoms, independent of and through the mediator, actual-ideal weight discrepancy. The figure depicts the direct effect coefficients, and Table 3 presents the indirect effects and total effects of the paths involving the mediator. Overall, the model explained a large proportion of the variance in eating disorder symptoms (59%). In support of our hypotheses, the direct path from self-esteem to eating disorder symptoms was significant. Further, the indirect path from self-esteem to eating disorder symptoms via actual-ideal weight discrepancy was significant. The paths from self-esteem to actual-ideal weight discrepancy, and from actual-ideal weight discrepancy to eating disorder symptoms were also significant.

Fig. 1
figure 1

Path model from perfectionism, self-esteem and self-compassion via actual-ideal weight discrepancy to eating disorder symptoms. Note. Black double headed arrows represent covariances between predictors, grey dotted paths are non-significant and significant paths are represented by solid black lines and bold text. Although age was controlled for, it was not a main predictor and is not included in the diagram. All coefficients are standardised. 95% CI intervals are in parentheses

Full size image
Table 3 Indirect and total effects of the model predicting eating disorder symptoms in adolescents with eating disorders
Full size table

Contrary to our hypotheses, the direct and indirect paths (via actual-ideal weight discrepancy) from perfectionism and self-compassion to eating disorder symptoms were not significant. Notably, this was in contrast to the zero-order correlations which showed significant associations between perfectionism and eating disorder symptoms, self-compassion and eating disorder symptoms, and self-compassion and actual-ideal weight discrepancy (Table 2). This absence of unique associations in the path analysis might be explained by the nature of regression, which captures only the unique contributions of each predictor, and not shared variance between predictors. Given there was a high degree of correlation, shared variance and overlap between the constructs, it is likely that when combined in the model, it was difficult to detect an effect for perfectionism and self-compassion beyond self-esteem. Further, the measure of perfectionism may not have been sensitive (see Discussion for further details).

We tested a second model for exploratory purposes to better understand the association between self-esteem and eating disorder symptoms via actual-ideal weight discrepancy. In this model, we split the EDE into subscales to better understand how self-esteem and actual-ideal weight discrepancy predicted various symptoms. The rationale for this exploratory analysis was to gain a more nuanced understanding of how self-esteem and weight discrepancy may relate to specific eating disorder symptoms. The path model and the direct effects are depicted in Fig. 2. As shown in Table 4, the indirect paths between self-esteem and all four EDE subscales via actual-ideal weight discrepancy were all significant.

Fig. 2
figure 2

Path model from self-esteem via actual-ideal weight discrepancy to eating disorder symptoms. Note. Black double headed arrows represent covariances between outcomes, and significant paths are represented by solid black lines. Although age was controlled for, it was not a main predictor and is not included in the diagram. All coefficients are standardised. 95% CI intervals are in parentheses

Full size image
Table 4 Indirect and total effects between self-esteem and eating disorder symptoms
Full size table

Discussion

The current study aimed to determine whether self-related constructs are associated with eating disorder symptoms in adolescents, independently and through the mediator of actual-ideal weight-discrepancy. Lower self-esteem and self-compassion, and higher perfectionism, were associated with greater eating disorder symptoms, consistent with previous research (Bills et al., 2023; Drieberg et al., 2019; Johnston et al., 2018; Jones et al., 2020; Morgan-Lowes et al., 2023; Turk & Waller, 2020). Lower self-esteem and self-compassion were related to greater actual-ideal weight discrepancy, consistent with previous studies in non-clinical samples (Argyrides & Sivitanides, 2017; Jansen et al., 2022).

The current study highlights a role of actual-ideal weight discrepancy in the association between self-esteem and eating disorder symptoms given that self-esteem was the most salient predictor. Building on previous clinical and non-clinical research showing an association between actual-ideal weight discrepancy and eating disorder symptoms (Boyd et al., 2018; Gardner et al., 2000; Jansen et al., 2022; Mason et al., 2016), the present study shows self-esteem may be important in this association. The findings align with studies that have found that self-esteem and actual-ideal weight discrepancy are related in non-clinical adolescents (Argyrides & Sivitanides, 2017). A future research direction would be to evaluate the prospective associations to establish causal risk pathways and whether changes in self-esteem and actual-ideal weight discrepancy correlate with and occur before change in eating disorder symptoms.

Self-compassion and self-esteem, although different constructs, are closely connected (Neff, 2012). In our sample, self-compassion and self-esteem were very highly correlated (r =.75). The overlap between self-compassion and self-esteem might explain why self-compassion was not a unique predictor of actual-ideal weight discrepancy or eating disorder symptoms. Specifically, if an individual feels unworthy and has a negative self-view, then they are likely not to be understanding and kind towards themselves when they feel inadequate (low self-compassion; Donald et al., 2018). A longitudinal study with adolescents found self-esteem to be a crucial antecedent for developing self-compassion, but self-compassion was not a critical antecedent for developing self-esteem (i.e., unidirectional effect) (Donald et al., 2018). The data from this longitudinal study suggests that self-esteem is a more important predictor of the development of eating disorder symptoms in adolescents than self-compassion (Donald et al., 2018). Given the significant correlations between variables, shared variance and overlap between variables, it is plausible that self-compassion is associated with eating disorder symptoms through the pathway of actual-ideal weight discrepancy, potentially due to shared variance with self-esteem.

In contrast to previous literature (Landa & Bybee, 2007; Wade & Tiggemann, 2013) perfectionism was not correlated with and did not predict variance in actual-ideal weight discrepancy. Nonetheless, consistent with previous research there was a significant positive correlation between perfectionism and eating disorder symptoms (Bills et al., 2023; Egan et al., 2011; Limburg et al., 2017). The EDI-P assesses both socially prescribed perfectionism, the belief others expect you to be perfect and self-oriented perfectionism, setting high standards for one’s own performance (Hewitt & Flett, 1991). It is likely that different aspects of perfectionism are more strongly related to actual-ideal weight discrepancy (e.g., concern over mistakes, parental criticism; Wade & Tiggemann, 2013). Future research could investigate which aspects of perfectionism are associated with actual-ideal weight discrepancy. It would also be helpful for future studies to use measures of perfectionism more closely linked with eating disorders, e.g., clinical perfectionism (see Shafran et al., 2002) with the Clinical Perfectionism Questionnaire (Fairburn et al., 2003b) which has demonstrated reliability and validity in adolescents (Shu et al., 2020) and eating disorder samples (Egan et al., 2016; Prior et al., 2018). The potential lack of findings regarding perfectionism in the path model may be due to the EDI-P, which is supported by the results of Jones et al. (2020) in an examination of the transdiagnostic model of eating disorders (Fairburn et al., 2003a) when the EDI-P was used in adolescents with eating disorders and a direct association between perfectionism and eating disorder symptoms was not found, in contrast to the substantial literature on this association (Bills et al., 2023). Jones et al. (2020) argued that the EDI-P may not have adequately captured perfectionism and suggested the CPQ should be used. In a subsequent study of the transdiagnostic model in a non-clinical sample of adolescents Jones et al. (2024) used the CPQ to measure perfectionism and found the expected significant association between perfectionism and eating disorder symptoms. Hence, we agree with Jones et al. (2020) that future research should use the CPQ rather than the EDI-P to examine perfectionism and eating disorder symptoms in adolescents as it may be a more sensitive instrument and is in line with the construct of clinical perfectionism in the transdiagnostic theory of eating disorders (Fairburn et al., 2003a).

Although self-esteem was salient as a unique predictor of eating disorder symptoms, the significant correlations between perfectionism, self-esteem, and self-compassion with eating disorder symptoms, and the significant correlations with each other, suggest these self-related traits cluster together and possess shared variance. Conceptualising self-related constructs as overlapping and not delineates is relevant for future research, especially given the high degree of variance explained by the combination of these variables in the path model and prior findings on each of these constructs (Egan et al., 2011; Limburg et al., 2017; Turk & Waller, 2020).

Given self-esteem was the most salient predictor of eating disorder symptoms, our findings provide support for the utility of future research examining self-esteem as a therapeutic target in the treatment of adolescent eating disorders. This direction for future research is supported by data indicating self-esteem was the most salient variable from the transdiagnostic model of eating disorders (Fairburn et al., 2003a) in adolescents with eating disorders (Jones et al., 2020). A focus on self-esteem in treatment of eating disorders is also supported by meta-analyses of longitudinal data demonstrating that self-esteem predicts eating disorder symptoms over time (Colmsee et al., 2021; Krauss et al., 2023).

In Fairburn et al.’s (2003a) transdiagnostic model of eating disorders, self-esteem is proposed to maintain eating disorders as the individual with low self-esteem sets goals in the domain of eating, shape, and weight to improve their self-evaluation related to appearance. In addition, individuals with core low self-esteem have a pervasive negative self-view (Fairburn et al., 2003a). In CBT-E (Dalle Grave & Calugi, 2020; Fairburn, 2008), if an individual has core self-esteem and it interferes with treatment of the eating disorder, a broader version of the treatment that is focused on additional maintaining factors is implemented (see Fairburn, 2008 for further details). The CBT-E module for core low self-esteem draws on an established protocol of CBT for low self-esteem (Fennell, 2016) that is effective across a range of psychological symptoms (Kolubinski et al., 2018). Techniques include cognitive therapy to reduce negative self-schemas, behavioural experiments to test negative evaluative self-beliefs, and recording diaries of positive qualities (Fennell, 2016).

A clinical implication of our study is that future research could examine if self-esteem is a useful target for reducing actual-ideal weight discrepancy and eating disorder symptoms in adolescents. Despite FBT (Lock & Le Grange, 2019) being the treatment of choice for adolescents, there is preliminary evidence for the effectiveness of CBT-E (Dalle Grave et al., 2021). In a non-randomised study comparing FBT and CBT-E, the treatments demonstrated similar effectiveness, except for a favouring of FBT in restoring weight in underweight adolescents (Le Grange et al., 2020). However, a key difference is that CBT-E has specific modules which target low-esteem when elevated. Self-esteem may be a particularly pertinent maintaining process to target in adolescents with eating disorders given the developmental focus in this age range on self-esteem (Jones et al., 2020). Future research should continue to determine the efficacy of treatments targeting low self-esteem in adolescents with eating disorders, including CBT-E (Dalle Grave & Calugi, 2020). In adolescents CBT-E could potentially be refined with an increased focus particularly on self-esteem and incorporating additional sessions based on Fennell’s (2016) CBT for low self-esteem. This is important given the significance of self-esteem as the predominant predictor in our sample for both actual-ideal weight discrepancy and eating disorder symptoms.

There were numerous limitations of the study. Many participants (55%) did not respond to the question related to “desired weight”, making it impossible to calculate actual-ideal weight discrepancy for these individuals. The question regarding what they wish their ideal weight to be was asked at the baseline admission to the eating disorders service, and the typical responses to those who could not provide an answer were stating that they did not know or were unsure. Unfortunately, it was not possible to retrieve this missing data from participants since the question was only asked at baseline assessment and was not repeated when the participant did not provide a specific answer. This is an interesting clinical finding why participants did not respond, which reflects why actual-ideal weight discrepancy in adolescent eating disorders is an important area of research. Future research could examine patients who did endorse actual-ideal weight discrepancy compared to those who did not, to find out why this question is seldom answered. There are a number of hypotheses as to why this is not answered, including patients denying or under-reporting actual-ideal weight discrepancy due to a fear of hospital admission. Future studies may include a question regarding their ideal weight as part of a self-report questionnaire. The advantage of a self-report question is that it may reduce anxiety about the disclosure of an ideal weight they feel that the professional may think is unsuitable that could have potentially led to the high degree of missing data in the current study. A self-report question could be compared to interview reports to determine whether self-report may be a less anxiety-provoking approach to obtain a measure of actual-ideal weight discrepancy and increase response rates.

A significant limitation of the study was that it was a female only sample. This was because almost all participants in the database identified as female. The female only sample limits the generalisability of our findings to males and individuals who identify as non-binary or gender diverse. This is an important limitation given eating disorders affect all genders, including males (Shu et al., 2015), and gender diverse youth have an increased risk of eating disorders (Watson et al., 2017). Future research should be conducted across all genders instead of a female only sample. Another major limitation was the cross-sectional design, and therefore causation could not be established. The cross-sectional nature of the study with participants who have an eating disorder diagnosis prevents causal inferences. It is not possible to ascertain whether low self-esteem represents a risk-factor for the development of eating disorder symptoms, or a consequence of the core psychological disturbances of an adolescent with an eating disorder given the study design. Future longitudinal research could assess self-esteem, actual-ideal-weight discrepancy and eating disorder symptoms over time to determine time-order effects and whether these variables have a unidirectional or bidirectional relationship in adolescents.

In summary, actual-ideal weight discrepancy is related to a range of self-related constructs including perfectionism, self-compassion and particularly self-esteem. This study addressed an important gap in in the eating disorder field. Despite ongoing efforts, adolescent treatment outcomes remain unsatisfactory. There is a compelling need to test promising treatment and theoretical models while addressing the maintaining processes that contribute to eating disorders. The findings have the potential to inform future research directions on innovative treatment approaches. Actual-ideal weight discrepancy is an important construct to examine further in adolescents with eating disorders.