Anne-Sophie Cousteaux } Laboratoire de sociologie quantitative CREST Timbre J350 3 av. Pierre Larousse 92240 Malakoff  01.41.17.57.35  [email protected] Jean-Louis Pan Ké Shon } INED 133 bd Davout 75980 Paris Cedex 20  01.56.06.22.63  [email protected]

The Relationship between Gender, Living Arrangements and “Ill-Being”: Apparent Contradictions in Suicide, Suicidal Tendencies, Depression and Alcohol Problems Please do not cite or quote without the authors’ permission. Comments welcome.

ABSTRACT “Ill-being” is the result of the tensions that arise between an individual and society. The intensity and nature of these tensions are many and varied, and ill-being adopts various forms according to the players, who themselves possess social particularities that make them to varying extents subject to a given tension. Each gender has its own way of expressing its ill-being: suicidal tendencies and depression mainly affect women, while suicide and alcohol dependence are more often male. Any interpretation that focuses on one of these expressions is likely to be erroneous. Whereas the divergences between these expressions reveal the particularity of the various forms that ill-being adopts and its differential impact on specific human groups, the convergences between them produce sound conclusions that can be generalised to individuals’ ill-being, whatever the particular markers used. By comparing gendered indicators two at a time, a new view is provided of greater female protection against suicide and the protection traditionally ascribed to the couple and children.

Suicide is the most radical expression of emotional distress. Faced with an unbearable situation the individual deliberately chooses death, a solution that appears to them to be the way of ending their suffering. This is not the only way of expressing distress, and there may be others. In Tocqueville’s chapter on the “causes of the restless spirit of Americans in the midst of their prosperity” he notes that “[c]omplaints are made in France that the number of suicides increases; in America suicide is rare, but insanity is said to be more common than anywhere else. These are all different symptoms of the same disease.” (Tocqueville, 1840, vol. 2, ch. 13). If Americans do not often commit suicide, says Tocqueville, it is not because of their greater prosperity. It is condemnation by religion that largely prevents suicide, but religion cannot solve what makes life unbearable for Americans. Their distress takes another form, mental illness, which religion cannot directly address. Some factors that protect

1

against suicide, such as religion, do not for all that protect against distress in general1. It is tempting to apply Tocqueville’s comparison to men and women. Suicide and alcohol dependency are mainly male, but attempted suicide, suicidal tendencies and depression mainly affect women. The idea of women’s greater well-being, based on their relative immunity to suicide, is hard to defend. This apparent contradiction between the various expressions of emotional distress indicates in fact that each expression has its own specific features, one of which is gender. Consequently, interpretation is likely to be erroneous if it focuses on one of these expressions rather than the others. Only a parallel study of the various expressions will provide a satisfactory account of the differences in well-being between the genders and enable valid conclusions to be drawn. Any study restricted to a sole expression, such as suicide, is more informative about that particular indicator than about emotional distress in general (Aneshensel, Rutter and Lachenbruch, 1991). The study of gender differences requires therefore the examination of various expressions of distress, some of which are traditionally male, such as suicide and alcohol dependency, and other mainly female, such as serious suicidal tendencies and depression (Aneshensel, Rutter and Lachenbruch, 1991; Horwitz and Davies, 1994; Horwitz, White and Howell-White, 1996; Simon, 1998; Simon, 2002; Umberson, Wortman and Kessler, 1992; Umberson et al., 1996). Parallel examination of these various forms of distress makes it possible to reject explanations of greater natural depressive tendencies among women, based on convergent, repetitive observations. It is important to take up a critical stance towards statistical results that tend to reify the data as evidence of a more delicate female constitution, confusing cause and effect. Indeed Lovell and Fuhrer’s review of the literature (1996) demonstrates a greater female prevalence of affective disorders and anxiety states, and at the same time a higher male frequency for disorders linked to drug consumption and antisocial behaviour. One reason for female emotional distress more often taking the form of attempted rather than actual suicide may be the internalisation of various gendered habitus, such as women’s repugnance towards violent acts. Indeed the ineffective, less violent methods women employ have often been used to explain the different suicide rates of men and women. In this case, it would be useful to inquire why women fairly systematically choose ineffective methods. Gender differences in suicide methods also overlap with gendered inequalities in access to those methods. For example, men have more frequent access to firearms in their jobs (security, police, army) and hobbies (hunting, shooting); women, more subject to depression, have access to tranquillisers, which are the leading method of attempted suicide for both sexes (Davidson and Philippe, 1986). But above all, although some attempts are “failed” acts, most suicide attempts correspond to the expression of a social phenomenon other than suicide. Attempts are more often a desperate call for help to cope with overwhelming distress than an intention to kill oneself. They are less a refusal of life than an “intense need to ‘live differently’, even at the cost of dying to make themselves understood” (Davidson, 1986, p. 152). With Halbwachs, we should remember that “nothing proves intention, nothing proves that the victim knew that their act would result in death, except the fact that they carried it through” (1930, p. 66). Actual and attempted suicide are therefore distinct expressions of distress, one male and the other female. Out of a corrected estimate2 of 13,000 suicides and 195,000 attempted suicides in 2002 (Mouquet, Bellamy and Carasco, 2006), three-quarters of the suicides were male, and two-thirds of the attempted suicides were female (Davidson, 1986; Badeyan et al., 2001; Mouquet, Bellamy and Carasco, 2006). There are also fifteen times as many attempts as actual suicides, which demonstrates an incomparable disparity and a clear separation between the two phenomena. On the one hand, depression may be seen as an internal distress expressed in private life, traditionally reserved for women. Conversely, alcoholism is more an externalisation of distress, compatible with men’s public 1

We have checked this using a logistic model (not shown here) with data from the French Health Barometer (see below). Controlling for other characteristics, the practice of and feeling of belonging to a religion has no significant impact on serious suicidal tendencies. The effect of religion on other indicators could not be tested, since the health survey does not provide this information. 2

Under-reporting was estimated at roughly 20%-25%, so corrected figures for 2003 would raise the reported number of 10,660 suicides to roughly 13,000 (Mouquet, Bellamy and Carasco, 2006). 2

role (Horwitz and Davies, 1994; Williams, 2003). These dichotomies between internal and external, private and public, and therefore female and male are strikingly present in the social representation of female alcoholism as a solitary secret practice linked to neurosis (Clément and Membrado, 2001). Furthermore, the hypothesis that would reduce the contradiction between the higher rates of suicide and alcohol dependence among men and of attempted suicide and depression among women is that there is a gendered expression of distress. In that case, the social construction of gender underlies the observed differences between the sexes in their ways of expressing distress. On the male side, there is emotional retentiveness, aggressiveness, acting out distress by violence, breaking the law, risky, defiant behaviour, plus the social role of taking on the responsibility of family breadwinner and more generally the behaviour that corresponds to representations of manliness. On the female side, there is the management of home life, children, social relations, qualities of gentleness and delicacy, the expression of feelings, self-realisation via family success in forming a couple and having children (notably Belotti, 1974; de Singly, 1987; Bourdieu, 1998; Baudelot et al., 2003). These distinct values attached to male and female structure people’s identities and their most private behaviour, including internal conflicts. It is hardly surprising that reactions to various negative situations and stimuli adopt the forms appropriate to the sex concerned. It is this gendered construction of value that Margaret Mead observed in Oceania when she writes, “What originally was a slight difference in temperament became under social influence an essential and inalienable characteristic of one sex. Children were to be brought up according to this standard: boys must overcome their fear; girls may show it openly.” (Mead, 1963, [retranslated]). On this hypothesis, individuals’ distress is expressed in behaviour that conforms socially to the gender they belong to. Although the differences between the indicators show primarily the various ways distress may be expressed and reveal specific human groups, conversely the similarities validate conclusions that can be extended to individual distress. Men commit suicide more than women. There are few observations in sociology that are so regular. The higher male suicide rate is a fact that has been established since the first studies published in the 19th century and it is observed in virtually every country, with the notable exception of China (Baudelot and Establet, 2006). After Durkheim’s cautious start, the tradition of analysing gender differences in suicide has neglected to explain this basic division between men and women and concentrated on the conflicting interests of the two sexes in marriage. Despite a systematically gendered analysis in examining marital status, Durkheim devotes little space to the reasons that might explain the gendered difference in the “tendency to suicide”. He briefly puts forward women’s lesser participation in social life as one reason for their immunity against suicide. This unconvincing explanation is clearly disproved by the fact that the gap between male and female suicide rates has remained the same despite women’s gradual entry into the labour market. Durkheim prefers to look at the differential benefits of marriage for the man and the women, and shows that the “conjugal society” formed by the couple is mainly beneficial to men. Observing that childless married women commit suicide more often than unmarried women, he states that “in itself conjugal society is harmful to the woman and aggravates her tendency to suicide” (Durkheim, 1897). Women’s relative preservation, he claims, is only ensured by the presence of children and therefore by the integration3 of “domestic society” more than by marriage. The conflicting interests of the members of the couple are clearly apparent in the opposing effects of divorce on marriage. The introduction of divorce reduces the coefficient of preservation against suicide for married men compared with the unmarried, but it increases the protection of married

3

Durkheim attributes the benefits of integration to two complementary features: the density of the family and collective sentiments. He writes, “But for a group to be said to have less common life than another means that it is less powerfully integrated; for the state of integration of a social aggregate can only reflect the intensity of the collective life circulating in it. It is the more unified and powerful the more active and constant is the intercourse among its members. The conclusion we reached may therefore be completed as follows: just as the family is a powerful preservation against suicide, so it preserves more effectively the more strongly it is constituted” (1897, pp. 213-214). 3

women. Durkheim fails, however, in his explanation of women’s excessive matrimonial regulation4 by putting forward a theory that naturalises male and female sexual desire. As a critical reader of Suicide, Besnard returns to this “unfinished theory” and examines the fatalistic suicide5 of married women due to the high social expectations attached to female roles in marriage: “in fact it is not only childless married women (a negligible quantity in Durkheim’s eyes) but married women as a whole who suffer the consequences of excessive regulation, even if the presence of children does partly compensate for the ill effect of matrimonial discipline” (Besnard, 1973, p. 41). But, like Durkheim, he does not attempt to explain female immunity and even admits his “inability to imagine a plausible sociological explanation” (Besnard, 1987a, p. 138). Although male and female conflicting interests in conjugal society are a crucial issue, nevertheless the disadvantaged position of married women cannot logically explain their greater immunity to suicide, rather the reverse. The marriage effect introduces only relative and limited differences between the two population groups, where the initial absolute difference is of quite another magnitude. It is primarily this question of female immunity to suicide, the main difference between men and women, that Baudelot and Establet consider when they formulate the hypothesis of women’s protection being linked to their greater family integration: “in France, the woman is statutorily more committed to family relations than the man. Statutorily more integrated” (Baudelot and Establet, 1984, p. 101). Unlike men, women’s family integration depends less on their marriage but is maintained throughout their lives, which may also explain their lesser benefit from marriage: “the woman, on the other hand, ensures generational continuity: she is never discharged from family obligations. Male autonomy in this respect involves the cost of a greater risk of loneliness” (Baudelot and Establet, 1984, p. 104). This hypothesis is part of a theory of gendered identities differentiated by male and female roles — son, daughter, husband, wife, father, mother — and socially constructed values (Dubar, 1987). As Besnard points out in his debate with Dubar (Besnard, 1987a, p. 378), this hypothesis amounts to considering that although women are dominated, they benefit from a marginal advantage in this domination, taking the form of a lower suicide rate. This hypothesis, which Besnard does not support, is expressed as an observation by Goldberg (1976). “The idea that the male sex is privileged… does not resist for a moment the examination of the statistics of risks run by individuals: if we consider factors as diverse as life expectancy, propensity towards illness, suicide and crime rate, number of accidents, alcoholism, drug addiction, etc., women are on average better off than men” (quoted in Giddens, 2004, pp. 185-186 [retranslated]). According to these authors, the “hazards of being a man” involve a high cost, particularly in suicide. The divergence between these two positions, Durkheim and Besnard on the one hand and Baudelot, Establet and Dubar on the other, is not as insurmountable as it may appear. In reality, their differences stem from the fact that each of these conceptions answers complementary but different questions, stressing either the differential benefits of marriage or women’s immunity to suicide. By adopting a gender problematic that unifies the matter, the two questions need to be addressed but separately. In the sociology of mental illness, as in that of suicide, the question of gender differences has often been approached from the angle of the differential benefits from marriage. The family protects its members and is at the same time the prime locus of inequalities between men and women. Does the male advantage in marriage revealed by Durkheim on the basis of the suicide rate and confirmed by Besnard (1997) persist when other expressions of emotional distress are examined, particularly as the institution of marriage weakens and other family forms develop? We propose therefore to re-examine the conclusions concerning greater female protection and male benefit from marriage using the concept of “mal-être” (ill-being) that can cover its various

4

According to Durkheim, matrimonial regulation corresponds to the moderation of the passions.

5

According to Durkheim, fatalistic suicide only concerns childless married women and those with “young husbands”. In his typology, fatalistic suicide is at the opposite pole from anomic suicide. It is the result of excessive regulation. It is the suicide of those whose future prospects appear to be permanently closed (Besnard, 1987b). 4

expressions, which taken separately would bias any conclusions based on one particular expression. The article is structured as follows. First, we discuss the concept of “ill-being”, because the value of this examination depends on its validity. Then we describe the statistical methods and sources employed and the indicators and types of score used. We present results showing that women are not more protected by their condition but that each gender adopts specific ways of expressing its “illbeing”. Finally, we re-examine the benefits of marriage and establish that women profit mainly from marriage and men from cohabitation. WHAT IS THIS THING CALLED “ILL-BEING”? Since Bentham, well-being has mostly been defined by economists as a utility function corresponding to the satisfaction of the agent’s desires and preferences. The abundant literature on the economics of well-being, happiness and hedonic psychology (Kahneman, Diener and Schwartz, 1999) has gone beyond this initial simplified model and established the relativity of well-being. First, it has been shown that the same level of self-reported happiness is observed for the inhabitants of wealthy and medium-income countries (Easterlin, 1974). Then as a nation advances, a “hedonic treadmill6” is gradually set up, whereby people get used to the new situation, raise their aspirations and are caught up in an endless race for satisfaction (Kahneman and Sugden, 2005). It is also noted that women report higher job satisfaction than better-placed men, because they assess their position not as compared with men, since they are as yet unaware of a right to professional equality, but compared to their own mother’s job (Clark, 1997). For our purposes, we note that well-being is appreciated relatively, with respect, for example, to a standard based on the respective position of agents, especially according to gender, and thus ultimately on the values integrated. Unlike well-being, which is often discussed, ill-being has not been explicitly conceptualised. In terms of attitudes to work, Baudelot and Gollac (2003) show that happiness is distributed along two dimensions: being (and doing) and having. For some people, it means having a family, a house, a job, money; for others, being content, at peace, on good terms with one’s children, husband, being loved, etc., but the authors observe that in order to be and to be fulfilled, you need first to possess. It is not therefore enough to reverse the definition of similar concepts like well-being and happiness to describe ill-being or unhappiness. And yet the concepts are obviously related. Schopenhauer believes that happiness is not to be sought in the satisfaction of desires and the accumulation of pleasures but is the result of an absence of suffering (1818, p. 404). Ill-being cannot be restricted to the non-satisfaction of preferences or the absence of the enjoyment of desired “goods”, since, although frustration can cause suffering, ill-being is not just suffering. Ill-being is an emotional suffering that is dual in nature. A subjective suffering, in that a given situation will affect to differing degrees individuals who are distinct but similar, and this difference is a matter of psychology. An objective suffering, in that it takes on identifiable forms of measurable intensity and regularity through many different expressions affecting distinct population groups. In this respect, illbeing cannot be reduced to its purely individual and psychological impact but displays an eminently social dimension. The wild boy of Aveyron had only a restricted vocabulary of elementary emotions and was insensitive to the death of a sibling (Le Breton, 1998). In this sense, ill-being can only find its source in society. It has a social dimension also because the historical construction of affectivity by the self-control of violence in particular had as its first consequence the construction and modification of taste, disgust and shame (Elias, 1939). This makes it possible to advance the reasonable hypothesis that the sources of ill-being have been modified by the social changes that have occurred throughout history. For example, the contemporary change in the status of the child as a recent object of affection in the couple (Ariès, 1960) will potentially generate greater suffering in the event of separation or death than in the past. The social dimension also appears when tensions arise between an individual marked with social attributes — gender, age, social and family status, etc. — and society. Tensions that can no longer be released in violence as in the past, as Norbert Elias tells us (1939): “In a certain sense, the battlefield has moved inside each man. That is where he must now cope with some of the 6

[Term coined by Brickman and Campbell (1971) and conceptualised by Eysenck (ca. 1990).] For a literature review in French of well-being and happiness economics, see Davoine, 2007. 5

tensions and passions that used to be discharged in hand-to-hand fight when men confronted each other directly” [retranslated]. The example of the excessive mental burden in some businesses shows that it can lead to depression and sometimes suicide (Ehrenberg, 1998). The transgression of internalised representations and socially constructed norms also generates suffering. The greater frequency of suicide attempts by women who have had homosexual relations (Lhomond and SaurelCubizolles, 2003) can be understood in the context of a society marked by compulsory heterosexuality (Butler, 2005). A shock of varying gravity may lie behind the feeling of inadequacy, personal failure (breakdown of a relationship, criminal conviction, bankruptcy, job loss, school failure, etc.), revealing the gap between a socially desirable position (success in relationships, career, education, society) and the supposed failure of the situation as experienced. Other tensions may cause ill-being and are created by the slow, gradual erosion of self-confidence (frustrations, criticisms from others, feelings of guilt, repeated unemployment, excessive mental stress, disability, illness, ageing, etc.) especially when combined. Suffering often arises from the perceived gap between internalised values and reality as experienced, leading to a feeling of inadequacy, a loss of self-esteem. The ill-being caused by the death of a loved one appears not to fit this model of inadequacy. But a death is often associated with feelings of guilt due to self-reproach at not have clarified the unspoken, or shown more attention or affection, which can now never happen. The origin and intensity of individual suffering are many and various, so it is not surprising that the manifestations of ill-being are equally so. Ill-being cannot be observed directly but is assessed through manifestations of widely varying form that are the result of the interaction between a particular pressure, a socially characterised individual and a given society. This is what Elias says about the civilising process (1939), “But they [self-constraints] also provoke — depending on internal pressures, the state of society and its members — specific tensions and disturbances in behaviour and emotional life” [retranslated]. Econometricians call these not directly observable phenomena latent variables. These variables can only be approximated indirectly via their visible expressions, which are merely the measurable indicators of a wider phenomenon. We would therefore qualify ill-being as a social fact, a phenomenon that may assume various forms of expression. For example, suicide, suicidal tendencies, alcohol dependence, depression, perceived loneliness, bulimia, anorexia and various nondegenerative mental illnesses, are only some of the various expressions that result from the social tensions that affect a socialised individual caught up in the contradictions between a value system that constrains that individual, their perception of their situation and the aggressive stimuli of varying kinds and intensity to which they are subjected. This all-embracing concept is rather like that of a social relationship or social hierarchy, which can only be approximated by educational qualifications, income, socio-professional category, residential area, etc. It cannot be reduced to a continuous linear variable from lowest to highest values, or for our purposes, from perceived loneliness up to suicide via intermediate stages that would be alcohol dependence and depression. It is rather a discrete variable that may involve some overlapping (suicide and depression, perceived loneliness and depression), or disconnection (anorexia and alcohol dependence) but always with marked degrees of intensity. The social construction of gender and the values attached to gender cause specific responses to the various events and situations that each sex must face. DATA, INDICATORS AND METHODS Data and indicators Three sources of French individual data were used: the 2003 database of causes of death for suicide, the 2005 Health Barometer of the French national institute for prevention and health education (INPES) for suicidal tendencies and the 2002-2003 INSEE health survey for depression and alcohol dependence. The suicide data, held by the epidemiological centre on medical causes of death (CépiDc), concern some 11,000 suicides among 500,000 deaths. These sources were not originally designed for statistical studies. They contain little socio-demographic information, which greatly restricts the opportunities for analysis. For that reason, we present suicide rates only by marital status and age for men and women, and then, following Durkheim, we calculate the increased coefficient of the unmarried compared with married people of each sex. 6

The INPES Health Barometer covers some 30,500 people and contains a large amount of information on the health of the general population and a section devoted to mental health. It provides in particular an evaluation of serious suicidal ideation (SSI). This is defined either as a positive answer to the question, “Over the last year, have you attempted suicide?” or by positive answers to the two questions, “Over the last twelve months, have you considered suicide?” and “At any time in your life have you attempted suicide?” since a first attempt greatly increases the risk of a repeated one. The data on depression and alcohol dependence come from the 2002-2003 health survey. This survey of French state of health and consumption of healthcare is carried out by INSEE every ten years. It offers major opportunities for studying the state of health according to the socio-demographic characteristics of households and individuals. Over 15,000 households, corresponding to 35,000 individuals, were interviewed. Unlike in previous health surveys, the questions are strictly individual for the adults able to reply. In addition, the 2002-2003 health survey includes questions that can be scored on internationally validated scales and used in epidemiological studies. In twenty questions covering most of the criteria for diagnosing depression, the CES-D (Center for Epidemiological Study of Depression) scale seeks to identify the presence of pre-depressive symptoms and assess their severity. Since the depression is not diagnosed by a doctor, it is generally agreed that what is captured by this scale is an indirect measure of depressive tendencies. However, for convenience, we use the two expressions indiscriminately in this paper. The questions refer to the previous week, and depressive tendencies are therefore measured at the time of the survey. Scores vary from 0 for the absence of pre-depressive symptoms to 60, corresponding to major depressive tendencies. It is generally recommended to take two cut-off points undifferentiated by gender: a cutoff at 17 is evidence of depressive symptoms and a cut-off of 23 demonstrates depressive symptoms (Husaini and Neff, 1980). We took the higher cut-off, on the understanding that the choice of either automatically changes the number of depressives identified but not our conclusions. Alcohol dependence is established by the DETA (Diminuer, Entourage, Trop, Alcool) questionnaire. This is adapted from the American CAGE clinical test (Cut down, Annoyed, Guilty, Eye-opener). It is designed to identify individuals presenting a risk of alcohol dependence by their answers to four questions: (1) Have you ever felt the need to diminish your consumption of alcoholic drinks? (2) Have family or friends ever made remarks to you about your consumption? (3) Have you ever had the impression that you drink too much? (4) Have you ever needed an alcoholic drink in the morning to feel in shape? Clinicians generally consider that the individuals who give two positive answers have an excessive consumption of alcohol likely to cause disease. Statistical models and endogeneity tests Unlike the suicide data, those on suicidal tendencies, depression and alcohol dependence can be used for multivariate logistic analysis. The variables introduced include age, sex, type of household, socioeconomic status (educational qualifications, employment and household income), state of health (invalidity or disability), and key events during childhood and the previous year that are known to be linked to depressive states (Menahem, 1992). Although this information is not perfectly harmonised between the two sources, it is sufficiently similar to justify comparison of the various indicators of illbeing. The logistic models established for each risk evaluate simultaneously the risk for men and women by correlating each independent variable with sex so as to reveal any contrasting effects. Using this method it is possible to rigorously evaluate the significance of differences in the male and female parameters. Statistically unobservable variables, such as physical appearance, may affect both the likelihood of living as a couple and the risk of greater ill-being. The greater ill-being of people not living as a couple might therefore partly account for the unobservable variables and biased coefficients of logistic models. To correct for the endogeneity bias and properly estimate the coefficients, we generally used models that simultaneously estimated two equations: the probability of living as a couple and the probability of ill-being. Since the correlations between the residuals from the two equations are not

7

significant, we conclude on the basis of our tests that there is no endogeneity bias7. The international literature shows that the benefits of marriage cover more the protection provided than marital selection. According to studies of longitudinal data, marital selection relating to the state of mental health remains limited. The study of a cohort of young adults reveals that depressive individuals of both sexes are equally likely to marry; on the other hand, the alcohol dependent are less likely to remain on the marriage market (Horwitz and White, 1991). Real or self-reported effects? Before examining the results, the suspicion relating to self-reporting effects in mental health questions needs to be addressed. It is often maintained that women more easily report that they are suffering from mental illness than men. If this were true, the discrepancies measured would be more selfreported than real. Comparison of the unprompted statements of illness and depressive episodes identified by the CES-D in the health survey 2002-2003 indicates higher self-reporting among women and shows that the use of a depressiveness scale reduces the discrepancy between men and women. However, the female prevalence of depression remains twice that of men (Leroux and Morin, 2006). Opinions are divided as to whether there is a gender-linked self-reporting bias. Using a specific interrogation protocol, researchers conclude on the contrary that there is a general tendency to underreport health problems, particularly of mental health, whatever the sex, apparently less out of fear of the implicit judgement of others than ignorance of the illnesses. Contrary to conventional wisdom, this under-reporting of mental illness is even slightly more pronounced among women (Macintyre, Ford and Hunt, 1999). RESULTS APPARENT CONTRADICTIONS IN THE EXPRESSION OF ILL-BEING In 2003, men continued to commit suicide more often than women; the suicide rate for women was one-third that for men, namely 9.2 per 100,000 population, compared with 27.5. The regularity of this higher male suicide rate at all ages, whatever their marital status, confirms, as if there were any doubt, that suicide is above all a male expression of ill-being (Table 1). Against this it may be argued that the discrepancy corresponds rather to the protection involved with caring for children, which is still overwhelmingly done by women. Durkheim shows, admittedly on the basis of fragile data, that the presence of children is a decisive protection against suicide (1897, p. 207-208) and infers, “that the essential factor in the immunity of married people is the family, namely the entire group formed by parents and children”. Halbwachs, using data from the Soviet Union, refines this idea according to the number of children: “In summary, these statistics reveal that married men and women, especially women, are more protected against suicide the more children they have” (Halbwachs, 1930, p. 178). His observations cannot unfortunately be confirmed at present in the absence of adequate data, a point to which we return below. However, one may note that female immunity to suicide is observable even among young unmarried women aged 15 to 24, most of whom have had no children, which would tend to prove that this greater female protection is not linked solely to the presence of children and has other causes too (Table1). Baudelot and Establet’s attractive hypothesis is based therefore on women’s better integration within the family, and to go further, they propose a single theory based on Durkheim’s concept of integration, “the protection against suicide that an individual enjoys is a function of the number and depth of the relationships they maintain with their family circle. This is a hypothesis in which we redefine integration, adding the following sub-hypothesis that one may consider sex and age to be factor of integration in the family.” (Baudelot and Establet, 1984, p. 101). Unfortunately this hypothesis cannot be tested either against suicide figures in the absence of information about the family relationships of the victims.

7

For all the bivariate probit models with instrumental variable tested, only the equations between living as a couple and female alcohol dependence showed a moderate correlation significant at the 10% threshold. 8

Table 1. Suicide rate by sex, marital status and age in 2003 Per 100,000 population Male Female Unmarried

Married

Widowed

Divorced

Age

Male Female Male Female Male Female Male Female

15 - 24

12.4

3.6

17.5

5.0

-

-

-

-

25 - 34

30.0

8.9

15.2

3.7

310.5

28.9

38.7

15.4

35 - 44

49.4

16.5

27.4

6.9

130.6

30.4

76.4

22.3

45 - 54

58.3

20.8

30.7

11.3

98.6

33.6

75.3

26.6

55 - 64

55.9

16.6

23.0

10.4

87.5

17.4

54.8

22.3

65 - 74

66.7

16.1

29.2

10.0

90.7

17.4

61.4

23.4

Source: Centre d’épidémiologie sur les causes médicales de décès (CépiDc, INSERM), deaths in 2003. However, if we suppose that suicide is one form of a wider phenomenon, ill-being, we may expect the same causes to have the same effects. This makes it difficult to explain the significance of the apparent contradiction between fewer female suicides and more frequent female suicide attempts and suicidal tendencies like depression (Figure 2). Consequently it is hard to maintain the Baudelot and Establet hypothesis as it stands, because one cannot suppose that the benefits of interactions with the family network would apply solely to the act terminating in death and that these protective qualities would be ineffective on suicide attempts and serious suicidal ideation and depression which often precede suicide itself (Davidson and Philippe, 1986; Lemperière, 1999). In other words, interactions with the extended family, whether of support or constraint, cannot at once preserve against the most radical form of ill-being and yet be ineffective against, or even aggravate, that ill-being in its less lethal forms. If more numerous interactions with the extended family protect against one form of ill-being (suicide), why do they not protect against other forms? The fact remains that the family interaction hypothesis is inadequate to solve the paradox of women’s lower suicide rate and higher depression rate. Most of the discrepancies observed for each form of expression of ill-being would appear to be based not on variations in integration but primarily on the separate ways each sex finds to express its ill-being. However, researchers have convincingly shown that when relationships are supportive, involving confidence and support from parents, friends and associates, they moderate psychological distress (Umberson et al., 1996). They point out that without their network of relations that is denser than men’s, women would present even higher levels of depression. Conversely, they observe that relationships of constraint, such as regularly caring for older parents, aggravate depression levels. We are therefore inclined to consider that positive interactions with one’s close circle have a moderating effect on ill-being, but this effect is insufficient to counteract the discrepancies in depression and suicide among men and women. The suicide curves by age and gender bring us back yet again to differences of gender (Figure 1), and after the relative improvement that coincides with the early years of retirement (Delbès and Gaymu, 2004), the male suicide rate climbs steeply while the female rate stabilises. These facts should be seen as follows: at this period of life, men are faced with problems relating to the passage of time that incite them to suicide, while women faced with the same ageing are not excessively penalised, at least with respect to suicide. This phenomenon is difficult to understand if one analyses this single indicator alone. Are men’s declining mental and physical capacities and the loss of their power status and professional functions hard to reconcile with male representations of manhood, already weakened by a less wealthy social status due to retirement and a network of relations that no longer includes their work colleagues? Why then do women, more socially dependent on their physical appearance and 9

losing their attractiveness, not commit suicide more often at the same age? Might it be that, since women have a more distant relationship with work, they find retirement less difficult to cope with? Figure 1. Male and female suicide rates by age 80 70

Suicide rates

60 50 Men

40

Women

30 20 10 0 0 - 14

15 - 24 25 - 34 35 - 44 45 - 54 55 - 64 65 - 74

75+

Source: Centre d’épidémiologie sur les causes médicales de décès (CépiDc, INSERM), deaths in 2003. Graph by authors. Figure 2. Serious suicidal tendencies, depression and alcohol dependence by age. % 25

20

15

10

5

0 18-24

25-34

35-44

Suicidal tendencies Men

45-54

55-64

65-74

75 and more

Suicidal tendencies Women

Depression Men

Depression Women

Alcohol Men

Alcohol Women

Sources: Enquête Santé 2002-2003, INSEE and Baromètre Santé 2005, INPES Population of reference: individuals aged 18 and over. Graph by authors. In fact, female depression does rise after the age of 45, stabilises between 55 and 74, and then rises even faster, which implies that here too other phenomena are interacting during these periods of maturity and ageing, particularly retiring from employment, which may be experienced differently according to gender and the job held, and then widowhood and the appearance of disabilities that gradually restrict daily life (Figure 2). In addition, the striking parallel between the curves for female 10

serious suicidal tendencies and alcohol dependence would appear to indicate a similar expression in these two indicators, peaking between 45 and 54 and then falling. It will be recalled that female suicides peak in the 45-54 age group and then level out. There is therefore at this point in women’s lives an event, or more likely a series of events, that contribute to unsettling them. This age often corresponds to the end of a period, marked by children leaving home, the menopause and an awareness of lesser attractiveness. Figure 3. Serious suicidal tendencies, depression and alcohol dependence by type of household %

5

Serious suicidal tendencies

0

35 30 25

Depression

20 15 10 5 0

20

Men

Women 15

10

Alcohol dependence

5

Sources: Enquête Santé 2002-2003, INSEE and Baromètre Santé 2005, INPES Population of reference: individuals aged 18 and over. Graph by authors. 11

Other

more

children)

(one or

Married

(no

children)

Married

more

children)

(one or

Cohabiting

(no

children)

Cohabiting

family

parent

Single-

(living

alone)

Widowed

(living

alone)

Divorced

(living

alone)

Unmarried

0

Among men, alcohol dependence peaks between the ages of 35 and 64 and then declines permanently. This may indicate various phenomena combining among young men an alcoholism based on parties, sociability and peer pressure, typically linked to the male role, and later an ill-being expressed in alcohol abuse among older men. The lower rate of alcohol dependence among the oldest men may be explained by lower physical tolerance of alcohol, advice not to combine alcohol with drug treatment and medical pressure to reduce a long-standing habit (Canarelli, Cadet-Tairou and Palle, 2006). In practice, the undesirable effects of excessive alcohol consumption are much harder to tolerate past a certain age and the sharp fall in the curve would therefore not be due to an improvement in individuals’ well-being, which is contradicted by the other indicators, but a special aspect of the side effects of excessive alcohol consumption. The other two indicators are harder to explain by age alone. Male depression remains more or less stable at all ages. Male suicidal tendencies are extremely low and fall virtually to nil at the highest ages, which confirms the absence of a link with suicide. The curves of expressions of ill-being have greatly varying profiles. However, at whatever age, women do more frequently present serious suicidal tendencies and a higher degree of depression, while men are more alcohol dependent and commit suicide more often. These results do not conceal questions of family structure. They remain the same for various types of household. Whatever their living arrangements, women always have a much greater prevalence of suicidal and depressive tendencies, and men are more subject to alcoholic dependence; the only exception is depression among the widowed living alone, where the usual male-female discrepancy is not observed (Figure 3). The results also stand up to multivariate analysis. Controlling for other characteristics, women are 1.8 times more likely to present serious suicidal tendencies, 2.4 times more likely to be depressive and 4.8 times less likely to be alcohol-dependent than men (Table 3). This set of results supports the hypothesis of an expression of ill-being based on the social construction of gender and particularly the inculcation during childhood of the values that will define gender (Belotti, 1974) and will later structure the individual’s mental and physical behaviour. Men are involved in virtually all sexual violence, in 84% of physical assault and 93% of attempted murders (Jaspard et al., 2001). Men were also 85% of those convicted of various offences and theft from 1950 to 1992 (Robert et al., 1994). Similarly, in the United States, 94% of prison inmates and 90% of murderers are men. Social construction and the internalisation of these constructions apparently lead men to externalise tensions by alcohol and drug abuse, defiant and deviant behaviour, physical violence and suicide, while women internalise problems, use verbal violence, and suffer somatisation disorders and depression (Braconnier, 1996, p. 96). WHO STANDS TO GAIN FROM MARRIAGE? Since Gove’s research, the question of differential gains from marriage has determined the field of sociological studies of gender differences in mental health. Compared with unmarried people of either sex, married men suffer less mental illness than married women. Conversely, unmarried women appear to be less prone to psychological disturbance than unmarried men. Marriage would seem therefore to have a protective effect on men’s mental health but to be a burden for women (Gove, 1972). This difference is claimed to be due to the traditional male and female roles within the conjugal society. The generally undervalued domestic role always falls to women and, combined with low job satisfaction among women who go out to work, is apparently the origin of greater female frustration (Gove and Tudor, 1973). In addition to its questionable choice of neuroses, more frequent among women, as the sole approximation to mental health (Dohrenwend and Dohrenwend, 1976), Gove and Tudor’s theory is based on an analysis of social relations between the sexes that dates back to the early 1970s, before the development of female paid employment and the current weakening of the institution of marriage. Until recently, the assertion that marriage favours men but disadvantages women was rarely challenged, most probably because it chimes with our sociological understanding of the inequalities between men and women (Williams, 2003). The family has seen radical upheavals since the late 1960s: development of cohabitation, later date of first marriage and first child, more births outside wedlock, higher divorce rate following the introduction of divorce by mutual consent, etc. (Roussel, 1989). These changes to marriage are bound to have consequences for the division of benefits for each member of the couple. Durkheim was one of 12

the first to point out that “marriage favours the woman in terms of suicide when divorce is more common” (Durkheim, 1897, p. 302). In addition, the development of paid employment for women has helped to redefine the power relationship between husband and wife. Marriage no longer imposes the same constraints, particularly for women, since it is now only one form of union, admittedly the most frequent, alongside others (civil partnership, cohabitation) and when it is contracted, it is with the knowledge that it can be dissolved. As a result of the weakening of the institution of marriage, the protection marriage provides against suicide is affected although it has not entirely disappeared (Besnard, 1997; Surault, 1995). Recent studies of depression and excessive alcohol consumption, for instance, show that marriage has a positive effect on the mental health of both husband and wife (Ross, 1995; Horwitz, White and Howell-White, 1996; Simon, 2002; Williams, 2003). This result is one reason for re-examining the hypothesis of a mainly male gain from conjugal life. Taking inspiration from Simon’s hypotheses (2002) in her article on the relationship between gender, marital status and mental health, our study of gender differences in ill-being seeks mainly to respond to the following queries. If the various expressions of ill-being are basically gendered, then women should present more suicidal tendencies and depression, and men should have a higher suicide rate and more frequent alcohol problems, and these patterns should be observable whatever their age or marital status. Given the beneficial effect of living together, married or cohabiting individuals should be further removed from ill-being, by whatever indicator. And, if men gain more than women from marriage, the difference between the married and the unmarried should be more marked among men. LOOKING FOR THE WINNERS FROM LIVING TOGETHER The study of suicide gives an initial indication of the differential gains in marriage, even if it so happens that the use of legal status does not make it possible to distinguish clearly between what is due to marriage and what more widely to conjugal family life. As Durkheim noted in the 19th century, except for early marriages, the married commit suicide less often than the unmarried (Table 2). On the other hand, the widowed and divorced do not seem to gain from their status as formerly married, since their “coefficient of aggravation” for suicide is generally higher than for the unmarried. Indeed it is widowed men who commit suicide most often at present (Besnard, 1997). Despite the weakening of the institution of marriage, it does still protect against suicide. But does this beneficial effect still only work for men? In Suicide (1897), Durkheim concludes that the main benefit from marriage is for men. A century later, this assertion needs to be nuanced. First, it is only after the age of 55 that unmarried men present a higher coefficient of aggravation than unmarried women (Table 2). Compared with married people, being unmarried appears to be even more disadvantageous for women than for men between the ages of 25 and 44, the main childbearing age. In the eyes of society, and therefore of women themselves, female self-realisation still takes the form of bearing a child. Even though the correlation between being unmarried and living alone is less clear as forms of cohabitation have changed. What is likely to be revealed here is the social pressure on women who have not yet lived out their social destiny as mothers. Once the childbearing age is past, unmarried status appears to be less hard for women to live with. Whereas widowers do indeed have a higher coefficient of aggravation than widows, the male advantage in marriage has disappeared among the divorced. Despite what Besnard (1997) observed between 1981 and 1993, divorce now increases women’s tendencies to suicide as much as if not more than men’s, compared with the married. Women’s greater family integration, since they are overwhelmingly the partners to be awarded custody of children, does not appear now to be sufficient to compensate for the negative effects of divorce. As Durkheim predicted, the introduction of divorce has increased married women’s protection against suicide. But it has also, by the same token, increased the fragility of the category of divorced people it has created. How far do the other forms of ill-being confirm these preliminary results based on suicide? As we now see, the weakening of the institution of marriage makes it necessary to reconsider the assertion of male gains from marriage.

13

Table 2. Coefficients of aggravation for suicide compared with married people of each sex. Unmarried

Widowed

Divorced

Age

Men

Women

Men

Women

Men

Women

15 - 24

0.7

0.7

-

-

-

-

25 - 34

2.0

2.4

20.4

7.8

2.6

4.2

35 - 44

1.8

2.4

4.8

4.4

2.8

3.2

45 - 54

1.9

1.8

3.2

3.0

2.5

2.4

55 - 64

2.4

1.6

3.8

1.7

2.4

2.1

65 - 74

2.3

1.6

3.1

1.7

2.1

2.3

Source: Centre d’épidémiologie sur les causes médicales de décès (CépiDc, INSERM), deaths in 2003. Interpretation: Unmarried men aged 25-34 are twice as likely to commit suicide than married men of the same age. MALE GAINS FROM LIVING TOGETHER Controlling for other characteristics, individuals living alone, whether unmarried, widowed or divorced, are particularly prone to ill-being, especially in the form of suicide or depression (Table 3). There is therefore an observable gain for both men and women who are living together. But can one strictly speak of a “benefit from marriage”? Is it really the fact of being married that matters, or the fact of living in a couple with one or more children rather than alone? As we noted with respect to suicide, the possible benefits of marriage are lost once it breaks down, since the formerly married do not seem to do any better in respect of ill-being than the never married. Strictly speaking, the actual effect of marriage as a particular form of union can only be measured against other individuals living as couples. In practice, the gain is mainly due to living as a couple. Sharing accommodation with other people than one’s spouse and children if any (relatives, friends, flatmates) increases the likelihood of presenting serious suicidal tendencies in both sexes and episodes of depression in men. The observed gain corresponds therefore not to the mere fact of not living alone but to living with the particular partner who is a spouse (Gove, Hughes and Briggs Style, 1983). The presence of children has no effect, positive or negative, on the ill-being of the married. The gain depends less on the presence of children than on that of a spouse. Although both sexes gain from living together, men gain more. Unmarried men living alone and the heads of single-parent families present a higher likelihood of suicidal and depressive tendencies than their female equivalents (Table 3). Similarly, widowers living alone are significantly more prone to the various expressions of ill-being than widows. These results are in line with the observation of a greater gain for men from conjugal life. Conversely, the higher rate of alcohol dependence among unmarried women living alone may indicate a possible female gain from living together. However, the particular nature of this indicator must be pointed out. Whereas socio-economic status and type of household have little effect on alcohol dependence among men, alcohol dependence is mainly seen among women living alone, cohabiting or living as a couple after a relationship has broken down (Table 4), highly educated, with substantial income (Table 3), and therefore most likely women who are less subject to social control and diverge from the social behaviour expected of their gender. The endogeneity tests described above confirm the correlation between living as a couple and women’s alcohol dependence. The last result should therefore be interpreted with caution, because it may come from a selection of people living alone (Horwitz and White, 1991) or cohabiting (Horwitz and White, 1998).

14

Table 3. Probability of presenting serious suicidal tendencies, depression or alcohol dependence. Logistic models 1 Simultaneous for men and women Serious suicidal tendencies -6.09 ***

Constant Sex Male Female

Ref 0.57 Men

Women

Depression

Alcohol dependence

-3.95 ***

-2.32 ***

Ref 0.88 *** Men Women

Ref -1.56 *** Men Women

Age 18 – 24 25 – 34 35 – 44 45 – 54 55 – 64 65 – 74 Type of household Unmarried, living alone, no children Divorced/separated, living alone, no children Widowed, living alone, no children Single-parent family Cohabiting, no children Cohabiting, one or more children Married, no children Married, one or more children Other1 Qualifications None or no response CEP, BEPC only CAP, BEP Baccalauréat Higher Household income per consumer unit Bottom quintile 2nd quintile 3rd quintile 4th quintile Top quintile Employment status Working Unemployed Inactive Invalidity Adverse events during childhood Death, serious illness, disability or accident, father or mother Separation or serious disputes between parents Financial difficulties2 Adverse events during previous year Death of close relative Particular financial difficulties3 Job or school difficulties4 Percentage concordance Number of observations

-1.10 ** -0.45 Ref -0.25 -0.83 ** -1.57 ** 1.44 *** 1.99 *** 1.50 1.75 1.17 0.26 -0.37 Ref 1.76

* *** **

***

-0.34 -0.05 -0.05 Ref -0.87 * -0.14 0.23 Ref -0.02 0.25

-1.18 *** -0.18 Ref 0.26 -0.64 *** -1.45 ***

-0.69 ** -0.17 Ref -0.02 -0.25 -0.77 ***

0.68 *** 1.54 ***

1.09 *** 1.19 ***

0.71 *** 0.87 ***

1.88 *** 1.12 *** 0.07 -0.02 0.05 Ref 0.55 *

0.70 *** 0.66 *** 0.07 0.42 *** -0.11 Ref -0.18

0.54 * -0.01 0.12 0.18 -0.09 Ref -0.07

-0,30 0,23 0,59 *** 0,47 ** -0,05 Ref -0,24

0.29 0.27 0.11 Ref -0.20

0.33 0.32 0.23 Ref -0.22

-0.06 -0.08 0.07 Ref 0.09

-0,94 -0,33 -0,34 Ref 0,33

0.22 -0.02 Ref 0.15 -0.21

0.25 ** -0.03 Ref -0.17 0.00

-0.04 -0.06 Ref 0.03 0.19 *

0,17 -0,11 Ref 0,25 0,47 ***

0.97 0.72 1.04 -0.10 0.13 Ref 0.84

*** *** ***

**

0.30 0.43 * 0.30 Ref 0.17 0.02 -0.01 Ref 0.22 -0.32

-0.09 -0.11 Ref 0.12 0.02 -0.01

-0.19 -0.29 ** Ref 0.00 -0.01 -0.42 **

*** *** ** *

0.06 0.27

-0,26 -0,39 ** Ref 0,51 *** 0,01 -0,13 0,44 ** 0,44 *

*** * * **

Ref 0.44 0.09 1.69 ***

Ref 0.00 0.37 ** 0.81 ***

Ref 0.36 ** 0.38 ** 1.32 ***

Ref 0.50 *** 0.24 *** 1.11 ***

Ref 0.43 ** 0.06 0.35 ***

Ref -0,11 0,06 -0,18

0.45 *

0.64 ***

0.10

0.20 **

0.21 **

0,21

1.18 ***

0.83 ***

0.47 ***

0.27 ***

0.34 ***

0,49 ***

0.27

0.41 ***

0.66 ***

0.46 ***

0.26 **

0,12

/ / 2.02 *** 1.33 *** 1.55 *** 1.13 *** 82% 379 of 25,857 15

0.15 0.30 *** 1.25 *** 1.02 *** 0.93 *** 0.52 *** 78% 1,714 of 17,815

0.22 *** 0,22 * 0.44 *** 0,58 *** 0.15 0,67 *** 71% 1,350 of 17,815

Sources: Enquête Santé 2002-2003, INSEE and Baromètre Santé 2005, INPES Authors’ calculations. Population of reference: household reference persons and spouses ***, **, * significant at 1% 5%, 10% level Difference of male/female coefficients significant at 5% level 15% level 1 Households comprising people (related or not) other than spouse and any children 2 Money problems during youth for serious suicidal tendencies 3 Dispute with friends or money problem for serious suicidal tendencies 4 Difficulties, poor results for serious suicidal tendencies

Female gains from marriage Comparing the married and the cohabiting, we can observe more accurately the actual effects of the type of union, whether marital or not. Except for greater suicidal tendencies among cohabiting men with no children, it is married and cohabiting men with one or more children or none who have the lowest levels of ill-being in its various expressions. Overall there is no significant difference according to the status of the relationship between the spouses. This confirms the beneficial effect of a couple for men, whatever its form. Among women, even the lower gain from living as a couple is increased by a gain from marriage, perceptible in the rates of serious suicidal tendencies, depression and alcohol dependence. Marriage now is no longer an excessive constraint for women, since married women present the lowest level of ill-being. Unlike Durkheim’s observation a century earlier, the gain from marriage itself is now more female than male. Although the institutionalisation of divorce favours married women, as Durkheim predicted, divorce itself has created new family risks (de Singly, 1987). Marriage has become a legal protection for women against the possibility of separation. The family is one of the major places for inequality between men and women, since the birth of children usually involves the woman withdrawing partly or completely from the labour market and reducing her autonomy and relations, and greater social vulnerability in the event of separation. This is why the cost of marriage for women, largely their limited career prospects, appears to be particularly high in the event of divorce. Divorced women, especially if they have no children, present the highest rates of suicidal tendencies, depression and alcohol dependence, showing that they suffer badly from marriage breakdown. The disadvantage of unmarried men and divorced women is most probably due to the varying gains that each sex gets mainly from marriage. Although marriage provides social support and increases material well-being, the factor of family integration is more important for men, while women are more sensitive to the economic factor. In the event of marriage breakdown, women suffer more from a fall in living standards, and men more from loneliness (Gerstel, Riessman and Rosenfield, 1985; Umberson, Wortman and Kessler, 1992). But it is cohabitation, a less traditional relationship than marriage, that appears to be a form of union less favourable to women. The hypothesis of selection effects as between cohabiting and married couples cannot, however, be totally excluded. Religious belief, condemning suicide, may also affect individuals’ choice between marriage and cohabitation. However, we have seen that religion has no significant effect on serious suicidal tendencies. Similarly, a study of a cohort of initially unmarried young adults shows that cohabitation is indeed associated with excessive alcohol consumption (Horwitz and White, 1998). On the other hand, it is not linked with depression (Horwitz and White, 1998; Brown, 2000). Consequently the ill effects of cohabitation are not simply due to the selection of individuals more prone to ill-being. The situation of cohabiting women may be explained by the fact that they take on the burdens of family life without benefiting from the security and guarantees that come from marriage. Indeed, the greater rate of depression among the cohabiting compared with the married is mainly due to the feeling of instability in the relationship (Brown, 2000). Compared with married women, cohabiting women have a greater risk of alcohol dependence whether or not they have children, of suicidal tendencies if they have none and of episodes of depression if they have some. Among cohabiting women, children do not in general provide protection against ill-being but rather determine its expression. Although children are related to lower suicidal tendencies, the illbeing of cohabiting women does not disappear but takes another path. 16

*** The gap between men and women’s suicide rates comes therefore apparently not from some female immunity, however this might be explained. Because of differential socialisation during childhood and the places assigned and roles attributed to the two sexes, the gap is due more to the fact that each gender has its own way of responding to life’s various tensions. Depending on the data available, we could have added expressions as varied as drug abuse, violent behaviour, suicide attempts, bulimia, anorexia, etc., which are also indicators of tensions between a gendered individual and society. The social construction of gender shows how far the internalisation of inculcated values determines even our most personal reactions and those over which we have little direct control. Both men and women are dependent on the social roles assigned them. To take suicide as the only indicator of individuals’ ill-being or as a sign of “social happiness” or the “state of health” of the social system (Durkheim, 1897) leads therefore to a partial vision and conclusions that may be erroneous, particularly when comparing men and women. So the idea must be abandoned of opting for a single relevant expression of ill-being that would tend to imply the strange conclusion that women gain a marginal benefit from being socially dominated. Ultimately, it is apparently less the integration due to “family society” that provides protection against ill-being, but rather the integration due to the couple, which contradicts Durkheim’s hypothesis on suicide. Similarly, the quality of conjugal relations in fact takes on much more importance than the fact of living as a couple. Remaining married when the relationship is perceived to be unsatisfactory is worse for mental health than living permanently alone or being separated (Gove, Hughes and Briggs Style, 1983; Ross, 1995; Williams, 2003). Even when controlling for the gendered nature of the expressions of ill-being, if there is any benefit to marriage compared with cohabitation, it is rather for the woman than the man. In the space of a century, the nature of marriage has certainly changed greatly and we are comparing things that are quite different. At present, living together no longer needs to be marital to be considered as a legitimate framework for a couple and a family. Similarly, the introduction of divorce by mutual consent has greatly affected the content of marriage. Paradoxically, as women have become more autonomous and benefited from the advantage of marriage, they have also become the main victims, as expressed by ill-being, of the new fragility of the couple relationship. Even if women go out to work, their place is still often restricted to providing an extra salary and handling the practical management of the household. For women, therefore, a breakdown in the couple greatly increases their social vulnerability. Clearly, the introduction of divorce and the rapid rise in the number of separations are due to deeper changes in the relations between and expectations of the two people. But these observations should not involve idealising the married life of the past. One need only note that the pressures on women used to be expressed in frequent neuroses, complaints that gradually reduced during the 20th century, as depression increased among women, and, to a lesser extent, men (Ehrenberg, 1998).

Aneshensel S, Rutter C.M, Lachenbruch P.A, 1991. - “Social Structure, Stress, and Mental Health: Competing Conceptual and Analytic Models”, American Sociological Review, vol. 56, n°2, pp. 166-178. Ariès P., 1960. – L’enfant et la vie familiale sous l’Ancien Régime, Paris, Plon. Badeyan G., Parayre C., Mouquet M.-C., Tellier S., Dragos S. et Ellenberg E., 2001. - « Suicides et tentatives de suicide en France. Une tentative de cadrage statistique », Etudes et résultats, n° 109. Baudelot Ch. et Establet R., 2006. - Suicide : L’envers de notre monde, Seuil. — 1984. - Durkheim et le suicide, Coll. Philosophie, 6e édition, 2002, PUF. Baudelot C., Gollac M. avec Bessières C., Coutant I., Godechot O., Serre D. et Viguier F., 2003. - Travailler pour être heureux ? Le bonheur et le travail en France, Fayard, Paris. Belotti E. G., 1974. - Du côté des petites filles, éditeur Des Femmes. 17

Besnard P., 1997. - « Mariage et suicide : la théorie durkheimienne de la régulation conjugale à l’épreuve d’un siècle », Revue française de sociologie, XXXVIII, n°4, pp. 735-758. — 1987a. - « Les sociologistes et le sexe. Réponse à Claude Dubar », Revue française de sociologie, vol. XXVIII, n°1, pp. 137-144. — 1987b. - L’anomie : ses usages et ses fonctions dans la discipline sociologique depuis Durkheim, Paris, PUF. — 1973. - « Durkheim et les femmes ou le Suicide inachevé », Revue française de sociologie, vol. XIV, n°1, pp. 27-61. Bourdieu P., 1998. - La domination masculine, Coll. Liber, Editions du Seuil. Brown S.L., 2000. - “The Effect of Union Type on Psychological Well-being: Depression Among Cohabitators Versus Marrieds”, Journal of Health and Social Behavior, vol. 41, n°3, pp. 241-255. Butler J., 2005. - Trouble dans le genre. Pour un féminisme de la subversion, Paris, Editions La Découverte. Canarelli T., Cadet-Tairou A., Palle C., 2006. - « Indicateurs de la morbidité et de la mortalité liées à l’alcool en France », Bulletin épidémiologique hebdomadaire, n° 34-35. Chesnais J. -C., 1976. - « Les morts violentes en France depuis 1826 », Travaux et documents, Cahier n° 75, Ined, PUF. Clark A. E., 1997. - « Job Satisfaction and Gender: Why are Women so Happy at Work ? », Labour economics, vol. 4, n° 4, pp. 341-372. Clément S., Membrado M., 2001. - « Des alcooliques pas comme les autres ? La construction d’une catégorie sexuée », in : Aïach P., Cèbe D., Cresson G., Philippe C., Femmes et hommes dans le champ de la santé. Approches sociologiques, Rennes, Editions ENSP. Davidson F., 1986. - « Conclusions » in : Suicide et tentatives de suicide aujourd’hui. Etude épidémiologique, Collection Grandes enquêtes en santé publique et épidémiologie, Les éditions Inserm. Davidson F. & Philippe A., 1986. - « Les tentatives de suicide » in Suicide et tentatives de suicide aujourd’hui. Etude épidémiologique, Collection Grandes enquêtes en santé publique et épidémiologie, Les éditions Inserm, pp.33-68. Davoine L., 2007. – L’économie du bonheur peut-elle renouveler l’économie du bien-être ?, Document de travail, n° 80, Centre d’études de l’emploi, 27 p. Delbès C. & Gaymu J., 2004. - La retraite quinze ans après, Les cahiers de l'INED, 223 p. Dohrenwend B.P, Dohrenwend B.S, 1976. - “Sex differences in Psychiatric Disorders”, American Journal of Sociology, vol. 81, n°6, pp. 1447-1454. Dubar C., 1987. – « A propos de l’interprétation du Suicide de Durkheim par Philippe Besnard », Revue française de sociologie, vol. XXVIII, n°1, pp. 127-136. Durkheim É., 1897. - Le suicide, 9e édition, 1997, Paris, PUF, Quadrige. Ehrenberg A., 1998. - La Fatigue d’être soi – Dépression et société, Odile Jacob, Paris. Elias, 1939, La Civilisation des mœurs, trad. fr. 1973, rééd. Calmann-Lévy, coll. « Liberté de l'esprit », 1991. Elias, 1939, La dynamique de l’Occident, réed. 1990, Paris, Calmann-Lévy, coll. Agora, Pocket, 320 p. Gerstel N, Riessman C.K, Rosenfield S, 1985. - “Explaining the Symptomatology of Separated and Divorced Women and Men: The Role of Material Conditions and Social Networks”, Social Forces, 1985, vol. 64, n°1, pp. 84-101. Giddens A., 2004. – La transformation de l’intimité, Hachette Littératures. Goldberg H., 1976, The Hazards of Being Male: Surviving the Myth of Masculine Privilege, New York, New American Library, 195 p. Gove W.R, Hughes M., Briggs Style C., 1983. - “Does Marriage Have Positive Effects on the Psychological Well-being of the Individual?”, Journal of Health and Social Behavior, vol. 24, n° 1, pp. 122-131. Gove W.R, 1972. - “The Relationship between Sex Roles, Marital Status and Mental Illness”, Social Forces, vol. 51, pp. 34-44.

18

Gove W.R, Tudor J.F, 1973. - “Adult Sex Roles and Mental Illness”, American Journal of Sociology, vol. 78, n°4, pp. 50-73. Halbwachs M., 1930. - Les causes du suicide, rééd. 2002, Paris, PUF, Coll. Le lien social. Horwitz A.V, Davies L, 1994. - “Are Emotional Distress and Alcohol Problems Differential Outcomes of Stress? An Explanatory Test”, Social science quarterly, vol. 75, n° 3, pp. 607-621. Horwitz A.V, White H.R, 1998. - “The Relationship of Cohabitation and Mental Health: A Study of Young Adult Cohort”, Journal of Marriage and the Family, vol. 60, n°2, pp. 505-514. — 1991. - “Becoming Married, Depression and Alcohol Problems Among Young Adults”, Journal of Health and Social Behavior, vol. 32, n°3, pp. 221-237. Horwitz A.V, White H.R, Howell-White S, 1996. - “The Use of Multiple Outcomes in Stress Research: A case Study of Gender Differences in Responses to Marital Dissolution”, Journal of Health and Social Behavior, vol. 37, n°3, pp. 837-857. Hughes M.E, Waite L.J, 2002. - “Health in Household Context: Living Arrangements and Health in the Late Middle Age”, Journal of Health and Social Behavior, vol. 43, n°1, p.1-21. Husaini B, Neff J.A, 1980. - “Depression in Rural Communities. Validating the CES-D Scale”, Journal of Community Psychology, vol. 8, pp. 20-27. Jaspard M. et l'équipe ENVEFF, 2001. - « Violences envers les femmes : une reconnaissance tardive » in L'état de la France 2001-2002, La Découverte, pp.76-79. Kahneman D., Diener E. & Schwartz N. 1999. – Well-being: The Foundations of Hedonic Psychology. The Russel Sage Foundation. Kahneman & Sugden, 2005, « Experienced Utility as a Standard of Policy Evaluation », Environnemental & resource Economics, vol. 32, n°1, pp.161-181. Le Breton, D., 1998. – Les passions ordinaires. Anthropologie des émotions, Payot, Coll. Petite bibliothèque Payot, 347 p. Lemperière, T., 1999. - Dépression et suicide, Thérèse Lemperière dir., Programme de recherche et d’information sur la dépression, Masson, Coll. Acanthe. Leroux I., Morin T., 2006. - « Facteurs de risque des épisodes dépressifs en population générale », Etudes et résultats, n° 545. Lhomond B., Saurel-Cubizolles M.J., 2003. – « Orientation sexuelle, violences envers les femmes et santé. Résultats de l’enquête sur les violences envers les femmes en France », in : Homosexualités au temps du SIDA, dir. Broca et al., ANRS, pp. 107-130. Lovell A. & Fuhrer R., 1996. - « Trouble de la santé mentale. La plus grande « fragilité » des femmes remise en cause », in La santé des femmes, Marie-Josèphe Sauvel-Cubizolles, Béatrice Blondel dir., Coll. Médecine-sciences, Flammarion.", pp. 252-283. Macintyre S., Ford G., Hunt K., 1999. - « Do women “over-report” morbidity? Men’s and women’s responses to structured prompting on a standard question on long standing illness. », Social Science & Medicine, 48(1), pp. 89-98. Mead M., 1963, Mœurs et sexualité en Océanie, Paris, Plon. Menahem G., 1992. - « Troubles de santé à l’âge adulte et difficultés familiales durant l’enfance », Population, n°4, pp. 893-932. Mouquet M.-C., Bellamy V., Carasco V., 2006. - « Suicides et tentatives de suicide en France », Etudes et résultats, n° 488. Riessman C.K, Gerstel N, 1985. - “Marital Dissolution and Health: Do Males and Females Have Greater Risk?”, Social Science and Medicine, vol. 20, n°6, pp. 627-635. Robert P., Aubusson de Cavarlay B., Pottier M.-L., Tournier P., 1994. – Les comptes du crime. Les délinquances en France et leurs mesures, Éditions L'Harmattan., p. 61. Ross C.E., 1995. - “Reconceptualizing Marital Status as a Continuum of Social Attachment”, Journal of Marriage and the Family, vol. 57, n°1, pp. 129-140. 19

Ross C.E., Mirowsky J., Goldsteen K., 1990. - “The impact of the Family on Health: The Decade in Review”, Journal of Marriage and the Family, vol. 52, n°4, pp. 1059-1078. Roussel L., 1989. - La famille incertaine, Éditions Odile Jacob. Schopenhauer A., 2001. - L'art d'être heureux, Paris, Seuil. Simon R.W, 2002. - “Revisiting the Relationship among Gender, Marital Status, and Mental Health”, American Journal of Sociology, vol. 107, n°4, pp. 1065-1096. — 1998. - “Assessing Sex Differences in Vulnerability among Employed Parents: The Importance of Marital Status”, Journal of Health and Social Behavior, vol. 39, n°1, pp. 38-54. Singly de F., 1987. - Fortune et infortune de la femme mariée, (3e édition revue et remaniée, 1994); collection Quadrige, PUF, Paris, 2003. Steiner P, 1994. - La sociologie de Durkheim, éd. 2005, 4e édition, Paris, La Découverte, n°154. Surault P., 1995. - Variations sur les variations du suicide en France, Population, n°4-5, pp. 983-1012. Tocqueville A., 1840. - De la Démocratie en Amérique, tome 2, éd. 1951, Editions M.-Th. Génin, Libraire de Médicis. Umberson D., Chen M.D., House J.S., Hopkins K., Slaten E., 1996. - “The Effect of Social Relationships on Psychological Well-being: Are Men and Women Really So Different?”, American Sociological Review, vol. 61, n°5, pp. 837-857. Umberson D., Wortman C.B, Kessler R.C, 1992. - “Widowhood and depression: explaining long-term gender differences in vulnerability”, Journal of Health and Social Behavior, vol. 33, n°1, pp.10-24. Williams K., 2003. - “Has the Future of Marriage Arrived? A Contemporary Examination of Gender, Marriage, and Psychological Well-Being”, Journal of Health and Social Behavior, vol. 44, n°4, pp. 470-487.

20