Introduction

Although there has not been a consensus on the precise definition of compassion in the psychological literature [1], this concept may be defined as “a basic kindness, with a deep awareness of the suffering of oneself and of other living things, coupled with the wish and effort to relieve it” (p. xiii) [2]. It has been suggested that it involves a set of attributes and skills [3]. The attributes, such as empathy, sympathy, and distress tolerance, provide the means to engage, stay connected with, and understand our own, as well as other’s suffering. Compassionate skills (e.g. compassionate reasoning and behaviour) provide the knowledge to prevent or relieve suffering, as well as the courage to act on this knowledge [3]. According to Gilbert, compassion can flow in different directions [2]: feel compassion for oneself; feel compassion for others; and receive compassion from others.

There is growing evidence that experiencing compassion promotes well-being and improves mental health (e.g. [4, 5]). Furthermore, a recent study has demonstrated that self-compassion attributes and actions are associated with a lower susceptibility to adopt disordered body and eating attitudes and behaviours [6]. However, there are individuals whose capacity for experiencing and expressing compassion is not developed or accessible, thus they may not benefit from its protective effects [3]. Moreover, empirical research and clinical observations have highlighted that some individuals actively resist developing and cultivating compassion for themselves and for others, and receiving compassion from others [7]. These individuals can find these experiences as uncomfortable, difficult, or aversive, resulting in their avoidance and even fear of them [7]. These fears have been associated with various psychopathological indicators [7, 8], including disordered eating [9, 10].

It has been suggested that insecure attachment styles are major factors underpinning resistance to and fear of compassion, as these are related to difficulties with empathic engagement and effective care for self and others [7, 11]. Negative beliefs regarding compassion (e.g. receiving compassion may be viewed as a signal of weakness, self-indulgency, or as not deserved) are also described as pervasive factors underpinning these negative attitudes regarding compassion, particularly in individuals who present high levels of self-criticism [7, 12].

The presence of fears of compassion may imply an active avoidance of self-generated or social signals of warmth, kindness, care, reassurance, and support [7, 13]. Indeed, a recent study [14] has found negative correlations between fears of compassion and social safeness. According to the tripartite model of affect regulation [15], difficulties experiencing feelings of social acceptance and connectedness [16] are considered to underlie many forms of psychopathology [13, 15]. As such, when one feels socially unsafe, one is more focused on defending oneself or competing for resources. Humans present a fundamental motivation to seek for social acceptance and approval, since the establishment of social relationships has granted access to social resources (e.g. support, sexual partners, friendships, alliances), which guarantee survival and prosperity [17,18,19]. Empirical evidence has demonstrated that feeling socially accepted and connected is negatively associated with insecure striving [16, 20]. Insecure striving has been conceptualized as a social strategy to achieve social approval and acceptance, in which individuals feel a constant pressure to compete to be regarded as attractive. However, this strategy implies a sense of scrutiny associated with the fear of negative social consequences contingent to failures in such competition (e.g. criticism, rejection) [20, 21]. Accordingly, it promotes a competitive and rank-focused social mentality with heightened social comparison, fear of involuntary subordination (i.e. perceiving self as low ranked in a social hierarchy of attractiveness), and shame proneness [16, 20, 22, 23]. Moreover, insecure striving has been pointed out as a predictor of psychopathology [20, 24], including disordered eating (e.g. [23, 25]).

In modern societies, body image has had a key role in the social ranking of attractiveness, especially for women [26]. As such, most women regard it as a central dimension for self-evaluation, as well as an instrument to estimate their social rank via social comparison and to enhance it (e.g. [19, 27]). Over the last decades in Western societies, thinness has been established as the ideal female body image and linked to success, beauty, status, power and happiness (e.g. [28, 29]).

Women tend to compare themselves to other women who are considered representative of the ideal body image (e.g. models, actresses), [30]. This may elicit body image dissatisfaction, along with feelings of inferiority and shame [31, 32], particularly in women who present high levels of fears of compassion and low levels of social safeness [14]. Dietary restraint may ensue as a strategy to control body image, attain thinness, and enhance perceived low social rank (e.g. [32, 33]). In this sense, Ferreira and colleagues [23] point out that drive for thinness may be regarded as a competitive strategy to obtain a safe place in the social world (i.e. may be regarded as an insecure striving strategy).

Research shows that dietary restraint is highly prevalent among women (e.g. [34, 35]) and is associated with eating psychopathology [36, 37]. Dietary restraint becomes problematic when it presents as inflexible eating, i.e. the inflexible compliance with subjective eating rules, with disregard for external or internal contingencies, and generation of a sense of control when such rules are met, or feelings of distress when they are not [38].

In summary, considering that humans present a fundamental motivation to seek for social affiliation, mainly through attractiveness (e.g. [17, 19]), it may be hypothesized that individuals who fear compassion [7] may develop a shame-prone hypercompetitive drive to be attractive to others (insecure striving) [20]. Yet they may hardly feel accepted and connected to others, even if they succeed in such competition [20, 23]. Empirical research demonstrates that in women such drive is often body image focused, given that this domain is perceived as central in social rank estimation and enhancement [19, 27, 33]. Dietary restraint may be the elected body enhancement strategy, which may present as inflexible eating, shown to significantly predict eating psychopathology [36, 37].

The current study aimed at testing whether insecure striving and inflexible eating mediate the impact of fears of compassion on eating psychopathology’s severity. Although the relationships between these constructs have not been studied, it is expected that they will be associated positively.

Materials and methods

Participants

The sample of this study comprised 310 Portuguese women from the general population, with ages between 18 and 65 years (M = 26.85; SD = 9.84) and an average of 15.07 (SD = 2.09) years of education. The mean of the participants’ body mass index (BMI) was 22.31 kg/m2 (SD = 4.05). Thirty-three subjects (10.65%) were underweight (BMI < 18.5), 229 (73.87%) presented with normal weight (18.5 ≤ BMI < 25), 32 (10.32%) were overweight (25 ≤ BMI < 30), and 16 (5.16%) presented with obesity (BMI ≥ 30), [39]. The sample’s BMI distribution reflects the one found in the female Portuguese general population [40].

Measures

Body mass index (BMI). BMI was calculated using participants’ self-reported weight (kilograms) and height (metres), according to the Quetelet Index (kg/m2).

Fears of Compassion Scales (FCS; [7]: Portuguese version [41]). FCS comprises three self-report scales which measure specific fears of compassion: (i) fear of compassion for self (with 15 items) that measures fears of being self-compassionate (e.g. “I worry that if I start to develop compassion for myself I will become dependent on it”); (ii) fears of compassion from others (with 13 items) that assesses the fear of receiving others’ compassion and affiliative feelings or messages (e.g. “feelings of kindness from others are somehow frightening”); (iii) fear of compassion for others (with 10 items) that assesses fears of feeling and expressing compassion for other people (e.g. “being too compassionate makes people soft and easy to take advantage of”). Respondents rate their agreement to each item using a a 5-point scale (0 = “Don’t agree at all”; 4 = “Completely agree”). Higher scores reflect greater fear of compassion.

These scales presented good psychometric properties, with Cronbach’s alphas ranging from 0.78 to 0.92 in the original version of the scale [7], and from 0.88 to 0.94 in the Portuguese version [41].

Striving to Avoid Inferiority Scale (SAIS; [20]: Portuguese version [42]). SAIS is a self-report scale which measures the perceived pressure to compete to avoid inferiority. In the present study, it used the part one’s insecure striving subscale (which comprised 20 items) to measure beliefs about the need to strive and compete to avoid inferiority (e.g. “If I don’t strive to achieve, I’ll be seen as inferior to other people”). Respondents were asked to rate, on a 5-point scale (0 = “never” to 4 = “always”), how often they experience what is stated in each item. Insecure striving subscale presented very good consistency both in its original [20] and Portuguese versions [42], with Cronbach´s alphas of 0.92 and 0.91, respectively.

Inflexible Eating Questionnaire (IEQ; [38]). IEQ is an 11-item self-report scale that measures the inflexible adherence to subjective eating rules (e.g. “get worried when I do not follow my eating rules, even if it only happens occasionally”). Items were rated on a 5-point scale (ranging from 1 = “fully disagree” to 5 = “fully agree”), with higher scores reflecting heightened inflexible adhesion to subjective eating rules. IEQ revealed good psychometric characteristics in the Portuguese population (α = 0.95), [38].

Eating Disorder Examination Questionnaire (EDE-Q; [43]: Portuguese version [44]). EDE-Q is a 36-item self-report measure based on the Eating Disorder Examination interview that assesses eating psychopathology. EDE-Q comprises four subscales: restraint, eating concern, weight concern and shape concern. A global EDE-Q score can be calculated through a mean of these four subscale scores. The items were rated for their frequency or severity on a scale from 0 to 6, regarding the last 28 days period, with higher scores reflecting higher levels of disordered eating severity. This instrument showed good psychometric properties, demonstrating an α ≥ 0.94 in different study samples, for both the original [43] and the Portuguese versions [44].

The Cronbach’s alphas by these measures in the present study are presented in Table 1.

Table 1 Cronbach’s alphas (α), means (M), standard deviations (SD), and intercorrelation scores on self-report measures (N = 310)
Full size table

Procedure

This study respected all ethical requirements of research in human beings, according to the 1964 Declaration of Helsinki and its later amendments. The research protocol and proceedings were approved by the Ethics Committee of the Faculty of Psychology and Education Sciences of the University of Coimbra. Participants were recruited through online advertisements, private messages and a social network on Facebook. Potential participants were also asked to share this advertisement with two more friends (exponential non-discriminative snowball sampling method). In conjunction with the URL to access the survey, participants received a brief message about the study purpose, the voluntary and anonymous nature of their participation, and the confidentiality of the collected data, which would be used for research purposes only. Participants who agreed to take part in the study completed a secure online survey which included an informed consent, a demographic form, and a series of self-report scales described above.

Analytic strategy

IBM SPSS Statistics (version 22.0) was used to perform descriptive and correlational analyses, while the IBM SPSS Amos (version 22.) was used to perform the path analysis.

Descriptive statistics were calculated to explore the sample’s characteristics. The associations between fears of compassion, insecure striving, inflexible eating, disordered eating severity, BMI, and age were calculated using Pearson product–moment correlations and analysed according to Cohen’s guidelines [45].

Structural relations among variables were examined using path analysis. Specifically, links between fears of compassion (exogenous variables) and disordered eating (endogenous variable) were explored, as well as hypothetic mediator effects by insecure striving and inflexible eating (endogenous mediator variables), after controlling for BMI and age. Regression coefficients and fit statistics were computed using the maximum likelihood estimation method. The plausibility of the theoretical model was examined through a set of goodness-of-fit indices (CMIN/DF; TLI; CFI; RMSEA) [46].

Mediational paths’ significance was calculated using the bootstrap resampling method considering 5000 bootstrap samples and 95% confidence intervals around standardized estimated total, indirect, and direct effects [47].

Results

Descriptive statistics and correlation analysis

The descriptive statistics of the study’s variables (means and standard deviations) and intercorrelations for the total sample (N = 310) are presented in Table 1.

Results revealed that the three dimensions of fears of compassion (FCS_fromOthers, FCS_forOthers, and FCS_forSelf) presented moderate to strong positive associations among each other. Also, these fears of compassion presented moderate positive associations with striving (SAIS-Striving), and weak to moderate positive associations with inflexible adherence to eating rules (IEQ), and with EDE-Q subscales and global score. Moreover, results showed that striving presented moderate positive associations with inflexible eating rules and disordered eating (as measured by EDE-Q subscales and global score). IEQ presented strong positive associations with EDE-Q subscales and global score. Finally, BMI presented weak positive associations with IEQ and EDE-Q subscales and global score. Age only presented a significant association with fear of receiving compassion from others, which presented a weak negative association. Finally, EDE-Q subscales and global score presented very strong correlations between them (ranging between 0.78 and 0.94). As such, only the global score was used in subsequent analyses.

Path analysis

The path model was firstly tested through a fully saturated initial model (with zero degrees of freedom) which consisted of 44 parameters. The analysis of the path coefficients indicated that the five following paths were non-significant: the direct effect of fear of compassion for self on EDE-Q (bFCS_forSelf = 0.01; SEb = 0.01; Z = 1.04; p = 0.297); the direct effect of fear of compassion for others on EDE-Q (bFCS_forOthers = − 0.01; SEb = 0.01; Z = − 1.06; p = 0.289); the direct effect of fear of compassion for self on inflexible eating (bFCS_forSelf = 0.07; SEb = 0.07; Z = 1.10; p = 0.273); the direct effect of fear of compassion for others on inflexible eating (bFCS_forOthers = − 0.13; SEb = 0.08; Z = − 1.53; p = 0.126); and the direct effect of insecure striving on EDE-Q (bSAIS_Striving = 0.01; SEb = 0.00; Z = 1.74; p = 0.082). These paths were progressively eliminated and the model was readjusted.

Figure 1 represents the readjusted model, which accounted for 27, 23, and 48% of insecure striving, inflexible eating and EDE-Q’s variances, respectively, while controlling for the effects of BMI and age. Results indicated that all path coefficients were statistically significant (p < 0.050), and the model presented an excellent model fit to the empirical data [χ2(8) = 13,46, p = 0.097; CMIN/DF = 1.68; CFI = 0.99; TLI = 0.97; RMSEA = 0.05, (IC = 0.00–0.09; p = 0.491)].

Fig. 1
figure 1

Final path model

Full size image

Specifically, fear of compassion for self, fear of compassion for others, and fear of receiving compassion from others presented direct effects on insecure striving of 0.14 (bFCS_forSelf = 0.17; SEb = 0.08; Z = 2.06; p < 0.050), of 0.27 (bFCS_forOthers = 0.51; SEb = 0.11; Z = 4.82; p < 0.001), and of 0.23 (bFCS_fromOthers = 0.33; SEb = 0.10; Z = 3.27; p < 0.010), respectively. In turn, fear of receiving compassion from others and insecure striving showed positive direct effects on inflexible eating of 0.12 (bFCS_fromOthers = 0.13; SEb = 0.06; Z = 2.20; p < 0.050) and of 0.38 (bSAIS_Striving = 0.29; SEb = 0.04; Z = 6.84; p < 0.001), respectively. Moreover, fear of receiving compassion from others and inflexible eating showed positive direct effects on EDE-Q of 0.18 (bFCS_fromOthers = 0.23; SEb = 0.01; Z = 4.28; p < 0.001) and of 0.59 (bIEQ = 0.07; SEb = 0.01; Z = 13.61; p < 0.001), respectively.

The analysis of indirect effects demonstrated that fear of compassion for self and fear of compassion for others presented indirect effects on inflexible eating of 0.05 (95% CI − 0.01 to 0.13) and of 0.10 (95% CI 0.05–0.16), respectively, which were totally mediated by insecure striving. In turn, fear of receiving compassion from others also showed an indirect effect on inflexible eating of 0.05 (95% CI − 0.01 to 0.16), which was partially mediated by insecure striving. Furthermore, fear of compassion for self, for others, and from others presented indirect effects of 0.03 (95% CI − 0.05 to 0.08), 0.06 (95% CI 0.03–0.10), and 0.12 (95% CI 0.05–0.19) on EDE-Q, respectively, which were mediated through the mechanisms of insecure striving and inflexible eating. Finally, insecure striving presented an indirect effect of 0.22 (95% CI 0.14–0.31) on EDE-Q, which was totally mediated by inflexible eating.

Overall, after controlling for the effect of age and BMI, results revealed that the impact of fear of compassion for self and fear of compassion for others on the engagement in disordered eating attitudes and behaviours was totally carried by the mechanisms of insecure striving and inflexible eating. Regarding fear of receiving compassion from others, results showed that its effect on EDE-Q’s variance was partially mediated by these two mechanisms.

Discussion

Recent research has been evidencing the negative impact of fears of compassion on disordered eating and on treatment outcome [9, 10, 48]. Furthermore, Oliveira and colleagues [10] suggested that the fear of compassion from others may contribute to the engagement in extreme control over one’s body and eating. In this sense, the current study sought to explore the role of insecure striving and inflexible eating in the impact of fears of compassion on disordered eating in a sample of Portuguese women from the general population.

Correlation results revealed that all constructs associated positively, as hypothesized. Some of these results corroborate previous empirical research: the association of fears of compassion with disordered eating attitudes [9, 10], the association of insecure striving [23, 25], and of inflexible eating with disordered eating [36, 37]. Other correlation results provide unprecedented empirical observations: the positive associations of all dimensions of fears of compassion with insecure striving, and of insecure striving with inflexible eating.

All associations were further explored performing a path analysis which tested a mediator model where insecure striving and inflexible eating were hypothesized to mediate the impact of fears of compassion on disordered eating’s severity, while controlling for the effect of BMI and age.

Path analysis results revealed that the model presented excellent fit to the empirical data explaining 48% of disordered eating’s variance. Results revealed that, while controlling for the effects of BMI and age, the impact of fears of compassion for self and for others on disordered eating was completely carried by insecure striving and inflexible eating. This result seems to support that women who fear compassion may engage in insecure striving strategies focused on body image, such as inflexible eating, which may contribute to the development of disordered eating. Unexpectedly, the impact of fear of receiving compassion from others also presented direct effects on inflexible eating and disordered eating. One possible explanation is that these relationships may be mediated by another emotion regulation process, namely self-criticism, which has been positively associated with fear of compassion from others [7, 8], and potentially lead to interpersonal-focused strivings and achievement-focused striving [49].

The present study presents some limitations. Its cross-sectional design prevents conclusions regarding causal directions between the variables. Future studies should explore the associations between these variables using a longitudinal design. The use of self-report measures only is also a limitation, as it might have produced biases. Upcoming studies should include other types of measures, such as interviews, and counterbalance self-report measures. Moreover, this study’s sample only comprised women from the general population; replication of this study using clinical and male samples is thus recommended for future studies. Finally, this study’s scope was limited, not apprehending confounding variables that could be involved in the examined associations. As such, upcoming studies should explore the role of additional variables, such as social safeness, self-criticism, and body compassion.

The main contribution of this study consisted of the empirical observation of the association between fears of compassion and the engagement in insecure striving strategies. This finding suggests that fears of compassion may prevent women from experiencing affiliation signals [14, 17, 18] and thus become highly attuned to a social rank mentality, feeling they must compete to attain social acceptance and connectedness [20, 24]. However, these strategies paradoxically elicit perceptions of threat, failure, and negative affect [20, 24]. Furthermore, these strategies may be focused on body image [23, 25], which may present as inflexible eating [36, 37]. In turn, the latter is a predictor of disordered eating attitudes and behaviour, as previously evidenced in the literature [36, 37].

Fears of compassion may not just be involved in the development of disordered eating, as they may represent major blocks to the effectiveness of eating disorders treatment [9, 48]. As such, eating disorders treatment should involve an assessment of these fears. If present, one may use compassion interventions, such as compassion cultivation training program or compassion-focused therapy (CFT), which were empirically shown to significantly reduce these fears and promote soothing abilities [50, 51].