1Since the nineteenth century, various associations between health and social class have been gradually brought to light. All or almost all show that the social classes best favoured in terms of income, level of education, occupational prestige are also those best favoured in terms of health, that it is as measured by life expectancy or by more recent indicators such as disability or perceived health. Thanks to the studies on gender and their progress, sex is increasingly treated as a dimension of social position with consequences for health. In fact, health conditions differ between the sexes and contribute to inequalities between women and men. Age and state of health, for their part, are still very rarely the focus of a sociological approach, as the biological dimension of health seems to massively outweigh the social dimension. In this respect, age is not understood as generating social inequalities between age groups, since health disparities are mostly not explained by social mechanisms but by bodily ageing, and thus rejected from the social sphere. Yet there is a field of health in which age has been more often approached in its social aspect: mental health. This is particularly the case for depression or suicide whose link with different age groups was very early identified (Durkheim  1997). In fact, age and sex are more amenable to mental health conceptualization and interpretation as social markers. In general, differences in the prevalence of mental tension between social groups draw attention to the reasons for it and then play the role of a sociological alarm bell about social tensions concentrated within specific segments of the population. Thus Durkheim drew his cardinal concepts of regulation and integration from the higher number of suicides among single, divorced and widowed people.
2Although there is a long tradition of studying health inequalities between social classes, the traditional approach is not appropriate for eating disorders. In fact, inequalities in health are often presented as the result of inequalities in other areas of life: education, income, working conditions, access to care, etc. (Bihr & Pfefferkorn 2008, Aïach 2010). Inequalities in access to socially valued resources are embodied (Fassin 1996) and lead to disparities in terms of life expectancy, the disproportionate prevalence of disease, disability, and the feeling of being in good health. etc. Assuming that many mental tensions reflect the stresses of the social world, it is to be expected that the most disadvantaged sections of society will also be the most affected. And indeed, depression, schizophrenia, suicide, dependence on alcohol or other psychoactive substances are more often identified among the lower classes. In this sense, inequalities in mental health also logically contribute to the accumulation of social inequalities.
3The pathologies of eating behaviour, however, escape this analytical framework since they are more frequent among the more privileged strata of the population. In the case of anorexia nervosa and bulimia nervosa, the differences in prevalence between social classes have the inverse gradient to that usually observed: they affect girls from the affluent classes much more often than those of the middle classes and finally the working classes. These unusual disparities sociologically contest an overly linear interpretation of mental disturbance as being the result of wider tensions in the stratification of the social world from which the higher social groups generally come out as winners. Nor can they be considered as the consequence of unequal access to scarce resources since—in the case of anorexia—it is a question of restriction of consumption. Understanding inequalities in the prevalence of anorexia nervosa and bulimia nervosa as a result of social relations of class, gender and age makes it possible to relax the constraints brought about by using the narrow perspective of class inequalities. By diverting our gaze from the conventional context of inequalities to look at particular expressions of the relationship with food, with the self and with others, this apparent sociological “anomaly” compels one to look for a model that is less substitutive than complementary to the underlying social tensions that they express.
4The prevailing explanation for anorexia is that of a “hyper normative adherence” with harmful consequences, to an excessive compliance with the social criteria of the desirable female body, particularly that of thinness, criteria which would be equally applicable to anybody. Now, if anorexia seems to flow with relative ease into this logical mould, how does bulimia enter it? Bulimic behaviour—the compulsive absorption of food—can hardly be understood just as a simple search for thin-ness and therefore of a desirable body. Another form of interpretation is required.
5Faced with the various stresses of life, which build up and /or suddenly arise, eating disorders (ED) are pathological responses that tend to be adopted more frequently by young women of the affluent social classes. We need to understand the reasons why this group, located at the labile cross-roads of three fundamental social relationships, more often responds to tension by anorexic or bulimic practices, and what are the tensions produced by their social positions.
6The sociological analysis of these pathologies makes it possible to overcome the limitations of common sense and to show that the norms and values of distinction encountered in the well-to-do classes—the injunction to excellence and self-control—can be affected in a pathological manner. The hypothesis developed here is less reliant on the need for conformity, which is a shift towards the disease, than a pathological use of these norms (cf. infra). In addition, although young girls are much more likely to be affected, this is not the result of female genetic predispositions since no research has been able to prove their existence (Godart et al. 2012). The argument of a potential biological reason does not fit well with the social gradient displayed by these pathologies except to imply a decreasing biological deficiency according to social class … Therefore, it is necessary to come back to a simultaneous sociological analysis of class differences, of positions of sex in this particular moment of “adolescence” where authorisations and obligations can be distinguished from each other.
Social relations and eating disorders
7By social relations, following Danièle Kergoat’s work and Philippe Zarifian’s interpretation of it, we mean a mode of regular relations between the individuals of different and hierarchical social groups, structured around issues, generating conflicts and antagonisms, and taking three canonical forms: exploitation, domination, and oppression (Kergoat 2005, 2009, Zarifian 2010). Kergoat identifies three fundamental social relations: gender, class, race, which she calls “consubstantial”. They are nested within each other, rather than independent, and they form and produce dynamic, complex and mobile configurations and interactions. To these three fundamental relationships,  it is necessary to add a fourth and equally fundamental relationship: the social relations of age , based on sociological studies highlighting social stratification by age. Although life courses are presented today as de-institutionalized, de-standardized, non-linear (Beck 2001), so that it becomes impossible— and foolish—to set thresholds for the different life-stages, it remains no less the case that being 17 years old today in France is not associated with the same roles, the same norms, practices, daily activities, social status, possibilities and expectations, etc. as being 47 or 87 years old. In this study, the period we are interested in is that when individuals enter and are involved in secondary education, which is conventionally called “adolescence”. It constitutes at the same time a socially identified position in the current course of ages and a specific age group, involving relationships of economic, material and social dependence on adults, while benefiting from a relative cultural, relational and emotional autonomy, etc., in relation to children.
8The set of social relations is concretized in particular by practices specific to each position in social space, reflecting differentiated socialization, and by interactions with others involving relations of domination, exploitation or oppression. Compared with an approach to health inequalities in terms of inequalities of resources, the use of social relations does not presuppose a given gradient of health inequalities. Insofar as social relations are constitutive of practices and interactions that do not necessarily have health implications, it is theoretically conceivable that these practices and interactions have deleterious health effects on those who are at the cross-over points with higher positions.  The concept of social relations is particularly relevant and adapted to the analysis of the particular social profile of those people affected by anorexia nervosa and bulimia nervosa.
9These pathologies constitute two “eating disorders,” according to the medical classifications currently in use.  Anorexia nervosa is characterized by weight loss that is not of organic origin and which goes below the threshold of thinness, a refusal to gain weight and a distortion of the perception of corpulence. Bulimia Nervosa combines very substantial and repeated food intake over a short time and weight control techniques, via vomiting or food deprivation. These controls are also observed in anorexia nervosa. These two disorders therefore overlap with common practices and concerns around weight control. In addition, although they can occur independently of each other, they can also be combined. The most common pattern is the onset of bulimic behaviour as a result of anorexic behaviour. Finally, they have similar epidemiological profiles. These observations justify, if necessary, the benefits of a combined study of these two disorders
10Anorexia nervosa, first described medically in the early 1870s,  was initially described as a disease of upper-class girls (Lasègue 1873, Gull 1874). The age, sex and social categories are also present in the older versions of the International Classification of Diseases and the DSM. The epidemiological profile seems hardly to have changed. Clinical and general population studies still report a marked preponderance of women among patients, often from higher social backgrounds, and early onset of symptoms in youth, often between the ages of 15 and 20 (Hudson and Hiripi 2007; Preti et al. al. 2009). Adolescent girls in higher social categories may therefore be more likely to adopt anorexic practices. Medically recognized much more recently, bulimia nervosa (hereafter “bulimia”) appeared late, in the 1980s, in diagnostic textbooks and classifications. Incidentally, this lag in the emergence of scientific interest in bulimia was based on the social appeal of pathological fasting, associated with the emergence of ideas about asceticism, dark romanticism, ethereal elegance, spirit against matter (Brumberg  2000), that created a lack of interest, and more often a contempt, for the excessive consumption of food that could lead to guilty carelessness, vulgar voracity, and stigmatizing a lack of self-control revealing an inner moral bankruptcy. Restraint is therefore associated with a moral judgment rewarding those who do not take part in the “excess” of ingestion of food.  By an apparent paradox, which calls for analysis rather than avoidance, the sex, age and social class profile of bulimia sufferers hardly differs from those suffering from anorexia nervosa.
11Like other expressions of malaise (suicide, depression, anxiety, etc.), anorexia nervosa and bulimia affect only a minority of the population. As with the sociology of suicide, the sociology of eating disorders could be the sociology of an exception (Baudelot and Establet 1984). If the sociology of suicide as those of the EDs has a wider scope than just the population of suicides or people affected by the EDs, this is because the social experiences generating tension and leading to these expressions of malaise, go above and beyond the context of those people who die by suicide as well as those people who develop anorexic and /or bulimic disorders alone. In the case of suicide, the tensions experienced eventually decide some of those who suffer from them to commit suicide. It is one of Durkheim’s great merits to have unearthed the protective factors within populations and then to have included them in a unifying theory whose scope far exceeds the case of suicide. As anorexia and bulimia involve, repeatedly and overwhelmingly, girls who mostly come from the middle and upper classes, they pose sociological questions about the reasons for these pathologies that reach beyond the experiences only of those individuals affected by these disorders.
12Anorexia nervosa is usually interpreted as excessive conformity to the norms and values of thin-ness for girls, facilitated for the upper classes by their eating habits (Darmon 2003, 2009, Guillemot and Laxenaire 1997). In this perspective, socialization in terms of food and diet is the central explanatory base. This interpretation has several limitations. It does not understand differences in the prevalence of anorexia in terms of inequality, because disorders are not included in specific social relationships. Although this interpretation explains the occurrence of disorders in girls and mainly those from the upper classes, it does not explain why disorders occur most often in adolescence. It is even poorer at explaining the frequent association with bulimia, with which anorexia nervosa maintains many points in common, except for the social fascination that the latter exerts when the former is stigmatized. Focusing on food norms and practices, it does not account for the other behaviours usually associated with anorexia nervosa. Indeed, dietary restrictions are only one aspect of the anorexic syndrome and “excessive” involvement in intellectual and physical activities have been reported for a long time (Lasègue, 1873, Bruch 1973/1994) and confirmed since (e.g. Brumberg 2000 , Taranis & Meyer 2010). Finally, this interpretation does not shed light on the reasons why the behaviours to which some people have been socialized can be perceived as pathological, and appear much more as a transgression of norms than conformity with them.
13The argument that anorexia nervosa conforms to andro-centred norms is frequently used to support the common idea of an “epidemic” of anorexia nervosa, due to the expansion of thinness standards and the increasing attractiveness of female bodies that are slender and light. However, if one observes a lowering of desirable weight thresholds (ESCAPAD 2002–2008, Saint Pol 2010), no one seems to be able to provide evidence of an increase in the prevalence of anorexia nervosa over time (Van’t Hof and Nicolson 1996) but only of an increase in the demand for care. The prevalence of anorexia nervosa among women, like that of bulimia, remains low. In Europe, it is around 0.9% for each condition (Préti et al. 2009) and between 0.9% and 0.5% respectively in the United States (Hudson et al. 2007). If anorexia can be attributed to an excessive conformity with the dominant aesthetic norms to the point of becoming deviant, its low prevalence would be surprising … This is probably because anorexia is not a simple conformity with norms of thinness, even where it is “excessive,” but that it represents more of a “pathological use” which goes beyond the mere realm of slenderness.
14Eating disorders are part of a process that constitutes our framework of analysis and that we will present in a linear rather than strictly chronological fashion. Schematically, individuals are exposed to stressors  whose frequency and nature vary according to their position in social relations of gender, class and age because of the interactions in which these social relationships manifest themselves. The current weakening of the patriarchate is accompanied by a renewal of the differentiation of the sexes, modulated by age and social class, and forms of oppression and domination of girls, including through the control of their bodies. Young girls are thus less exposed than boys to certain forms of physical violence but more exposed to sexual violence (Cavalin 2007). They are subject to stricter parental control (Bozon 2008), are less satisfied with their appearance, and made to feel inferior in social interactions (Dornbusch et al. 1984), and have lower self-esteem than boys from the beginning of adolescence and up to entry into adult life (Block and Robins 1993, Mahaffy 2004). In addition, various forms of mediation (social support, denunciation, collective action, ritual, stigmatization, reinterpretation of stress …) are likely to reduce or accentuate the effects of stressors.
15Two types of responses may emerge from the process of interpreting and mediating stress, which are not necessarily disconnected and may coexist for a certain time. These responses can help to dissolve the effects of stressors or, on the contrary, maintain them and affect mental health. In this second case, the responses are “unsuitable”. The practices adopted (anorexic, bulimic in the restriction phase) bring some benefits (relaxation of tensions, feelings of control and status) but they do not help to resolve the conflicts, and put an end to the mental stress. They are therefore both perpetuated and even strengthened, or even become a dead-end and involve practitioners in a vicious circle. For example, in a context where a person is in a submissive relationship and experiences the feelings of inadequacy which result from it, the “answer” can reside in excessive control over food thus demonstrating a valorising will (control) over a valued object (a slender body). The motivating factor for this behaviour is thus to provide a sense of self-control and to improve self-esteem as a result. In contrast, restrictions of diet do not terminate the initial stimulus. Stressors cannot, therefore, be completely obliterated by the responses of food deprivation, so that deprivation appears “necessary” but is never sufficient. In a general way, the stresses find their answers in the pathological vocabulary acquired during the socialization of the individuals according to their socio-demographic characteristics, in other words in the specific practices that will crystallize their malaise: food deprivation, “bulimia crises,” alcohol consumption, drugs, scarification, etc., or the triggering of a pathology without practice (depression) or even suicide.
16This analytical model posits as its initial hypothesis that the inequalities of prevalence of EDs are explained sociologically by the intersection of the social relations of age, gender and social background, with which they are associated
17—A control over the body, in its aesthetic and sexual dimensions, which is exerted preferentially on women, especially when they are young, and their inferiority, reinforced for the young and those in well-off circles. Inferiorisation and control of the body can be internalized and motivate certain practices but also become the sources of stress;
18—Signs of distinction and social classification used differently according to social class;
19—Sources of stress for adolescents, especially girls, over which stressors gain an increased impact, coming from the conjunction of the transition from a normative system of obedience to one of empowerment and the disruptions caused by unsatisfactory body changes.
20Through an intersectional analysis, the article uses this model to show how these specific social relationships promote eating disorders.
Data, indicators, analytical approach
21The main source of data is the “Health and Consumption Survey” (Enquête sur la Santé et les Consommations: ESCAPAD). It was collected in 2008 during the Call for Preparation for Defence (Appel de Préparation à la Défense), by the French Observatory of Drugs and Drug Addiction (OFDT; Observatoire Français des Drogues et Toxicomanies). It was the first in France to report anorexic and bulimic symptoms nationwide. Self-administered, the questionnaire includes a module on anorexic and bulimic symptoms. This survey has two unusual advantages: firstly that it was focused on the general population of young people, that is to say a sample unbiased by the social selection induced by use of the health system; and on a large population that meant it was possible to identify infrequent eating disorder behaviour.  The sample consists of 39,546 17-year-old French respondents, comprising 19,658 girls and 19,888 boys. To our knowledge, this survey is currently the only one internationally able to have such a large sample of adolescents and allowing a retrospective evaluation of the prevalence of anorexic and bulimic disorders. Other eating disorders are listed in the epidemiological manuals but as ESCAPAD does not identify them, only the two central eating disorders were studied.
22This module was developed by a multidisciplinary team based on the criteria of the Diagnostic and Statistical Manual of Mental Disorders IV (DSM-IV). It identifies the occurrence of the four symptoms that characterize anorexia nervosa: refusal to gain weight, fear of gaining weight, distortion of body image (these three symptoms must be simultaneously reported), and amenorrhea (absence of periods) during at least three months for girls already having them; as well as age, height and weight at onset of disorders. The “yes” and “no” options were proposed for the questions below:
The whole of this module makes it possible to distinguish several degrees of severity of anorexia, including or excluding the criterion of amenorrhea and according to the body mass index (BMI) achieved : threshold  less than 18.5kg/m²: less than 17kg/m². In this article the thinness threshold chosen is 18.5kg/m², unless otherwise stated, in accordance with one of the specifications used by Godart and Legleye (2013), and amenorrhea, dependent on gender, is not taken into account as proposed by Attia and Roberto (2009). This last criterion has indeed disappeared from version 5 of the DSM. The probable anorexia nervosa identified by ESCAPAD is described as sub-syndromic because defined by the least restrictive diagnostic criteria (see below). Finally, it is sometimes associated with one of the types of bulimia.“Since the age of 12, have you experienced a period of time during which?
1. You were afraid of gaining weight, of becoming too fat;
2. You have lost a lot of weight or you refused to gain weight;
3. You felt uncomfortable with yourself because of your weight or your figure;
4. Have you been treated by a doctor or hospitalized for anorexia nervosa?
5. If you are a girl: did you ever stop having your period for at least 3 months (excluding pregnancy);
Then came three questions: “If you answered “yes” to one of the previous questions, answer the following questions:
1. How old were you at this time? …… years
2. During this period what was your height? … m … cm
3. During this period what was your lowest weight? … kg
23According to the DSM-IV, bulimia nervosa is characterized by a very substantial food intake over a short time, a feeling of loss of control over eating behaviour, an “excessive” concern for the control of weight resulting in the establishment of modes of compensation, and an excessive influence of weight and figure on self-esteem. In addition, the disorder does not occur only during episodes of anorexia nervosa. The modes of compensation being used by the sufferer distinguish between bulimia with “purging” (induced vomiting, laxative or diuretic intake) and bulimia with “restriction” (episodes of severe self-restriction of diet and/or intensive sport). The ESCAPAD questionaire module on bulimia opens with a filter question:
Finally, the recent occurrence of anorexia (in the vast majority of cases, one or two years before the survey), and probably bulimia (there is unfortunately no age indicator for the onset of this disorder), makes it possible to relate them with a reasonable approximation to the current situation concerning, for example, the school curriculum and mental health, especially since these disorders generally continue for several years (Hudson et al. 2007). In the United States, half of those identified were still clinically ill five years after the first diagnosis (Steinhausen 2002). Like the series of ad hoc questions, which are more likely to identify “depressiveness” rather than “depression,” whose diagnosis remains the responsibility of physicians, ESCAPAD identifies symptoms indicating probable anorexia nervosa or probable sub-syndromic bulimia nervosa, but does not perform an individual diagnosis of the disorders. To make the rest of our article more easily readable, the terms “anorexia nervosa” and “bulimia nervosa” are used in this restrictive sense. Although the approach adopted in this article is not a criticism of medical categories, we will avoid any naturalization of disorders by not referring to “anorexic” or “bulimic” people but to the disorders and practices, of anorexic or bulimic behaviour presented by individuals.“Since the age of 12, have you ever eaten very large quantities in a short time, without being able to stop (bulimia attacks)?”
In the event of a positive response, the questionnaire asked eight questions to help identify a probable bulimia:
1. Have you had bouts of bulimia several times a week for at least three months?
2. Have you had bouts of bulimia only when you were too thin or when others said you were (anorexia)?
3. Have you been treated by a doctor or hospitalized?
4. Were you in pain because of your weight or your figure?
5. Have you had severe diets or stopped eating?
6. Have you done a lot of sport to avoid getting fat?
7. Have you taken laxatives or diuretics?
8. Have you made yourself vomit so as not to gain weight?
24The analyses are also backed by the broad body of epidemiological and sociological research on ED published in scientific journals. The advantage of this approach is to improve the depth of the analysis while having rigorous data. For this we use the convergent results of several publications as much as possible to reinforce our findings.
An unexpected social profile of disorders
25Between the ages 12 and 17, anorexia and bulimia are overwhelmingly feminine (97% and 91%) to the point where the very rare male cases (20 out of nearly 20,000 …) seem to indicate exceptions that we would need to clarify if there was sufficient information available. With a prevalence of 3.1%, girls are 31 times more affected by anorexia than boys (Table 1). As for bulimia, it represents a disorder only slightly more frequent than anorexia nervosa, affecting 3.7% of girls and 0.4% of boys. Purging (vomiting, laxatives, etc.) is the most frequently used method (60%) of the means of compensation for bulimic attacks compared to dietary restrictions or intense sports activity. These results do not support the widespread image of an “epidemic” of anorexia or bulimia, since these two disorders remain relatively rare even among the most-at-risk adolescent population.
Prevalence of disorders between 12 and 17 years*,**
|Girls (n=19,658)||Boys (n=19,884)||Relative risk|
|of which restrictive||275||1.4||43||0.2||7|
Prevalence of disorders between 12 and 17 years*,**Field: all respondents (age 17).
Note: * Refusal, Fear, Distortion, <18.5 kg/m2.
** Relative risk Girls/Boys = Prevalence girls/Prevalence boys.
26The over-representation of EDs among women, observed from ESCAPAD, appears even greater than that reported in other studies in the general population: 85% of women for anorexia nervosa (Garfinkel et al. 1996) and 90% for bulimia (Bushnell et al. 1994, Garfinkel et al. 1995). However, these surveys were conducted in North America with a sample of older people.
27Could the medical and social construction of ED in terms of female disorders contribute to low male prevalence by biasing responses? On the other hand, a complementary methodological study shows that boys tended to self-declare themselves lighter than their measured weight (Legleye et al. 2014). This low male prevalence should not mask higher frequencies among sportsmen, dancers, models, etc., where slenderness and bodybuilding are consubstantial to their profession. According to Chambry, the male rates of anorexia would be around 0.18/100,000 inhabitants year in psychiatry and 0 to 0.09% in the general population (2015). However, the gender gap seems far too large to be explained by these types of effects.
28Be that as it may, there are no other pathologies so sensitive to the social distinction of the sexes (apart of course from sexual pathologies such as breast cancer or prostate cancer). It is therefore in the expression of mental tensions, and specifically in eating disorders, that gender inequalities culminate in health.
Trouble is my business: classes and ED
29The EDs are significantly related to social background  and display a gradient where the highest prevalence is observed among upper-class girls (manager and higher intellectual profession, artisan, trader, entrepreneur), then among those from the middle class (intermediate occupation, employee) and remains the lowest among working-class girls (manual, unemployed, unskilled) and farmers (Table 2). Although net, the social gradient narrows more for bulimia. ESCAPAD reveals wide differences between social classes and confirms the reversal of prevalence inequalities compared to most other health inequalities. As the data was drawn from the general population, the results show that this unusual social trend is not based on a selection effect, which usually affects clinical data.
Prevalence of eating disorders among girls according to social class of parents
Prevalence of eating disorders among girls according to social class of parentsField: girls of 17
Note: * Refusal, Fear, Body image distortion, <18.5 kg/m2, ** Relative risk.
Remark: The number of boys with EDs is too small for a comprehensive study.
30Do the disparities in the declaration of the disorders come from the differences between classes in terms of perceptions of the normal and the pathological? It must be recognized that the borderline between these two notions fluctuates according to the various actors involved in the labelling of the disease (doctors, epidemiologists, “patients,” families, etc.) and results from their possible interactions. ED labelling is therefore likely to move in social space as well as over time , and to be associated with inequalities in recourse to the medical profession according to class. From this perspective, the higher prevalence of high-class EDs should be due to better screening rather than higher prevalence. The use of a logistic regression makes it possible to rule out this hypothesis because no statistically significant link is revealed between the medical remedy (medical treatment or hospitalization) and the social background of the individuals in our sample (results not presented). The labelling of anorexic disorders by the medical authorities is therefore insufficient to explain the inequalities of prevalence observed among girls according to social class.
31For 17-year-old girls, the average age of onset of anorexia nervosa is 15.6 years and the median age is greater than 15 years. According to other retrospective studies of a general population over 17 years of age in the US, half of the cases occur before the age of 18 and 50% of the cases are between 16 and 22 years of age (Hudson et al. 2007). This study indicates that bulimia begins more often before age 18 with slightly more dispersed initiation ages (inter-quartile range of 14 to 22 years). The EDs therefore start mostly in adolescence, or even when young people have recently become legally of age, but mostly at ages when most girls are still in education, reside in the parental home and are dependent on their parents, with all the constraints that this situation induces (Figure 1). 
Age at onset of anorexia nervosa (%)
Age at onset of anorexia nervosa (%)Field: girls with symptoms of anorexia
Body norms and social relations
32The questioning of the social norms of thinness is contained in the very definition of the EDs being studied since they are medically associated with an “excessive” concern with weight and thinness. This normative challenge is only relevant if we accept that people labelled as anorexic or bulimic adhere to these norms, like the other members of the social group to which they belong. The “ideal” body mass index (BMI) makes it possible to verify these claims empirically. This is conventionally defined by the BMI equalizing the proportion of people judging themselves to be too thin or much too thin, and that of people judging themselves to be too fat or much too fat (Saint Pol 2010).
Disparities and distances from ideal bodies
33Gender differences are the most striking result among those distributed by class and age (Figure 2). Indeed, the ideal BMI of girls is much lower than that of boys (BMI 18 vs 21.4) and is even below the thresholds defining thin-ness for girls (BMI of <18.25 between 17 years and 17.5 years of age, BMI of <18.38 between 17.5 and 18 years of age ) (Cole et al. 2007). It is remarkable that the “ideal” BMI of all girls at the end of adolescence corresponds to that of a body considered medically thin and whose weight is below that of “normal” corpulence.
Ideal BMI and BMI reported by sex and social background for 17-year-old
Ideal BMI and BMI reported by sex and social background for 17-year-oldField: All respondents
Note: Sup: senior manager, artisan-trader, entrepreneur; Ouvr: worker; Sans: unemployed, not given; Ens: together. F: girl, G: boy.
34Quite logically, boys and girls are distinguished not only by their expectations about their figure, but also by the possibility of satisfying them. While the declared BMI and ideal BMI are almost identical for boys, the gap averages 2.7 for girls … Boys feel overwhelmingly that their weight is correct (64%) and the rest of the answers are balanced overall between the opinions “a lot or a little too thin” and “a little or a lot too fat”. This is not the case for girls. Only half of them (51%) feel they weigh the right weight and if not, nearly 90% perceive themselves to be “a little or a lot too fat.” With a BMI of 21, one in two girls feels she is too fat as against less than one in ten boys. More than one in ten boys thinks they are too thin.
35These different results have the merit of objectifying the differentiation of body norms organized by gender and testify to the valuing of thinness in girls. They suggest the probable tensions to which they are subjected because of the frequent and often important difference between the real and ideal body. The ideal of corpulence for girls is, strictly speaking, a model which only a few women are able to satisfy. On the other hand, the masculine model seems less demanding, more pragmatic, in line with the experience of the majority of them. In addition, girls present a tendency to generalized dysmorphophobia. Whatever their weight, they often perceive themselves as too big, echoing the Duchess of Windsor’s formula: “A woman is never too rich, nor too thin.” Their dissatisfaction with their weight refers to the dissatisfaction with their body in general, in which weight has become of major importance.
36In addition, the valuation of thinness is more characteristic of the daughters of managers and entrepreneurs (BMI 17.9) than of workers (BMI 18.3). These results are consistent with the observation that, from adolescence, the slim body is valued more by women, and even more so by those at the top of the social hierarchy (Saint-Pol 2010). The latter are closer to the ideal figure (the smallest difference between ideal BMI and declared BMI), which illustrates their efforts and the effectiveness of their techniques in the pursuit of this desirable body. The results blatantly demonstrate that ideals and the distance to the ideal are differentiated primarily by gender, and then staggered according to social background, mainly for women. Differentiation by social background is much less important among young men.
37The first organizing principle of bodily norms—as the most powerful factor of differentiation—lies in the social relations of gender. If relations of sexual attractiveness do preside over the imposition of female body norms, basically they do not account for the place of thinness in our society … In addition, they only partially explain the connections between gender relationships and class relationships. The fruitful way of resolving these questions is through the radical abandonment of explanations in terms of thinness by considering what is being socially underpinned in the standards of thinness and valued by the affluent classes: self-control.
Slimming and relationships of sexual attraction
38If men value a muscular and powerful body, which is neither too fat nor too thin, then thinness is on the contrary desired by women. This difference highlights a gendered construction of bodies. Muscle is however not absent from female models. It eliminates the hateful fat and characterizes in this respect the feminine lipophobic models of the twentieth century (Fischler 1993, Poulain 2009, Saint Pol 2010, Vigarello 2010). Female muscle must be located at the level of the “abdo-glutes” rather than the shoulders, the arms and the thighs. For women, the issue about the muscles of the waist and hips is sexual attraction, while for men the issue of a broad build is about the display of strength and virility (Louveau 2007).
39As a factor in sexual attractiveness, thinness without fat is perceived by women as necessary to succeed in their love relationships in a heterosexual context and studies showing the correlation between body-mass and marital status prove them right (Poulain 2009, Amadieu 2005). Even more than social class, thinness offers real benefits and punishes those who cannot achieve it. Men are penalized for being underweight (Saint Pol 2010). However, their height and social status are much more important qualities and forms of capital in forming couples (Bozon 1990, 1991). For them, possessing good capital in the search for a spouse seems all the more important as the injunction to form a couple remains a powerful one. Unmarried women are perceived as at risk (or by being one), and women’s social status remains partly defined by that of their spouse, with the opposite being less true.
40The work of Michel Bozon underlines the dissymmetry in the criteria in the choice of a spouse. Men prefer partners who respond to the canons of physical beauty, who are thus able to increase their social prestige. Women favour partners who are equal or greater in age, size and social position. In relations of sexual atraction, not only is the hierarchy of man and woman clearly established, but the female body has an importance which the male body does not. This body does not have to be endowed with physical strength, visible working ability, etc. but to be socially and sexually desirable. It appears even more clearly to be the vector of male domination when the modalities for achieving a desirable body are examined. If adult men can hardly affect their own height, women can and must act on their weight. It must be worked, whether it is to be thinner, or fleshier, or thinner here and plumper there. They are therefore required to shape their body so that it is desirable and attractive because it has relatively rare qualities.
41This injunction might well be stronger in adolescence but for reasons, paradoxically, that do not only relate to attractiveness. In fact, bodily appearance takes on an unprecedented importance in the definition of social status, which is to be understood in the Weberian sense of prestige or social esteem (Weber 1995 ). In this learning period, when individuals cannot derive prestige from their occupation and from their income as they would in adulthood, the classifying properties are, more than for adults, appearance and cultural tastes. For middle and upper-class girls, thinness is also part of classifying properties (Pasquier 2005). As a result, bodily demands will be higher for middle- and upper-class girls than for adult females because they would be able to determine a desirable social status. From this, it is necessary to draw this central observation: sexual relationships are only one of the explanatory elements of the search for thinness. The latter is also a source of social prestige for internal and external use (see below). It is a model of pragmatic self-appreciation of oneself and a model in itself. The thinness norms accepted by girls affected by the EDs are therefore related to both heterosexual seduction relations and the major role of body appearance in the definition of self and social status in adolescence.
Food control, self-control and gratification
42Thinness is not usually perceived as the result of social inequalities deriving from socialization to a balanced and low calorie diet, or due to physical exercise, or as a result of social privileges acquired by the upper classes. It is interpreted as “an opportunity,” a gift, or something deriving from individual responsibility. Being able to act on corpulence is supposed to be something that can be exercised mainly through control of food intake. The person who is able to control her diet is slim and, by analogy, someone who is able to control herself. In affluent societies, hunger and satiety are among the bodily urges to be mastered and thus constitute a fundamental dimension of self-control. Guillemot and Laxenaire emphasize that people diagnosed with anorexia present “almost constantly subjective associations between thinness and self-control on the one hand, obesity and loss of control on the other” (Guillemot and Laxenaire 1997: 57). In this respect, there is an abundant epidemiological literature highlighting the control concerns associated with anorexic practices (Ibid Sassaroli et al. 2008, 2011, David & Button 2011, Bruch 1973) and other eating disorders (Katzman and Lee 1997). What is valued and contained in the search for thinness is the ability to control eating and, beyond that, to exercise self-control. This self-control is particularly expected of women, whilst on the other hand men have the privilege of controlling others (Mc Sween 1993). To meet the standards of thinness, girls must “teach the body” to incorporate the norms of food and self control in a habitus that predisposes them to the main symptom of anorexia, namely an excess of food control (Darmon 2003). Self-control provides external gratification by displaying a desirable thinness made visible to all and a symbolic manifestation of self-control. It is coupled with an internal gratification by the satisfaction obtained by the success in controlling oneself. As seen earlier, these symbolic rewards can materialize with advantage in the choice of a spouse and the formation of a couple. The disparities of control over bodies thus constitute an explanatory dimension of the inequalities of prevalence of anorexia nervosa according to gender and social background.
43Anorexic practices therefore draw on socially distributed predispositions to food control. How then should we introduce bulimic practices to this theoretical framework? Bulimia attacks are characterized by very substantial food intake in a short time and without the person being able to stop. In fact, they can be interpreted as an extraordinary relaxation of the control that people usually exercise over their diet, a breakdown in this control system for various reasons and through psychological mechanisms (e.g. anxiety about failure, discouragement linked with depressive symptoms, etc.). In this logic, people who rigidly control their diet will be more willing to binge and indeed this risk is increased for women, especially from the higher classes.
44The key role of the internalised norms of food control is confirmed by the implementation of methods for the compensation for excess. Bulimia has two closely related phases, that of relaxation in the bulimic crisis, which breaks the strong habitual control that people exercise over their diet; and that of the “resumption of control” which is exerted by vomiting, laxatives, diuretics, dietary restrictions or other methods of compensating for relaxation. Contrary to common sense, bulimia is strongly linked to norms of perfectionism and self-control (Vohs et al. 1999), but this disorder imparts an oscillating movement between relaxation and recovery, leading from a crisis of bulimia to a crisis of restriction or purging. Often stigmatized because of the lack of control it seems to manifest, bulimia, in terms of the findings of epidemiological research, is actually favoured by the internalization of norms about self-control (Ibid.).The consubstantial research into thinness and self-control helps to understand why thinness has become a sign of social distinction: it ranks women and places them in a hierarchy. For women from a working class background, an instrumental use of the body in occupational and domestic spheres, favouring a enveloping and muscular body; for women of the higher classes, an aesthetic use, or even one of attraction, requiring a shaping of appearance (Bourdieu 1979). In fact, physical appearance depends on the objective conditions of existence, but in a highly tertiarized society, where strength only plays a modest part, it depends upon and serves above all to classify women socially by acting as a factor of distinction. The division of labour, not only between men and women, but also between social classes among women, makes thinness a distinguishing feature of upper-class women and places it in female social class relations. Thibaut de Saint Pol offers an explanation of what might be called the “distinction of weight”: “the “civilizing process” described by Norbert Elias, which originally explained the evolution of bodily manifestations such as sputum or table manners, can also apply to body shapes. One of the reasons for the new attention paid to weight at the beginning of the 20th century may well be the desire to define new criteria of distinction that separate the better-off from the rudeness and sloppiness of the poorest “(Saint Pol 2010, 102). This hypothesis needs to be extended by articulating gender relations and class and age relationships. If weight is a factor in the social status of men (Amadieu 2005), it “weighs” much more on the social status of women, especially in adolescence. In the end, it is the young women of the dominant classes who most often display a valued figure, because of the division of labour between them and between the sexes.
45These representations of feminine thinness are effective: as a sign of a dominant position, a dominant social position. In addition to the marital market, thinness is profitable in the labour market, as thin women can more easily access a hierarchically higher job (Ibid.). Young girls from the upper classes benefit from their social background during their studies and on the labour market because they have inherited the educational, cultural and social capital of their parents (Bourdieu and Passeron 1970). They also benefit from their corpulence, which is a real “bodily capital,” that is to say, an investment which is expected to return increased profits. Access to a lifestyle and a diet inclining towards thinness, diet control provisions, enhancement of body aesthetics, and the higher demands they derive from them, lead them to be slimmer than young girls from a working class background. Thinness thus participates, in a hardly visible manner, in the reproduction of the social structure.
Excellence: a poisoned chalice
46Other practices that had been described very early in the clinical records, are often associated with dietary restrictions and weight control. These are the intensive practice of intellectual activities and / or physical activities (Lasègue, 1873, Gull, 1874, Bruch 1973, Guillemot and Laxenaire 1997, Brumberg 2000 ). The epidemiological and psychiatric literature continues to show links between EDs and physical over-activity (Beumont et al. 1994). Sports activities can, of course, be interpreted as means of compensation with the aim of eliminating the calories ingested but commitment to intellectual activities thwarts this logic. The search for social excellence is the common denominator of these intense activities and the concern for thinness. Anorexia and bulimia are especially restrictive and thus can be understood as an imaginary form of success and social excellence, through attributes that are not just physical but also intellectual  and moral. In this respect, Hilde Bruch observes that “hyperactivity and the efforts made by patients to achieve ideal fulfilment […] are essential aspects of anorexia nervosa” (Bruch 1973: 321). Girls affected by anorexia commit themselves, not necessarily in a calculating or conscious manner, to an exaggerated mastery of their bodies, to working hard at school, and self-control practices (Halse et al. 2007). In return, they enjoy the feelings of success produced by these forms of control that are about asserting social prestige, or at least one that is fantasized about (there is a distortion of the perception of the body, and a “too” thin body is less desirable than stigmatizing) and this encourages their continued efforts in this direction. In an individualistic society where the status of individuals is defined primarily by their personal qualities, and achievements (much less by their lineage or social background), where subjects feel they are called upon to be “a distinct and powerful self” (Ehrenberg 1998), individuals are classified in “higher” social status if they show signs of social excellence. Academic success, as much as physical appearance, is a particularly sought after asset in upper class families and a marker essential to the social status of adolescents. Anorexic practices, both of food deprivation and intense intellectual activity, thus appear in step with the norms and values of their class.
47This prestigious ranking, these symbolic benefits, produce an increase in self-esteem  that, for various reasons, can be weakened in people with anorexia (Godart et al. 2013). Therefore, the process favouring anorexic “deviance” can begin. The objectives set are always increasingly ambitious and tend to induce a constant feeling of dissatisfaction with what is accomplished. The results never meet the expectations, as they are always one-off and likely to be questioned and, if they can be measured, are never enough. Psychologists call this form of functioning involving unrealizable goals and self-dissatisfaction “perfectionism” and observe it more frequently in people with anorexia and bulimia (e.g., Vohs et al. 1999, Lilenfeld et al. 2000, Bastiani et al. 1995, Castro et al. 2004, Bardone-Cone 2006).
Responses to mental stress
48Anorexic and bulimic practices are labelled as pathological and socially deviant. If based on group norms, they go “too far,” and cross the threshold (quantitative and qualitative) of social conventions separating the normal from the pathological or the socially accepted behaviour from deviance. According to the model presented at the beginning of this article, practices of control or relaxation of control over food, and intense educational commitment are likely to be part of a deviant trajectory when they constitute inappropriate responses to stressors. These practices provide, in the first instance, some benefits but the “amount” varies according to the pathology and the compensation methods used. Nevertheless, these benefits remain unsuited to the resolution of tensions, so that without external intervention (medical, psychiatric, etc.) a vicious cycle leads the person to sink deeper into the disease and sometimes to their death.
Malaise and the quest for prestige
49The concomitance of EDs and various other psychiatric symptoms argues for an explanation in terms of the pathological response to stress. Suicidal thoughts, suicide attempts, extreme depressive symptoms, the consumption of antidepressants, anxiolytics, and hypnotics are significantly higher for girls with ED than for others (Figure 3). These indicators confirm that anorexia and bulimia are forms of mental pathologies and not merely forms of normative judgments imposing a single normality of food behaviours that should not be deviated from. Depressive symptoms associated with anorexia nervosa easily go unnoticed due to intense intellectual or physical activity that counteract the usual slowdown observed in the case of depression.
Mental health indicators by type of ED
Mental health indicators by type of EDField: 17 year old girls.
Note: Depressive symptoms are estimated based on a scale designed for adolescents, the ADRS: Adolescent Depression Rating Scale (Revah-Levet al. 2007)
50The underlying logic of ED and compensation modes is striking (Figure 3). The implementation of dietary restrictions in anorexia or bulimia is accompanied by lower levels of prevalence of indicators of malaise, appearing to reveal a beneficial psychological impact of restriction. Thus, the prevalence of almost all the signs of malaise or poor mental health that the study has available is increasing:
- girls without ED,
- in case of bulimia compensated by restrictions,
- in cases of strict anorexia nervosa,
- anorexia associated with bulimia compensated by vomiting or other purges,
- bulimia with purges.
51Girls using purging are significantly more affected by different mental health problems than girls who engage in restriction, regardless of the nature of the ED. The mode of compensation for bulimia attacks is more important than the anorexic or bulimic disorder itself. Using a socially disregarded mode of control such as provoked vomiting is sanctioned by a more intense malaise than when a more rewarding mode of control like the restriction of food is used, which testifies to the successful exercise of willpower. Compensation methods thus have a differentiated return in terms of well-being, because of their implicit social stigmatization and especially by the exercise of the rewarding control of hunger. However, these remarks seem insufficient to sociologically explain better mental health in cases of restrictive bulimia than strict anorexia, the latter escaping the stigma of bulimia attacks.  Should we thus recognize that although these two disorders proceed from the same rationale, affect similar populations, maintain frequent interactions, they remain nonetheless two distinct pathologies whose initial levels of malaise are hierarchized unequally and where anorexia is more pathogenic (and rarer)?
52Because they are part of a valued mastery over the self, restrictions will reinforce and increase an abusive and deficient self-esteem (Sassaroli et al. 2008), especially among girls (Block and Robins 1993, Mahaffy 2004). Thus we understand better that the pursuit of prestige has as its intimate motive the search for self-esteem through the acquisition of a desirable figure, by hard work, by various markers of excellence, finally by the power to exercise self-restraint. Thus, “corpulence is the object of issues whose horizons are at the same time self-esteem, occupational success and success in love” (Poulain 2009: 124). The practices that are used provide various benefits to those who adopt them, which explains the difficulties in disengaging from them.
Stressors and mental tensions
53Two sociological models compete to explain the factors that promote disease. The first uses social demands and pressures (in terms of the body, self-control, etc.) and amounts to considering the symptoms of ED as a slide in the pathology due to an “excessive” conformity to norms. The second tends to identify the ED as unconscious and pathological uses of the norms and values to which individuals have been socialized, as so many responses to stressors and which implicitly aim at a desirable social classification, the exercise of the control over one’s own existence and the restoration of self-esteem. These models are understood simultaneously in order to propose a general framework of interpretation, coherent for both anorexia nervosa and bulimia nervosa. This sociological interpretation is obviously not intended to define the aetiology of these pathologies.
Violence and ED
54Girls with anorexia nervosa or bulimia appear to be more exposed to various violent or conflictual social interactions. In particular, girls who have anorexic symptoms more commonly complain about being in a bad relationship with their father and mother (where parental social class, school curriculum, repeating a year’s education, cohabitation with parents are controlled). In addition, they have been more often assaulted in the past 12 months (Table 3). For those with bulimia we can add a parental misunderstanding and being involved in a fight. For circumstantial reasons, they are less concerned by issues about agreement with their mother (they do not know her or have no relationship with her, or she has died). While girls of higher social class are less likely to be exposed to these family difficulties and violence (results not shown), those affected by EDs, mainly recruited from these classes, appear on the other hand more likely to be confronted by this than the others.
Stressors and eating disorders (logistic regressions)
|Relationship with father: good||Ref.||Ref.|
|Relationship with mother: good||Ref.||Ref.|
|Relationship between parents: good||Ref.||Ref.|
|Has been in a fight (ref. no)||1.1||1.3*|
|Has been attacked (ref. no)||1.5***||1.7***|
|Has been threatened (ref. no)||1.8***||1.9***|
Stressors and eating disorders (logistic regressions)Field: girls of 17
Note: 1: Refusal, Fear, Body image distortion, <18.5 kg/m2
* p < 0.05; ** p < 0.01; *** p < 0.001
Control variables: SEG of parents, defined by the occupational category of the parents (as shown in Table 2), school curriculum, repeating a school year, living with parents.
55Of course, it is impossible from cross-sectional data to determine the temporal succession of events. Anorexic and bulimic practices could as easily be a response to violence, conflict and family difficulties, as the opposite. Thus dietary restrictions can lead to conflict with parents when they can no longer be hidden;  thinness, and deviant behaviours can be subject to ridicule but also a morbid response to them (Menzel et al. 2010) making girls vulnerable to aggression. However, other studies show that women who have experienced violence, including rape, are at greater risk of developing psychiatric disorders, including ED (Waller 1991, Vize & Cooper 1995, Rayworth et al. 2004), which seems to support the hypothesis of responses to stressors.
Adolescent mental stress
56Bodily changes at puberty are often implicated in the development of anorexia nervosa at these ages. At the macro level, the appearance of ED does not coincide with puberty as defined by age at first menses. In France, this average age was 12.6 years in 1994 (the median age being 13.1 years; cf. de la Rochebrochard 1999) whereas the periods of girls identified as anorexic in ESCAPAD begin most often after 16 years.
57Periods are, however, only a stage of puberty and corporal changes continue beyond this point. Hairiness, breasts, body fat continue to develop and more in girls than in boys. On average, the latter gain more muscle mass, the former more fat mass, while on average, the bodies of boys and girls are not different before puberty. The contemporary disapproval of fat (Vigarello 2010) is not socially neutral. It is even more effective now that women are expected to work on their bodily appearance. Since the 1920s, the focus has been on cellulite, a form of fat considered feminine, and against which women are supposed to fight, particularly by physical exercise (Ghigi 2004). The passion for bodily effort, noted in the cases of anorexia and bulimia (Gull 1874, Bruch  1994), is thus a reflection not only of the search for thinness but the rejection of a fat that could paradoxically come to lodge itself even in thin female bodies. It is not only bodily changes that are at stake, but the social interpretation of masculine and feminine forms of differentiation and their inclusion in asymmetric gendered relationships that are capable of generating internal stresses. Studies have thus highlighted the greater dissatisfaction with their bodies of teenage girls, in particular because they see their fat mass—something that is particularly devalued—increase (Dornbusch et al. 1984, Park 2003).
58Body maturation is not just a biological development. It also has a significance in terms of social status. The importance of bodily appearance in the definition of status in adolescence, already emphasized above, is one of the aspects of the influence acquired by peers in judgments about oneself (and by oneself) (Mardon 2010), in the adoption of norms and values that are sometimes at odds with those transmitted by parents. From the distances, or even oppositions, between normative family and friendship systems—school and media-based—inevitable tensions will arise. In addition, the status of being an adolescent requires more autonomy, and involves making choices that engage one’s responsibility and future as a progressively independent individual (in terms of educational path, affirmation of sexuality, etc.). Given the normative tensions that these changes sometimes bring about, adolescents can opt for safe and dominant norms and values, by submitting to the expectations of the school, and those of parents and peers. While they are enjoined to assert their “identity” through their choices and achievements, they relate themselves to an “ideal self,” excellent in body, education and ultimately in their job. However, the pursuit of excellence contains the seeds of constant dissatisfaction because it simultaneously contains a gap between the lived self and the ideal self.
59Parental guardianship, which is physically burdensome and restricts autonomy in areas where aspirations may be strong, is another source of tension (Bruch 1994 , 299). Again, despite a generational rapprochement in the education of both sexes, girls remain more closely supervised by their parents. Going out alone, in particular, is more often strictly controlled, and frequently still prohibited before the age of 18 (Bozon 2008). The restrictions imposed by their parents, the unequal treatment according to the sex of the children can be a source of conflicts, but can especially give girls the feeling that they have less control over their existence. The EDs are responses by which they can re-establish close control over a part of their existence: food, body appearance and ultimately “one-self”. 
60If these tensions about the definition of standards, the acquisition of control, etc. concern a wide adolescent audience, the EDs are much more selective and circumscribed. Why? The general idea underlying the markers of this form of ill-health is that the intensity and the diversity of the tensions of the social world and their accumulation end up being objectivised in various expressions of mental tensions. Thus, it is necessary to take eating disorders, like other mental disorders, as so many specific responses to stressors of different natures, and of varying intensities. They come in the form of shock (e.g. rape, break-up, death of a loved one) and/or in a more insidious and repetitive form (e.g. parental criticism, peer mockery, humiliation, family conflict, etc.) and life-cycle disturbances (in this case adolescence) which, in combination, prepare the ground for ED installation. The pathological responses to these stressors are produced by individuals interwoven in social relationships defined mainly by gender, age, and social class. These influence their exposure to certain stressors and determine the values and norms of their group(s) and consequently their “pathological vocabulary,” that is to say their modalities of response to stress. For example, in France, suicide is three times more likely to involve men (especially elderly men), while depression is twice as often experienced by women (with variations according to age), anorexia affects more than 95% of young girls, and so on. This logic makes it possible to understand why the EDs, like suicide, go beyond the sociology of an exception and recount the singularity of the social world as the tip of an iceberg whose essentials seem hidden.
61Even though the EDs recruit more frequently among the relatively privileged classes, young women of the lower classes are not spared. The values of excellence which prevail in the higher social classes, do not stop at their borders. Firstly, because the dividing lines between the different social categories are unclear and interpenetrate at the points of contact, not only because of the nature of the capital that they can avail themselves of—economic, cultural—and coincide only partially, but also by their relative spread in a part of the working classes. The aspiration to academic success is frequent in the upper classes but is not strictly circumscribed. Anorexia and bulimia are sometimes developed by young boys, by adult men or women and even children. The analysis is based on what a given position generates in the gender, class and age relationships, but their effects can be observed (more rarely) in other positions. The modes of corporal socialization are sometimes out of the ordinary, especially in the work carried out on appearance and sports practices. The very high inequalities in the prevalence of ED between the sexes nevertheless suggest that this gender atypism is rare. The class one is less so.
62* * *
63In the end, where does the sociological study of anorexia nervosa and bulimia nervosa lead? It leads first of all to validating the relevance of the simultaneous taking into account of the social relations of class, gender and age in order to understand the reasons why these mental pathologies are not “mad” ones but obey a logic that sociology helps to unveil. Danièle Kergoat is right to insist on the dynamics of social relations, their contextualization, their historicisation. Our study confirms her intuition when she suggests that to the three fundamental social relations, gender, class, and “race,” one must add age (and generation) (Kergoat 2010). The configuration they take in eating disorders is both singular in terms of the relationship with age and the relative importance of each of these social relationships in these disorders.
64Thus, even by going beyond the reductive vision of these EDs by the imposition of andro-centered norms on the female body, gender relations take precedence over class ones in this context. The differential between bodily socialization and appearance induces a drastic control of feelings of hunger and widens a gap between men and women, that is much more marked than that between social classes. This is hardly surprising since body differences are thought of in our culture as the obvious and founding element of distinctions between the sexes. The education of women requires a control of their diet, a permanent restriction to thinness and thus the preservation or increase of their sexually attractive capital. The slimming of women over time probably reflects a shift in expectations about gender relations. The modern criteria of distinction have moved away from a robust peasant corporeality and roundness. In affluent societies (at least in terms of food), full bodies now mean looseness, “junk food,” bad taste and further stigmatize the lower classes. Thinness makes women’s bodies desirable because it is associated with the upper classes and is an element of enviable distinction. Not all women are equally thin and do not place the slimming slider at the same point. It is important to emphasize, especially since this is rarely done, that the social differences in corpulence and ideal corpulence are relatively high in women compared to those observed in men. To the extent that body fatness and control are important women’s issues, they become criteria of social distinction between women, external signs of prestige or elements of their status—when male status is even more signalled by the socio-occupational position and external signs of wealth. The role of corpulence in the social status of adolescent girls appears preponderant when it undergoes strong transformations reclassifying in the eyes of all the sexual operability of their bodies. Along with academic results, slenderness means excellence. It relaxes its grip when women later enter an occupation that will weigh more in the esteem given to them, and to a conjugal and family status that reduces the stakes of sexual attraction.
65Corpulence, food control and self-control do not take on the same meaning according to the social relations from which they are considered and this is the second achievement of this study. The emphasis placed on women’s corpulence and self-control therefore appears as an instrument of men’s domination over women, who are constrained—explicitly or not—to shape their bodies to conform to male desire. The search for thinness has perverse effects because it maintains the representation of “fragile” women, who are physically weaker than men, with the implications of their “natural inability” to perform certain tasks, the “need” to protect them or inversely the possibility of physically dominating them. By shifting the configuration of social relationships, when considering thinness and control within class relations between women, they become instruments of superiority. Women who display these attributes of success are more likely to be perceived as belonging to the dominant classes and to gain access to them. According to the relationships in which these practices and signs are involved, they currently mean to be dominated or to be dominant. They cannot therefore be understood in a general social context but in the context of the plurality of particular social relations. If they were only associated with dominant positions, the EDs could be interpreted as “costs of domination”. Associated at the same time with dominated positions, and of young dominant women, because they come from well-to-do classes, they are in fact carriers of a strong contradiction of meaning for those who are at the intersection of these social relations. Analyzing how this contradiction is perceived, undergone or thwarted according to the circumstances opens a way for promising qualitative investigations.
66Finally, the findings suggest that the range of life norms from which adolescent girls must define a desired and feared “autonomy,” the injunction to make choices, the maintenance of parental control, the negative views focused on their bodily changes,  and the violence encountered in their lives, generates tensions to which the EDs provide answers, by commitment to rewarding practices, excellence and control over oneself. In other words, although the norms of thinness, excellence, etc. are internalized during primary socialization, they are activated and consolidated during adolescence with the transformation of sociability networks, the increasing influence of peers (Mueller et al. 2010), the injunction to autonomy, and choices to be made for the future. In the case of anorexia, the behaviours of young girls seem to firstly conform to the norms in force, but in reality they exceed them without any explicit desire for transgression. In bulimia, they are moreover hidden, and repaired in more or less gratifying ways. The work of bodily appearance and control (sensations, nutrition) appear above all as a way of acquiring a form of prestige particularly “adapted” to girls, because their self-esteem is undermined by their inferior position in the social relations of gender and age, because they make of their body and its control two major dimensions of their status at the moment when their body changes meaning and becomes a subject of dissatisfaction. So we are not dealing with a simple normative script that runs uniformly on a reel throughout life but one that is activated in the unique conditions of adolescence. If these socially desirable qualities—thinness, control, excellence—lead to eating disorders, it is not because of an extreme conformity to these norms but because of a pathological use, coming from a need to re-acquire a self-esteem undermined by the convergence of various types of stress. Mental pathologies or certain expressions of malaise (suicide, for example) play a role in revealing social tensions in complex configurations. Analysis in terms of social relations is particularly adapted to this complexity and thus makes it possible to better understand why these social tensions affecting a priori favoured categories adopt this particularly singular pathological form.
Translated by Peter Hamilton with the support of CNRS-INSHS.
“Race” relations are not dealt with here in the absence of adequate information about this subject in the survey.
Classically, age effects combine with generational effects (for example, at the same age, the generation of baby boomers may show specificities that following generations do not have) and period effects (wartime, post-war boom, etc.). Nevertheless, adolescents from different periods share the experience of parental dependence, even though this may have been modulated or even transformed over time.
This perspective has the merit of updating the analysis of the life of women over the long term and makes it possible to understand it as the product of social relations in which they are dominated.
At the date of the survey, these classifications came from the 10th revision of the International Classification of Diseases of the World Health Organization (ICD-10) and the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) of the American Psychiatric Association. Other and sometimes more common eating disorders, are not discussed here, such as those referred to as the “Eating Disorders Not Elsewhere Classified” (FED-NEC). These disorders have close links with anorexia and even more so with bulimia (Fairburn et al. 2009).
Much older cases, notably that of St. Catherine of Siena (†1380), have been reported. However, they enter the field of medieval mysticism and associated practices including extreme mystical fasting (Maître 2001). It would be risky to equate mystical fasting with contemporary anorexia nervosa because of the misinterpretation that such an anachronism could generate after more than six centuries.
For “thinness” (minceur), the French dictionary of synonyms in Word offers: finesse, lightness, grace, beauty, distinction, elegance, charm, softness, insight, intelligence, sagacity, subtlety, etc.
The notion of stress is frequently used in the English-language sociology of mental health and differs from the common sense meaning it has in France. “Stressors” are negative stimuli encountered during life that in turn lead to mental stress that can lead to psychological or somatic pathology (Aneshensel 1992, Pearlin 1989). These stressors are for example a period of unemployment, harassment at work or in private life, financial difficulties, a relationship breakup, etc. The definition of the French term “tension” (i.e. “stress” in English) in the CNTRL, CNRS or Petit Robert dictionaries is similar: it is a “psychic state where the need for relaxation is felt.” The French and English definitions of stress and are therefore relatively close. A mental tension (a stress) represents a psychic state that induces a reaction of variable intensity. However, “tension” can also be understood as “specific tension” generating stress (a “stressor”) creating a blurred feeling. These terms are used interchangeably in the rest of the article to avoid ambiguity.
ED is used restrictively here to refer to anorexia nervosa and bulimia nervosa.
Body-mass index is the result of the formula weight/height2 and is expressed in kg/m2
These thresholds correspond respectively to the “underweight and thinness” thresholds defined by epidemiologists for people aged 18 and over; thresholds are lower below age 18 and vary by age and sex (Cole et al. 2007). It is these thresholds, adjusted for age and sex, that were used to rank respondents.
This is based on the highest socio-occupational category of the parental couple.
For example, the level of weight loss suggestive of anorexia nervosa has been lowered between DSM-III and DSM-IV to promote diagnosis in children. DSM-V now excludes the amenorrhea criterion. One can also point to the strategies of deviant dissimulating behaviours within the home, of the role of the publicity given to anorexia and of the social norms of the group in the labelling of deviance (Darmon 2008).
The module on anorexia nervosa was the only one to include a question about age at onset of the disorder. The rate at each age is termed its incidence.
After age 18, the threshold of 18.5 is common to both sexes.
Escapad reveals that among girls in school, those who have experienced anorexia are more often in the general and technical streams than in occupational streams (80.8% compared to 71.8%), and that they repeat a year of classes more rarely (24% versus 35.6%). This may also be due to the statistical selection from the more affluent social classes linked simultaneously with academic success and ED.
There are many definitions of the concept of self-esteem, but overall they refer to “the positive or negative overall individual attitude towards self” (Simmons et al. 1973).
The confidence intervals between these two categories overlap, indicating that there is no statistically significant difference (Figure 3). However, the systematic gradient (five out of six indicators) suggests that this ordering is probably not random.
It must be added that disagreements with parents deprives them in addition of their social support that is recognized as a protection from expressions of malaise.
Hilde Bruch notes that: “The need for autonomy and independence that all adolescents come up against seems to provoke an insoluble conflict within them, after a whole childhood as an obedient robot. They are not aware of their own resources and they do not trust their feelings, their thoughts, their physical sensations.”
Having an attractive appearance at age 14 is even negatively correlated with increasing self-esteem at age 23 (Block and Robins 1993).