Emotional intelligence (EI) is a relatively new construct related to emotions that comprises a set of basic emotional skills. From an ability perspective, EI is defined as “the ability of people to perceive, use, understand and manage emotions” (Mayer and Salovey, 1997; p. 532). This EI theoretical framework involves four basic emotional dimensions: emotional perception and expression, emotional facilitation, emotional understanding, and emotional regulation. From this perspective, emotional regulation is defined as the ability to manage both positive and negative emotions in themselves and in others, integrating emotion and cognition effectively (Mayer et al., 2016). These emotional skills can be developed in clinical settings through systematic and comprehensive EI training (Mayer et al., 2000; Kotsou et al., 2018).

In the last decade, an emerging line of research has typically shown the significant relationship between EI and ED symptoms. For example, some prior research has found that individuals with some EI deficits are more prone to display disordered eating attitudes and behaviors (Costarelli et al., 2009; Pettit et al., 2009; Filaire et al., 2010; Zysberg and Rubanov, 2010; Hambrook et al., 2012; Zavala and López, 2012; Zysberg, 2013; Zysberg and Tell, 2013; Gardner et al., 2014; Koch and Pollatos, 2015; Cuesta et al., 2017; Peres et al., 2018; Foye et al., 2019). In addition, recent systematic reviews have reported that these abilities are relevant factors in both the development and maintenance of ED (Romero-Mesa et al., 2020; Giusti et al., 2021). These findings provide some preliminary support for the role of emotions in disordered eating attitudes with a view to the prevention and management of ED and point to the potential use of EI measures to identify individuals at risk of ED.

Beyond this direct association between EI and ED symptoms, other potential underlying processes have been theorized through which EI might impact ED symptoms. Two of these individual psychological mechanisms considered to be relevant mediators might be self-esteem and anxiety. A growing body of research has supported the theory that deficits in emotional skills are significant predictors of reduced self-esteem domains (Fernández-Berrocal et al., 2006; Hasanvand and Khaledian, 2012; Bibi et al., 2016). Likewise, it has been found that people with high levels of anxiety typically report difficulty in accurately perceiving, using, understanding, and managing their own emotions (Fernández-Berrocal et al., 2006; Connor and Slear, 2009; Kousha et al., 2018). Taken together, these findings suggest that people with high levels of EI feel more security and less stress and also believe in their abilities, showing higher feelings of self-worth, goodness, and self-respect.

On the other hand, there is some theoretical and empirical support that self-esteem and anxiety are risk factors for ED symptoms. As mentioned above, improving self-esteem in combination with other components resulted in a more effective treatment for ED patients compared to CBT therapy alone. In addition, several studies have consistently corroborated a negative relationship between self-esteem and ED criteria (Silverstone, 1990, 1992; Silvera et al., 1998). Besides, numerous studies have found that high levels of anxiety are a well-recognized symptom in individuals with ED (Arnow et al., 1995; Levinson and Rodebaugh, 2012; Menatti et al., 2015) and that anxiety worsens the ED psychopathology (Brand-Gothelf et al., 2014). Given the high comorbidity between ED and anxiety disorders and the fact that both disorders share the same components (evaluative, affective, and somatic; Kaye et al., 2004), the robust relationship between ED and anxiety is not surprising.

Along with the individual role of self-esteem and anxiety in ED symptomatology, several studies have tested the combined role of self-esteem and anxiety on ED. For example, a recent study found that self-esteem (together with mood dysregulation) moderated the association between levels of anxiety/depression and greater deterioration of ED (Sander et al., 2021). In addition, Aloi and Segura-García (2016) found that low self-esteem had an indirect effect on the risk of developing an ED through the mediating action of anxiety.

There is also evidence that anxiety and self-esteem (considered independently) play a relevant mediating role between EI and ED symptoms. For example, prior findings have found that social anxiety mediated the relationship between EI and ED risk (Li, 2018). Thus, Hambrook et al. (2012) found that self-reported anxiety levels mediated the observed relationship between EI and anorexia nervosa (AN).

In sum, there is empirical support for the link between EI, higher self-esteem, and reduced anxiety, as well as between levels of self-esteem and anxiety and ED symptomatology. It has also been pointed out that individuals with higher EI use their ability to maintain a global feeling of self-worth when appropriate and effectively manage a distressed mood when faced with negative events that are considered key in the development of ED symptoms. Hence, the assessment of self-esteem and anxiety as potential mediators in the association between EI and the symptomatology of ED seems to be justified (Hambrook et al., 2012; Li, 2018). However, most of these studies only examined the role of a single mediator in the linkage EI-ED symptomatology and have rarely evaluated effects of the mediators concurrently. Specifically, and to the best of our knowledge, no prior research has examined the cumulative effect of self-esteem and anxiety on the relationship between EI and ED symptomatology. Including these mediators in the serial model would help unveil the underlying mechanisms through which EI influences ED and would also help clinicians and educators to focus on the factors with the most clinical relevance for the prevention and treatment of ED symptomatology. Therefore, the purpose of the present study is 2-fold: first, we sought to examine the relations between EI, self-esteem, anxiety, and ED symptomatology. Second, we sought to determine whether self-esteem and anxiety mediated the relation between EI and ED symptoms in sequence. Since prior studies have found that individuals with higher self-esteem are more likely to have lower levels of anxiety (Sowislo and Orth, 2013), and that self-esteem has an indirect effect on ED risk, as mediated by anxiety (Aloi and Segura-García, 2016), we expected that both mediators might act in this sequence; that is, individuals with high positive cognitions about themselves would have lower levels of anxiety, which in turn would lead them to an amelioration of ED symptomatology. Overall, considering prior research on the significant associations between EI, self-esteem, anxiety and ED symptoms, we developed the following research hypotheses:

The study sample consisted of 516 Spanish undergraduate students and community adults (319 females and 197 males) located in Southern Spain. Their ages ranged from 18 to 77 years, with a mean age of 38.89 years (SD = 14.76) (see Table 1). Given that the prevalence of ED in elderly women (aged 65–94 = 3.25%) has been found to be comparable to young women (Conceição et al., 2017) and that, according to a systematic review, late-life onset ED do continue to occur in the elderly (Lapid et al., 2010), we have not excluded older people from our study.

The educational level in the present sample was as follows: 15 (2.9%) had no formal education, 73 (14.1%) had a primary-level education, 120 (23.3%) had not completed secondary school, 57 (11.0%) had completed secondary school, 186 (36.0%) had completed University Studies, and 65 (12.6%) had post-graduate studies.

The percentage of at-risk participants (i.e., those scoring ≥ 20 in the EAT-26) was 6.2% (32 out of 516).

University and community participants were solicited using non-probabilistic convenience sampling techniques via an online survey format. A student-recruited sampling methodology was used following guidelines by Wheeler et al. (2014), which allowed us to access a community sample from a University setting. The community participants were recruited with the assistance of students enrolled in a psychology course at university, who were asked to recruit at minimum two adults over the age of 30 through their personal network and then administer the online version of the questionnaires to them. The online survey was designed so that incomplete questionnaires could not be saved, which allowed only the whole completed questionnaires to be received. Student participants earned points for their participation in the study. Before getting to the online survey, participants were informed that the survey was about eating habits and emotions, and that their participation was entirely voluntary. All participants provided written informed consent according to the Declaration of Helsinki. The procedure was approved by the ethics committee of the University of Málaga (104-2020-H). The administration procedure lasted for ~30 min. The sample was obtained from May 2019 to November 2020. The percentage of participation among students was of 78.28%.

The following well-validated measures were used.

After calculating descriptive statistics and computing the bivariate correlation between EI, self-esteem, anxiety, and ED symptoms, the SPSS macro PROCESS (Hayes, 2018) was used to conduct multiple mediation analyses for testing the potential mediating role of self-esteem and anxiety in the linkage between EI-ED symptomatology. A bootstrapping method with 5,000 esteem resamples was used to calculate overall indirect effects and specific indirect effects. The direct and indirect effects are considered to be statistically significant if 95% of the bootstrap confidence intervals do not contain zero. Thus, this procedure enables multiple mediators to be examined and determines the independent effect of each mediator while controlling for the other. Preliminary ANOVA and t-tests showed differences according to age and sex, respectively, in ED symptoms, self-esteem and anxiety. Female participants scored significantly higher than males in ED symptoms and anxiety, and lower in self-esteem (all ps < 0.01); younger participants scored marginally significantly higher than older ones in ED symptoms and anxiety (ps < 0.10) and significantly lower in self-esteem (p <0.001). Therefore, sex and age were entered as covariates to control any potential confounding effects.

Descriptive statistics, response ranges, normative data, reliability coefficients, and correlations among the study variables are presented in Table 2. The column “Response ranges” indicates the minimum and maximum possible scores in the questionnaires. The column “Normative data M (SD)” indicates means and standard deviations from normative data. For EI, self-esteem, and anxiety, normative data correspond to the information provided by Spanish adaptations of the questionnaires. For the EAT-26, since the Spanish adaptation of EAT-26 was performed with adolescents, we have included the data from Johnson and Bedford (2004), who employed a Canadian sample comparable to ours (men and women, age range 18–94). As Table 2 shows, EI was positively correlated to self-esteem and negatively associated to anxiety. As expected, self-esteem was significantly and negatively associated with ED symptoms, while anxiety levels showed significant and positive associations with symptoms of ED. Finally, EI was negative and significantly linked to ED symptomatology.

We examined whether the relationship between EI and ED symptoms was sequentially mediated by self-esteem and anxiety. Both age and sex were added as control variables. Results of the mediation analyses are presented in Table 3.

As shown in Figure 1, the three hypothetical mediating effects were supported. First, the specific indirect effects of EI on ED symptoms through self-esteem [EI → Self-esteem → ED symptoms] were supported (B = −0.09, SE = 0.03; 95% CI = −0.15, −0.04). Second, anxiety was found to mediate the association between EI and ED symptoms [EI → Anxiety → ED symptoms] (B = −0.03, SE = 0.01; 95% CI = −0.066, −0.009). Third, the sequential pathway of EI → Self-esteem → Anxiety → ED symptoms, was significant (B = −0.02, SE = 0.01; 95% CI = −0.031, −0.005). Accordingly, higher levels of EI were serially associated with higher self-esteem, lower anxiety, and finally lower ED symptoms. Thus, the residual direct pathway between EI and ED symptoms was no longer significant (b = −0.03, p = 0.578). Therefore, self-esteem and anxiety fully mediated the link between EI and ED symptoms. This final serial mediation model was significant, accounting for 11% of the variance in ED symptoms [R² adj = 0.11; F_{(5, 510)} = 12.08; p <0.001].