CAIRN-INT.INFO : International Edition

1The idea that society is becoming increasingly pathogenic and stressful is not a new one. As Tocqueville noted in 1840, “The soul is more stricken and perturbed” amongst the peoples of the new democracies, than in the “aristocratic ages” ([1840] 1951: 194). Nearly a century later, Janet, using his experience as a psychiatrist, made a similar observation: “Modern society demands of us that we expend increasing amounts of psychic energy… things change quickly, and require a higher pressure to adapt our activities to new situations, in terms of what is original about them. Our civilization is exhausting.” (Janet 1923-24, [1] 1929). More recently, Ehrenberg has analyzed the causes of changes in the subject during the second half of the twentieth century, which led to changes in the nature of the stress which the subject must now face. According to him, three forms previously regulated conduct: “disciplinary rules, which found their definitive forms at the beginning of the 20th century with the inventions of Taylorism and Fordism; rules of conformity (to opinion, tradition and authority); and prohibition rules” (2001: 27). Ever since, individuality itself has been involved. The individual is no longer a member of a community as before and is thus under a pressure which is then concentrated on this single individuality. He says that the “change in the concept of depression was made in a context of regulatory change that became evident in the late 1960s. Indeed, the traditional rules for the management of individual behavior were no longer accepted and the right to choose one’s life began, if not yet to be the norm in the individual-society relationship, at least to be part of custom. By the time that depression had begun to spread in general medicine and manners, French society had indeed entered its great transformation: it had left the world of the gentry and the peasant farmers, and the immobility of class destiny” (ibid. 34). From the 19th to the 20th century, the mental universe of the individual had changed: “Depression began its ascent when the disciplinary model for behaviours, the rules of authority and observance of taboos that gave social classes as well as both sexes a specific destiny, broke against norms that invited us to undertake personal initative by enjoining us to be ourselves.” (Ehrenberg 1998: 10-1, trans. 2009: 5).

2Ehrenberg, in focusing mainly on change in the normative system of the contemporary individual, fails at the same time to see that apart from the regulation [2] that derives from changes in constraint, authorization and commands, the integration of the subject is greatly reduced, especially in terms of its conjugal integration. The increasing mass of people living alone or without a partner is a characteristic of contemporary Western societies. In his work on suicide, Durkheim demonstrated the particular harm of living without a partner ([1897] 1997), a finding that persists more than a century after his seminal work. Therefore, the question arising is that of the causal candidate for the rise of depression between 1970 and 2000. Is it the change in normative regime following changes in the social regulation of the subject including change in the employment structure? Is it the structural change in the population due to the increase in the proportion of single people or both together, and to what extent?

3The issue of increased prevalence of depression comes into its own in that it reveals the changes in society and its members that have stronger effects on some people that affect their mental health and even drive them to suicide. The various expressions of discomfort thus play a role in revealing the more or less evident social tensions, that are experienced by people with certain social characteristics. By contrast, the expected effects of increased stress may not in fact occur. Some of them show the change over time in the social meaning of an object. For example, in moving from what was once called the “single mother” to what is known today as “head of a single-parent family,” it is easy to understand that the stress related to this may have been mitigated by the changes in how society perceives it. Unemployment may have undergone this transformation of its social meaning, because the deleterious effects associated with high unemployment appear to have reduced over the years.

4The article attempts to answer some questions about the impact of loneliness and changes in paid employment by objectivizing them in terms of depression and suicide. In the first part of the article, the theoretical context of the changes of normative regimes is summarized, emphasizing the adaptation of individuals to the stresses brought in by new social configurations. The demographic increases in the share of single people are traced and illuminate the social changes of the second demographic transition in which they are located. In the second part of the article the share of the prevalence of depression due to the weakening of conjugal integration, including solitary living in the second half of the Twentieth Century, is empirically assessed. For this the development of depression between 1970 and 2002 is traced through the ten-yearly “Santé” (Health) survey and through the regular Baromètre Santé (Health Barometer) survey.

5Over this period, a statistical decomposition of a non-linear model makes it possible to estimate what changes are due to the progression of lifestyles of those without a partner and those that appear to derive from other causes. The relationship between depression and the single, widowed, and separated are analyzed. Each type of person living alone shares a common risk of depression because of their very lack of conjugal integration due to the absence of a partner. At the micro level, the difference between the rate of solitary depressive categories is then due to the shock of the impact of separation or widowhood. Excluding the impact of single people and changes in their demographics, the remaining part of the over-prevalence of depression between 1970 and 2002 is thus due to other causes, including changes in social regulation. The links between depression, employment and occupational categories are examined.

6The results drawn from the examination of depression, this singular expression of a predominantly female malaise, needs to be examined against a male counterpoint??”suicide??”able to challenge the uniqueness of conclusions about depression (Aneshensel, Rutter and Lachenbruch 1991). Although the nature of the suicide data does not allow micro-econometric investigation, data are however available over a longer period and seem more stable temporally in their construction than those for depression. They are used to test Ehrenberg’s hypothesis about modification of the normative regime of the individual over a period that includes and goes beyond that begun during the 1960s.

Changes in normative regime

7The Ehrenberg thesis does not trouble with the distinction between integration and regulation. It must be said that regulation seems to be an outdated concept, almost forgotten in any case in the explanation of suicide. However, Durkheim ([1897] 1997) implicitly introduced a truly innovative element with regulation. It lies in the idea of the intervention of social norms, even demanding ones in the mental balance of the individual. Thus, he says, the man marrying agrees to limit his horizons to one sexual partner within monogamous marriage (social norm) and draws from this state a stability whose benefit is objectified by a lower rate of suicide. The interference of the social in the interiority of the subject is also one of the illuminating insights to be drawn by Halbwachs: “Love, hate, joy, sorrow, fear, anger, were first experienced and manifested in common, in the form of collective reactions. It is in the groups to which we belong that we have learned to express them, but also to feel them.” (1947: 9). Individuals are endowed not only with a system of acquired preferences, principles of vision, cognitive structures and patterns of action (Bourdieu 1994), but also a system of acquired feelings and affective expressions integrated through learning, through primary socialization and refined and then reinforced according to a particular social location (sex, age, social position, origin, etc.).

8A discourse about the intensified lives of individuals, and the mental stress that results is a recurring theme in modern times. Can we accept that, from generation to generation, the requirements of societies on individuals have been continuously more pressing and personal? Elias laid the blame for this on the process of civilization itself: “The question that arises is that of a possible link between the high level of efficiency that the monopoly of physical violence attained in parliamentary nation-states, and the high incidence of psychosomatic disorders.” ([1988] 2010: 122). But perhaps the essential issue is not located there. At base, it appears that in reacting to the apparent acceleration and intensification of stress, individuals end up??”willingly or unwillingly??”by adapting to their new conditions of existence in society, even if it means developing new pathologies. Baudelot and Establet stressed that in respect of suicide, “the loosening of traditional constraints do not make societies more enlightened, or individuals more fulfilled” ([1984] 2002: 42). The old stresses disappear with societies of the past and are simultaneously replaced by new pressures corresponding to modern societal configurations. Ehrenberg takes us a big step forward in showing that the individual, while avoiding the old rules of restraint and taboo (the old social norms) eliminates mental stresses that were related to them, in this case prohibition neuroses to the benefit of new diseases.

9In return, the injunction to be a distinct and powerful self, the unconscious adoption of new norms based on the individuation of the subject, added to the “decline of discipline to the benefit of individual initiative” has given way to new tensions whose nerve centre has become the inadequate individual (Ehrenberg 1998, 2001). In fact, ideologies and structuring beliefs such as religious certainties and leftwing “utopias” have been eroded. Job-insecurity, the undermining of occupational groups and a more aggressive management have fragmented employees who have become powerless against performance demands that are both more intense and more individualized (Boltanski and Chiapello 1999; Dejours and Begue 2009). Moreover, the welfare state, the other protective support of the individual, tends to fade away, removing support that is effective for some and potential for others. The combination of these various lines of tension, the emergence of an individual exposed to the evaporation of collective groups have worked together to create an empty field to be covered by a new subject in translation between two normative systems. Depression, a sort of precipitate of modern societal changes, then crystallizes the mutation of the regulatory regime of the subject during the twentieth century (Ehrenberg 1998, 2001). It thus remains to empirically confirm the theoretical analysis using data on depression and suicide.

Data and methods

Sources and measurement of depression

10Depression is in reality different pathologies that can be seen as forming a hierarchy according to the severity of the disease in terms of the inability of individuals to fulfill their social roles [3] (OMS 1996). Depressivity covers depressive conditions often identified in the general population by the diagnostic tools used in epidemiology (depressivity questionnaires using scales such as the CES-D score [4]). For convenience, we use the terms depressivity and depression interchangeably.

11Estimates of depressivity vary greatly according to the survey protocol, the method of detection and the indicator adopted. They do not produce the same level of prevalence even for a similar or identical year (Pope and Lecomte 1999; Leroux and Morin 2006; Lamboy, Leon and Guilbert 2007; Sapinho et al. 2008). The goal here is less to reach a hypothetical measure of the “real” level of depression than to highlight developments that make sense in terms of the weakening of family integration and changes in the regulation of individuals. The “Santé” (Health) surveys for 1970 and 2002 considered depressives to be those who reported a disorder linked to depression, either spontaneously or as a reason for treatment. The indicators thus constructed are comparable between the two years.

The “Santé” (Health) surveys

12The decennial surveys on health and medical care in Metropolitan France were set up in 1960 to measure the medical consumption and the state of health of its inhabitants. A first visit was made to collect data on the medical conditions declared by the members of the household. From three to five following visits, according to the surveys, were made to record medical consumption. This study used the oldest available survey (1970) and the most recent (2002), [5] from which we could build an indicator close to these two dates, which meant that we could not use the interim surveys of 1981 and 1991. By construction, these two latest noted in a related manner the consumers of anti-depressants and those reporting depressive conditions, which was not the case in 1970 and 2002. However, the identification of depressed people by the use of anti-depressants raises several methodological problems. The population of consumers only partially coincides with that of depressed people. [6] In addition, the increase in consumption over the years cannot be considered as reflecting the progression of depression. Indeed, prescription (of anti-depressants) depends heavily on pharmacological advances and medical prescription practices (Amar and Balsan 2004). The inability to distinguish depressed people as identified by reasons for care outside of those using antidepressants has led us to avoid the use of the 1981 and 1991 surveys.

The indicators of depressivity being used

13In 1970, one person answered for all household members. The 1970 survey had no questions about mental health. Depression does not appear explicitly in the list of diseases proposed to respondents, increasing the risk of under-reporting. The wording of the question was: “Here is a list of chronic diseases and disabilities, is any member of your household suffering from one of them?” Twenty-two items of diseases or groups of diseases were proposed, but none referred to a mental illness, only a twenty-third item, “Other diseases: which one?” allowed the respondent to declare any disease not included in this list, including depression. To improve the reliability of information, we have narrowed the field to respondents aged over 18 years with the status of reference person or partner. All diseases reported or identified via the “reasons for medical care” were coded according to a specific classification. The category “depressive syndrome” in the surveys allows to identify people with depression (although we cannot distinguish spontaneous reports of reasons for medical care).

14In 2002, the question used was: “Do you currently have one or more chronic disease(s) (i.e., an illness that lasted a long time [or will last long] or returns [will return] regularly)? If so what is it?” Then a second question completed the first: “Apart from this or these chronic diseases, do you currently have other illnesses or health problems?” These two questions were open. Following collection, the diseases and disorders reported in the first visit or identified during the investigation as “reasons for medical care” were classified according to the tenth revision of the International Classification of Diseases (ICD-10). In order to approach closer to the definition of “depressive syndrome” used in the 1970 survey, the ICD codes we selected concern mood disorders related to depression: depressive episodes (F32), recurring depressive disorders (F33), bipolar disorder (F31), persistent mood disorders (F34), mixed anxiety and depressive disorders (F41.2). To optimize the comparison between these two surveys, the people considered to be depressives are those who declared this spontaneously or gave it as a reason for consultation. This indicator is more restrictive than those used today. [7]

The Baromètre Santé (Health Barometer)

15To test the hypothesis of a differential impact between the event of change of marital status and the state resulting from this change over the longer term, we used data from the Baromètre Santé (Health Barometer) 2005 of INPES (Beck, Guilbert and Gautier 2007), which has the date of widowhood and makes it possible to distinguish the impact of widowhood from the period following it. The date of separation was unfortunately not available for those who had separated. In a conventional and instrumental way, we have used the first eighteen months following the death of a spouse to have a period approaching an average period of mourning [8] and to keep a sufficient number of respondents in the sample. The depressivity indicator that we used in this study is the Duke score, ranging from 0 to 100, calculated from the answers to five questions about the state of mental health (Appendix 4). The higher the score on this measure, the greater is the probability of being depressed (Guillemin et al. 1997). One of the advantages of using a quantitative indicator is to allow the implementation of a more robust linear regression that does not require specification of a conventional threshold for depressivity, a method that is still controversial.

Analytical methodology

16It is difficult to ignore the sex of individuals in the analysis of depression. Pathological manifestations of malaise are known to be gendered. Thus, sexual differences in exposure to negative stimuli have been successively highlighted (Gove 1972; Gove and Tudor 1973), as have differential vulnerability to stress (Pearlin and Schooler 1978) and gendered responses to stressors directed by the structures of interiority constructed differently for each sex during primary socialization (Coster 2005; Cousteaux and Pan Ké Shon 2008, 2010). There is therefore a need to assess changes in the prevalence of depression by sex of individuals as it is twice as great for women. Logistic models of depression simultaneously evaluate the risk of men and women by crossing each independent variable with sex. The introduction of an additional indicator of the sex of the individual in the regressions neutralizes its specific effect. In other words, the men-women variables are cleared of the initial difference owing to gender.

17The level of identification of depression depends on the questionnaire protocol, the nomenclatures of classification of the disorder, knowledge of symptoms and the degree of its social acceptability for his statement. The person interviewed must know they are suffering from the disease in order to declare it. But with medical advances, the development of prevention and treatment of health problems, knowledge and diagnosis of diseases have evolved over the thirty years between the two surveys. Different biases are thus associated with the measurement of disease between two distant dates (Sermet and Cambois 2002), especially with regard to mental disorders because of the popular and negative representations of “madness.” Descriptive statistics of depression are not immune to these biases.

18However, an instrumental way to partially circumvent this problem is to use the statistical Blinder-Oaxaca Decomposition modified for nonlinear models by Fairlie (2005) (Appendix 1). This econometric method makes it possible to separate out what is due to the increase in numbers of single persons in the calculation of the rise of depression, from what is due to the other causes changing the levels of prevalence of depression at a given date. It is then possible to evaluate “all other things being equal” what part in the lack of integration is due to the increase in numbers of single people without the effects specific to each period. This econometric technique serves to undifferentiate the impact of depression on the two dates selected as the regression coefficients from the two surveys are stacked up. Statistically, changes in contributions to the increase in depression derive from the substitution of variable after variable of the distribution in 1970 by that of 2002 (Appendix 1).

So that a thousand lonely people might bloom…

19Basing his views on the observation that the greatest number of suicides are among the single, divorced and widowed compared to married persons, and concerned about the direction taken by society, Durkheim was against divorce. Vain hope. The post-World War II period saw the flowering of intermittent couples and of the lonely life without a spouse. In Western countries, the decade 1965-75 was a pivotal period in which the privacy of individuals and their individual behaviours were profoundly reworked. These changes could be seen in objective terms by a major shift in demographic and cultural trends, marking the beginning of what some have called the “second demographic transition” (Lesthaeghe 1995). Thus fertility decreases, family size is reduced and homogenized, the rates of cohabitation and the number of births outside marriage are amplified, age at first marriage rises, the rate of married people is shrinking, family breakups and divorce are spreading, and episodes of solitary life multiply. These demographic changes come at a moment when there was an exceptional combination of economic, social and cultural circumstances. Society was undergoing profound changes with two major features: the empowerment of the individual and the empowerment of women (ibid,).

20This period of change is characterized by the massive influx of baby boomers into adulthood within an extremely favourable economic environment of virtually full employment and rising standards of living for families. The arrival of these generations of young people (jeunes) was accompanied in France by an ideological challenge to the institutional and integrative triptych Work-Family-Nation (Travail-Famille-Patrie) which de-dramatized the solitary life and provided an indictment of the “bourgeois” family. Moreover, women were being empowered through the synergy created by a higher level of education, leading to new requirements for their partners, and by the rise in female employment allowing some financial independence. By gradually entering the labour market on a large scale in the second half of the twentieth century, women were simultaneously exposed to the inherent tensions of paid employment, particularly those faced by those in modest positions. To this was added better birth control than in the past, allowing more control over their fertility calendar. In addition, the dissemination of feminist ideas made it possible to take a fresh and less complicated view of female autonomy, that had posed many problems in the past (Kaufmann 1999). Since 1960, life expectancy has increased and the mortality gap between the sexes has widened, increasing the number of older women who, therefore, are forced to live alone more often than men (Gaymu et al. 2008), since finding new partners is more difficult for them than for the latter (Cassan, Mazuy and Clanché 2001).

21Under the combination of new requirements that reflect and permit these economic, demographic and cultural changes, marital integration (and partner integration) has weakened, couples have become rarer and their ties more contractual. The modern family is more aptly characterized by individualization, the quality of relational links and its “nationalization” (Singly 2007). Between the French censuses of 1962 and 2005, people living alone have gradually increased from 6% to over 14% of the French population (Table 1), and even 16.2% of people 15 years and older in 2008 (INSEE 2012). In this period, the proportion of single-parent families has more than doubled and now represents 8.1% of French households. Today, 41.4% of households consist of people living alone or without a partner. This relative mass of single, separated, widowed and single parents represents a population that is particularly permeable to malaise, regardless of the forms it can take to express itself: suicide, suicide risk, depression, alcohol dependence, feelings of loneliness, etc. (Cousteaux and Pan Ké Shon 2008, 2010). Therefore, the increase in the number of single people should logically be accompanied by a corresponding increase in expressions of malaise.

Table 1

Changes in the proportion of people living alone in France (in %)a,b

Table 1
1962 1968 1975 1982 1990 1999 2005 Persons living alonea 6.3 6.6 7.7 9.1 10.6 12.9 14.2 Households of persons living alone 16.6 20.4 22.2 24.5 27.2 31.0 32.9 Single-parent households - 2.9 3.0 3.6b 6.6 7.4 7.6

Changes in the proportion of people living alone in France (in %)a,b

Field: Private individuals and households.
a Individuals living in private households in metropolitan France.
b Children over 25 years of age have been included in the definition of a single-parent family since 1982.
Source: Recensements de la population (Census), enquête annuelle de recensement (Census survey) 2005, INSEE. Calculations by the authors

22But to describe single people as if they were a specific entity does not help in understanding their particularities. Indeed, each category of single people??”unmarried, separated, widowed??”tendentiously emphasises three distinct moments in the life cycle, outlined with a broad brush: youth, maturity, old age. They form groups with unequal proportions. In the 1999 census, among the 9.1 million French people living without a partner in their home, 37% were single, 27% widowed, 12.5% divorced, 4.5% married and 19% single parents. Since the 1982 census, the increase in the number of single people was nearly 60% due to the increase in the number of single people and accounted for 20% of those who are divorced (Pan Ké Shon 2002).

23Strictly speaking, the lack of conjugal integration that is shared by the single, the separated and the widowed, is expected to crystallize at the same level in an expression of malaise, and in this case depressiveness. Specifically, the lack of conjugal integration derives from the inability to alleviate the stresses of everyday life through a partner. There is both a “structural” deficit of social support, and of interactions which are limited in frequency and duration (Hughes and Gove 1981). However, as we will be able to verify there are levels of depressivity that can be seen to be differentiated between these three categories of people living alone. We must recognize that this difference is due to another factor than marital or partner integration since each of these three categories lives without a partner. It should be recalled that it could be the presence of children which explains the differential since it can be shown that children do not have any protective effect on depression or on other expressions of malaise (Cousteaux and Pan Ké Shon 2008, 2010). Implicitly, people without a partner embrace attitudes towards life whose horizons distance them from each other because the social order that has to be achieved within a couple is clearly not as acute an issue for the single, widowed and divorced. However, the lower pressure of social gaze focused on these states of singleness, widowhood, separation, or divorce was not accompanied by any lesser harm resulting from a solitary life. An alternative hypothesis would simply start from the lack of the impact of widowhood or separation for single people. Differential prevalence of depression would then be due not to the state of the person without a partner, but to the event or the impact of separation and death of a spouse.

Combined increase of depression and single people between 1970 and 2002

The spread of depression

24Depression affects nearly three times more women than men (Table 2). This is what is generally observed regardless of surveys and indicators (Leroux and Morin 2006; Sapinho et al. 2008). The tripling of the rate of depressivity between 1970 and 2002 tends to confirm the thesis that work has become more pathogenic in society because of the individualization of the subject (Ehrenberg 1998). These findings are also supported by other analysts of changes of management of employees for the years 1980-90 (Beaud and Pialoux 2003; Dejours and Begue 2009). These changes correspond to the negative aspects explored in the model of work stress of Karasek and Theorell (1990): high psychological demands, low autonomy of the employee, less social support derived from colleagues because of their competitive setting. Boltanski and Chiapello (1999) had already noted the profound changes in the management of employees, that justify describing them as the expression of a “new spirit of capitalism.” Our results do not conflict with this statement because depression really increases for men until their sixties and then declines at the ages when there is large-scale retirement in France. For women, the increase of depression is much more dramatic and consistent with their large-scale involvement in the labour market over the past three decades. If changes in depressive prevalence between 1970 and 2002 are unreliable because of remarks made earlier (see Data and Methods), the gender gaps remain relevant, as these two populations are subject to the same potential biases. It is observed that whereas the curves are generally parallel between the sexes, in 2002 there was an increase in female depressiveness for those in their thirties which continued until the age of 60 (Figure 1).

Table 2

Proportion of depressive people according to sex (in %)

Table 2
1970 Survey 2002 Survey Ratio 2002/1970 Men 0.6 2.3 3.8 Women 1.8 6.0 3.3 All 1.3 4.2 3.2

Proportion of depressive people according to sex (in %)

Note: People identified as depressives.
Field: Reference persons and partners aged at least 18.
Source: “Health” (Santé) surveys 1970 and 2002.
Figure 1

Proportion of depressive people

Figure 1

Proportion of depressive people

Field: Reference persons and partners aged at least 18.
Source: “Health” (Santé) surveys 1970 and 2002.

25Women are more often employees in positions and in jobs where work-demands can be heavy and autonomy limited, two characteristics deleterious to mental health. Thus, the OECD report on mental health notes: “These jobs involve more often than in other occupations, high cognitive demands and low decision latitude??”a combination that can generate a situation of tension at work, that is to say, an unhealthy workplace stress, which is a recognized mental ill health factor.” (OECD 2012: 223). However, female depression peaks between 50 and 70 years old when economic activity rates are low. [9] This finding questions the part of Ehrenberg’s thesis on the individualization of employees which would have resulted in more numerous depressions, since the sharp increase focuses on women and occurs more among inactive people… However, the results do not formally contradict the thesis, since depression appears to increase for all ages, including those who are economically active.

Single people, depression and selection bias

26The strong growth in living alone denotes upheaval in relationships with others, especially in intimate relationships. Institutional constraints are lower than before, the mutual needs of the couple are higher, and episodes of loneliness consequently more frequent. However, in the relationship between living alone and depression, there is still the hint of a potential selection bias by which the depressives could be disabled in terms of their ability to form a couple or that the union would be weaker for those having formed one. Therefore, it would not be single people who would be more receptive to depression, but depressives who are more frequently alone. Studies using longitudinal data indicate that marital selection related to mental health is either limited (Waldron, Hughes and Brooks 1996), or nonexistent (Simon 2002). A study of a cohort of young adults indicates that there are equal chances of marrying depressives for both sexes. Only alcoholics are further apart in the marriage market (Horwitz and White 1991). The marriage break-ups show that married people who have subsequently separated report more depression and alcohol problems before failure than those who remained married (Simon 2002). However, the circularity of the relationship (disagreement–depression–break-up–depression) makes the search for an initial cause an unrealistic one (ibid.).

27Others note that no evidence so far has been provided to support this long-standing suspicion of selection bias (Hughes and Gove 1981). On the contrary, they highlight the unique and preservative benefits of the partner of which single people are deprived: the most frequent and often daily interactions, more than with any other partner, the longest durations, the more intense commitment, the mutual relations that bond and support and that inevitably lead to greater integration. These are what others have summarized as emotional and instrumental support (e.g., Jong-Gierveld and Tilburg 1987). Furthermore, social support is as an element of preservation or acquisition of self-esteem by promoting itself through privileged interpersonal relations (Turner and Turner, 1999). Undermining self-esteem promotes mental disorders including depression (Bibring 1953; Jacobson 1971; Fredén 1983; Tousignant 1992). Finally, recent widowhood is a random event that is beyond the effect of selection of depressed people and the greatest difficulty in the formation or durability of the couple. The fact is that widows are not only more often depressed in the months after the spouse’s death, but still when the shock is more distant in time.

The increase in single people

28People living alone or without a partner??”unmarried, single parents, widowed, separated??”make up a heterogeneous group with experiences and perspectives on life that distance them more from each other than make them alike. In 1970, the widowed were almost half of all single people, but gradually it is the unmarried who have become the dominant group (Table 3). Today, children no longer stop co-habiting with their parents in order to create a new household as a couple, as they did before. The period that opens after they have left the nest is one of experiences, and often proclaimed as such (Van de Velde 2004). This new “freedom,” these diverse experiences of self-building can be more destabilizing than before and consequently be expressed by mental illness. Simultaneously, the proportion of separated and single parents have greatly increased, and more moderately for widows, favouring an increased prevalence of depression.

Table 3

Distribution and proportion of people living without a partner according to the “Health” surveys (in %)

Table 3
1970 Survey 2002 Survey Proportion Distribution Proportion Distribution Unmarried 3.2 24.7 7.9 33.7 Widowed 5.9 46.4 6.8 28.9 Separated 1.1 8.3 3.9 16.5 Single parent over 65 years old 0.9 6.9 0.5 2.0 Single parent under 65 years old 1.8 13.7 4.4 18.9 Total 12.8 100.0 23.5 100.0

Distribution and proportion of people living without a partner according to the “Health” surveys (in %)

Field: Reference persons aged at least 18.
Source: “Health” (Santé) Surveys 1970 and 2002.

Relationship between depression and single people

29A logistic model estimating the risk of depression was used to highlight the links between depression and living alone when they are freed from the effects of other sociodemographic characteristics. The model is controlled by sex, the situation of the individual in the household, age group, occupation and level of education. Each characteristic is crossed with sex in order to take into account the gendered nature of the indicator and identify gender differentiated effects. An indicator of the sex of the respondents was introduced, and it reduces the initial depressivity differential between men and women.

30In 1970, single men and single women aged under 65 living alone with children were the only groups for which the risk of depression was significantly higher compared to those in couples with children (Table 4). Overall, the men’s results for 1970 are not significant because they count insufficient numbers of depressives. Thirty years later, with the exception of single parents aged 65 or older, all people living without a partner??”regardless of their gender, marital status and responsibilities for children??”are at significantly higher risk of depressivity than those in a couple with or without children. In thirty years, not only the number of single people has increased, but their risk of depressivity has been accentuated.

Table 4

Extract from logistic models of risk of being depressive

Table 4
1970 Survey 2002 Survey Intercept - 4.3*** - 4.3*** Sex 0.3 1.5*** Men Women Men Women Single 1.2 ** 0.2 - 1.1 *** 0.8 *** Widow(er) 1.1 - - 0.1 - 1.5 *** 0.7 *** Separated - 12.2 - 1.0 - 1.2 *** 1.1 *** Single parent 65 years old or over - 10.9 - 0.1 - 1.6 - - 0.1 - Single parent 65 years old or under -12.3 - 1.0 *** 1.3 *** 0.8 *** Couple with children Ref. - Ref. - Ref. - Ref. - Couple without children 0.0 - 0.1 - 0.4 * 0.2 - Other - 0.2 - - 0.6 - 0.2 - - 0.4 -

Extract from logistic models of risk of being depressive

Note: *** p < 1 %; ** p < 5 %;* p < 10 %
A logistic regression model is established for each survey, each variable of the models is crossed with sex (from the full model presented in Appendix 2).
Field: Reference persons and partners aged at least 18.
Source: “Health” (Santé) surveys 1970 and 2002.

31Recent results obtained from different depressivity indicators show that in respect of depression widowhood seems the most pernicious form of the three categories of living alone for men (Table 4 and Cousteaux and Pan Ké Shon 2008, 2010). The life experiences of loneliness do not have the same meaning or the same consequences for men and women (Umberson, Wortman and Kessler 1992). The departure of a partner frees women of extra chores whereas the opposite is the case for men who must take on new activities which are all the more difficult as they have rarely practiced them in the past. Yet even the slightest familiarity with housework seems an insufficient explanation as separated men who experience similar responsibilities and conditions are less adversely affected by depression than widowers. It is particularly after the break-up, and it is assumed by extension after the death of a spouse, that women suffer from a reduction in their standard of living and men from loneliness (Gerstel, Riessman and Rosenfield 1985; Umberson, Wortman and Kessler 1992; Dykstra and Fokkema 2002).

32The heavy penalty of loneliness for the widowed compared to other situations of loneliness is not due to the lack of a partner alone, because they share this with single people and the separated. Other hypotheses may be advanced to explain this heavy penalty. 1) Men become aware at the time of widowhood, rightly or wrongly, that finding another partner will be unlikely. This perspective would be more deleterious for them, rather than for women, who have already internalized this fact, as the marriage market at this time of their lives is not likely to work in their favour. 2) In a schematic feminist vision, the loss of a partner is not the same for both sexes: widow leads to the loss of a dominant partner, whereas the widower loses a dominated partner. It is therefore understandable that the loss of a partner is more harmful for men and this is crystallized in a higher prevalence of depression. However, these rational assumptions have yet to be confirmed, as we shall see.

Shock and the state of widowhood

33It can be deduced from the difference in the risk of depression between each type of lone person that it is not something based on a lack of conjugal integration. Indeed, the unmarried, the separated, the single parents and the widowed share the same experience of the absence of an integrative partner. We must therefore seek another dimension to explain these differentials. If all widow(er)s [10] suffer the loss of a partner, separations, in turn, often occur at the initiative of one member of the couple, and this person does not suffer the same degree of toxicity from the break-up. One might think that, as they affect overall one in two of those separating, their depression prevalence would logically be much lower than those seen in widowers, who, themselves, suffer all of the separation from the deceased partner. In fact, the combinations between the initiative in break-up and reactions of the partners are more complex. Separation may as readily come about following a gradual degradation in relationship, represent a satisfactory situation and a logical outcome for each partner, even an opportunity for release or renewal, as being experienced in terms of betrayal, fear of loneliness and the loss of income (Aseltine and Kessler 1993; Kaufmann 1999; Singly 2011). Situations of break-up are thus more heterogeneous than those of widowhood.

34Within widowhood, it is necessary to differentiate between two distinct phenomena: the shock felt during the period covering a few months after the event, and the state, which is the normal state of life without the dead partner beyond the impact of the event itself. It seems reasonable to think that the shock of widowhood is composed of various elements. The emotional shock of grief, feelings of hopelessness, and the disturbance due to the “sudden change in the social universe of reference” (Besnard 1987: 103). Durkheim’s anomie, understood as the loss of bearings, can be seen in outline form as a two-speed process. There is a violent destabilization at the time of the shock, and then a stabilization over time. Here we find an illustration of the distinction between “acute anomie” and “chronic anomie” (Durkheim [1897] 1997) or “crisis anomie” versus “institutionalized anomie,” in the words of Besnard (1987: 102).

35The temporal dynamics of widowhood make it easier to understand the interconnections between integration and regulation. The shock of widowhood is part of an “acute” deregulation, a sudden destabilization due to emotional shock, a violent reconfiguration of the mental landscape of the individual from the loss of their spouse. Add to this the powerful destabilization due to the loss of bearings, the lack of structure for the individual through the minor daily activities previously carried out in relation to the partner, the absence of this referential framework that is binding but also structuring. These feelings and this shock gradually decrease over time, as illustrated by the popular saying “Time heals the wounds.” Destabilization from the loss of structure of the individual’s daily activities will be felt particularly during the period of shock. In fact, the old habits will gradually be substituted for other ways to perform minor daily tasks, resulting in a new structuring of time. However, the regulation that had prevailed is one that becomes looser as marital stress is released, one that the individual imposes on themself, in terms of course of the norms and values they have internalized. This disruption is not just due to the vertigo of possibilities so dear to Durkheim (Besnard 1987), but also to the lack of the social support previously provided by the partner. Her/his presence protected, reassured, and helped to release tension, to communicate about things in life and which finally was what was of importance about ego in the eyes of this particularly significant other, or what in Durkheimian terms is called integration. In this sense, the state of domestic deregulation or “chronic anomie” corresponds to a lack of conjugal integration.

36Two important results can be drawn from the empirical evidence. Firstly, confirming our intuitions, the shock of widowhood has a much more powerful impact on mental health than the ordinary state of the widower, and this is regardless of sex (Table 5). Secondly, there is no significant difference between the degree of depressivity of single people and those in the state of widowhood, regardless of sex. In relation to depression, widowhood appears to have the same harmfulness as living alone. In other words, it is the common lack of a partner, the deficit of conjugal integration, that makes single and widowed people so similar. These findings would need to be consolidated with data from other studies. It can be assumed that if the data on separation could distinguish on the one hand, the impact of the state and, on the other hand, the initiating member of the couple who underwent separation, they would reveal similar trends, although break-up situations are more diverse and complex (Kaufmann 1999; Singly 2011). It is noteworthy that the increases in the number of widowers, separated and single parents also will be accompanied by correspondingly increased depression and the impact of the strong development of single life will be more limited over the thirty year period in which we are interested.

Table 5

Extract from the linear regression of the score of depressivity

Table 5
Baromètre Santé 2005 Constant 19.7*** Men Women Sex Ref. 4.2 *** Single alone 6.0 *** 2.5 *** Separated alone 4.2 *** 3.3 *** Widow(er) alone, state 6.2 *** 2.9 *** Widow(er) alone, shock 8.8 * 6.8 *** Single parents 3.6 ** 2.0 *** Couples Ref. - Ref. -

Extract from the linear regression of the score of depressivity

Note: *** p < 1%; ** p < 5 %; * p < 10 %
Depressivity indicator is built using the Duke score (Appendix 4). The results come from a single linear regression model where the variables are crossed with that of sex. The model is controlled by the sex, age, qualifications, employment status, income, important events during childhood and the past year (full model available from the authors).
Field: Individuals of 12 to 75.
Source : Baromètre Santé 2005 (INPES). Analysis by the authors.

Impact of the increase in single people on depression

37The statistical decomposition proposed by Blinder-Oaxaca extended to logistic models by Fairlie (2005) allows the assessment of the contribution of each variable to the development of depression prevalence between 1970 and 2002, while controlling for other variables in the model. It looks at changes in the structure of the population and in particular the increase in numbers of single people at risk of being depressed (Appendix 1). The establishment of similar models for making strict comparisons between 1970 and 2002 restricts us to using a source that contains less information, the 1970 survey. The model constructed is thus more descriptive than explanatory. However, it meets the aim of estimating the impact of the erosion of marital integration on the rise in the proportion of depressives, and avoids the bias of the intertemporal measure that was mentioned earlier. A model for each sex is established to account for any conflicting trends in selected independent variables. Let us recall that the indicator used here tends to minimize levels of depressivity and thus the proportions better reflect the changes.

38The changes in demographic characteristics of the population that took place between 1970 and 2002 contributed to an increase in the rate of depressivity of about 0.5 for men and 0.2 for women, representing 34% and 5% respectively of all increases for the period (Appendix 3). These totals are the result of the sum of the positive and negative contributions of the following characteristics to the prevalence of depression: an ageing population, changes in socio-professional categories and transformations of the family.

39The new balance in the distribution and the levels of couples and single people contributed more than a fifth of the overall increase in depression since the end of the postwar boom (Table 6), but this did not occur uniformly. Separated people make up the largest contributors (7.3% and 7.7% of the share of contributions for men and women), followed by the contributions of single men and single parents. In total, the intertemporal changes in household types account for over a fifth (20.4% for men, 22.2% women) of the increase of depression between 1970 and 2002. The marital situations that have made depression increase are bachelorhood and separation for men. For women, it is separation and the particular form of separation with child(ren), e.g. single parenthood, which is 85% female.

Table 6

Logit and logistic decomposition likely to be depressed between 1970 and 2002. Excerpts from Appendix 3 (in %)

Table 6
Men Women Rate of depressives in 2002 2.02 5.62 Rate of depressives in 1970 0.65 1.82 Difference 1.37 3.80 Logit Relative raw contribution Logit Relative raw contribution Single 1.03 *** 0.07 5.4 0.70 *** 0.09 2.3 Widow(er) 1.38 *** 0.01 0.9 0.60 *** 0.07 1.8 Separated 1.10 *** 0.10 7.3 1.11 *** 0.29 7.7 Single parent 65 years old or over 1.19 - - 0.01 - 0.4 - 0.10 - 0.00 0.0 Single parent 65 years old or under 1.14 *** 0.00 0.1 0.81 *** 0.21 5.4 Sub-total single people - - 0.17 13.3 - - 0.66 17.2 Couple with child(ren) Ref. - - Ref. - - - Couple without child(ren) 0.28 * 0.11 8.0 0.14 - 0.08 2.0 Other 0.11 - - 0.01 - 0.9 - 0.46 * 0.12 3.0 Total situation in the household - - 0.27 20.4 - - 0.86 22.2 All contributions - - 0.47 34.4 - - 0.19 4.9

Logit and logistic decomposition likely to be depressed between 1970 and 2002. Excerpts from Appendix 3 (in %)

Note: *** p < 1%; ** p < 5 %; * p < 10 %
This table is extracted from the complete model of statistical decomposition presented in Appendix 3. It presents the risk of being depressed (logit) calculated for each sex by stacking the two samples (see Appendix 1), then the contribution of each category to the rate of depressives and the relative share that each brings to the contribution to the difference between the rate of depression in 2002 and 1970.
Field: Reference persons and partners.
Source: “Health” (Santé) surveys 1970 and 2002.

40The worsening of depressivity related to weakening conjugal integration alone is mainly due to the increase in the number of people without a partner, 13% of men and 17% for women in the period, or about 15% for both sexes combined. If the increase of living without a partner is reflected in a more substantial prevalence of depression, its own impact, however, remains limited. Assessment of the weakening of conjugal integration between 1970 and 2002 shows that it has undermined part of the population, but that it did not push it into “widespread conjugal anomie.” In fact, 85% of the increase in depression is not explained by the increase in the number of single people, but is due to other causes.

The sick depression of unemployment

41What have been the consequences of changes in paid employment for individuals, of the intensification and modification of management, and are they visible throughout the social categories during the “Thirty pitiful” (“Trente Piteuses”) years that followed the post-war boom? The information available in the “Health” surveys is limited to the employment and socio-occupational status of the respondents. The economic sector is not available, nor the degree of autonomy or the intensity of work, or social support derived from colleagues, all dimensions that influence feelings about work (Karasek and Theorell, 1990). The data are insufficient to conduct a thorough investigation. But even when limited to these two variables, temporal comparisons in terms of depression are valuable. Changes in the occupational structure and the place of women have led to very different gendered contributions. First, the decrease in the number of housewives helped reduce depression by 16% between 1970 and 2002 (Table 7 Other inactive category). The massive influx of women into the paid labour market objectively contributed to their empowerment and their lesser dependence vis-à-vis their male partner. Unlike for men, the increase of wage-earning by women has resulted in a change of status through the gain in family and occupational autonomy (Maruani 2005). By accessing paid employment, women have acquired both interaction at work, relationships of sociability with work colleagues that are more rewarding than their confinement to the home, a sense of empowerment especially when compared to their mothers (Baudelot et al. 2003), a more valued symbolic role, and a step towards the rebalancing of power within couples. Work is also an opportunity for personal development, at least for some of them.

Table 7

Logit and logistic decomposition of the risks of being depressive between 1970 and 2002. Appendix 3

Table 7
Men Women Rate of depressives in 2002 2.02 5.62 Rate of depressives in 1970 0.65 1.82 Difference 1.37 3.80 Logit Contri-bution Relative share (%) Logit Contri-bution Relative share (%) Constant - 0.89 *** - - - 1.14 *** - - Farmer - 0.39 - 0.04 2.9 - 0.16 - 0.00 0.0 Business owner - 0.30 - 0.02 1.5 - 0.34 - 0.02 0.5 Senior manager - 0.23 - - 0.01 - 0.4 -0.66 *** - 0.06 - 1.5 Manager - 0.24 - 0.01 0.6 - 0.34 ** - 0.05 - 1.2 Employee Ref. - - - Ref. - - - Manual worker - 0.14 - 0.06 4.0 - 0.26 - 0.05 1.3 Unemployed 0.91 *** 0.11 8.0 0.44 *** 0.15 3.8 Retired 0.56 - 0.05 3.8 0.38 *** 0.28 7.2 Other inactive 1.81 *** 0.30 21.6 0.38 *** - 0.61 - 16.0

Logit and logistic decomposition of the risks of being depressive between 1970 and 2002. Appendix 3

42The development of women’s work was actually beneficial for women, but this benefit is unevenly distributed. Thus, the contributions of women in senior and middle management have reduced the level of depressiveness between 1970 and 2002 more than for the junior categories (Table 7). Their greater autonomy and greater control over their work are recognized factors promoting good mental health (Karasek and Theorell 1990; Baudelot et al. 2003). However, this benefit appears more feminine (the coefficients of the logistic models in 2002 are negative and significant for women of the upper classes) than masculine (the coefficients of occupational category are not significant) (third and fourth columns of Appendix 2). Subordinate female jobs (reference category in the model) are also those that are most often part-time, less valued and most poorly paid (Maruani 2005). It is thus this contrast which has probably been captured.

43For both sexes, although it is especially true for men, being unemployed is more damaging for mental health than work. This updates a longstanding result of destabilization of the individual by the loss of active employment at a time in the life cycle when work remains the norm (Rushing 1979). In other controlled variables, the institutional or actual situations of inactivity (retirement, unemployment, other inactivity) lead to an over-increased depressive risk for women and men (over and above that of retirement for the latter) (Appendix 2). The male Other inactive category which includes students, those with disabilities, people who have never worked, show significantly increased contributions to depressiveness. In a previous study, it was shown that the people who are afflicted with a penalizing disability who are most affected by various expressions of malaise are more often men (Cousteaux and Pan Ké Shon 2008, 2010). Apart from this case, it is mainly the increase in the number of unemployed and, to a lesser extent, of retirees, who have increased the level of depression identified in this period following the post-war boom.

44This shows that the impact of changes in work on the mental health of individuals cannot be interpreted unambiguously because, despite more aggressive management, work??”when compared to situations of inactivity??”remains a vector for good mental health for varied, changing and sometimes incomplete reasons depending on the social status of the individual (including their sex): the assertion of a prestigious status, self-realization, the increases in autonomy and self-esteem, relational exchanges, structuring of time, etc. (Demazière 1996). This reflection is obviously not intended to hide suffering at work, especially in areas where management has the explicit aim of “easing-out” some of the most expensive workers by making them “crack” psychologically (Dejours and Begue 2009; Du Roy 2009) or areas with high intensification of work without autonomy for employees. The lack of significance of the coefficients related to socio-occupational groups of men in contrast to those high and significant levels associated with single people suggests that the pathological intensification of work is far from having yet reached all employees and all sectors of activity (Appendix 2). The lack of paid work is even more detrimental (like unemployment or disability) when it is suffered by men, whose social role is still that of the “breadwinner.” For women, economic inactivity means a return to dependency on the partner and to devalued tasks, which results in a significant rate of depressivity for all three categories.

45It must therefore be concluded, firstly, that the harmful situations for males are mainly forms of solitary life. They are more harmful than position on the social class scale, but almost on a similar level to unemployment; and secondly, with regard to women, that lower social classes are more penalized. Thirdly, unemployment shows itself to be more pathological than work for men and women. For the latter, even retirement is more depressive than being employed. At this stage, we would therefore be led to believe that the casualization of paid work leading to longer or shorter repeated periods of unemployment and the lack of conjugal integration are the main elements that have eroded the integration of individuals in this period. We can repeat verbatim what Bastide said: “Here we find again the problem posed by A. Comte, and that cannot be ignored: the correlation between the facts of personal disorganization and social disorganization, and the facts show, in an impressive way, that mental disorders are more prevalent in the disintegrated than integrated sectors of a given population.” (1965: 93).

Depression frustrated by suicide

46Nevertheless, we do not really know if the transitions in regulatory regimes since the mid-twentieth century have been accompanied by an increase in mental stress or whether the movement of tensions from one sector to another of society or from one disease to another, as a result of upheavals in nosography alone, have just made them more visible. For their part, the greater acceptability of mental illnesses and the improvements in their screening have tainted intertemporal comparisons of levels with an irreducible bias of doubt about whether we can assess its importance. Unlike depression, the collection of death certificates from suicides avoids this type of bias. Suicide seems to be a more stable indicator than depression and may provide a useful counterpoint. But it is not without problems, because it is also driven by gender (male suicides in France in 2006 were three times higher than those of women), the available data do not make it possible to conduct micro-econometric studies, and it is highly specific because of its radical nature. It is frequently argued that it is under-reported, by roughly 20% according to certain estimates, although a more recent study using a suitable protocol has concluded that this figure is about 10% without really significant gendered differences (Aouba et al. 2011). More importantly, under-reporting ought to have been stable over the past three or four decades (we do not see why it would not) and in this sense does not impede temporal comparisons.

47The evolution of suicide over time creates an unexpected complexity in seeing too linear a relationship between change in the regulatory regime, individualization and increased tension. Thus, the temporal evolution of mortality by suicide, this radical expression of the malaise of individuals, illustrates a change that seems to be better correlated with unemployment (Figure 2). During the postwar boom, the level of male suicide remained low and without significant changes, like the extremely low and stable unemployment rate. Once this period had ended, the rate of suicide and unemployment increased strongly until the mid-1980s and then decreased until the end of the observation period, with a jump in the mid-1990s. As for women, their suicide mortality increased continuously from 1945 to 1986, then fell (with the same jump but of much lower amplitude) to stabilize, unlike that of men, at the high level it had reached before the Second World War. Observers have regularly been tempted to connect these developments automatically with economic disruption, especially rising unemployment, as the profile of its curve is similar to that of male suicides (Figure 2). Thus the psychiatrist Michel Debout claimed in an interview with Le Monde that “the correlation between unemployment and suicide is quite undeniable” and that an increase was to be expected soon (Le Monde 08.02.2012).

Figure 2

Changes in gross rate of suicide mortality from 1925 to 2006

Figure 2

Changes in gross rate of suicide mortality from 1925 to 2006

Note: Data has only been available since 1975. The profile of curves of male and female unemployment is similar, only the level varies, it is higher for women.
Field: Population of France.

48Chauvel has shown that the profiles of the curves of male suicide and unemployment have two peaks rising in the middle of the 1980s and 1990s. These peaks in suicide coincide with the dark years of the economy, whereas on the contrary 1985-90 saw a period of relative economic upturn and a sharp decline in suicide rates. He explained further: “There is a temptation to arrive at the simple idea of an ‘economic’ suicide and reduce the causes of suicide to unemployment or economic hardship alone, i.e., the most material aspects of the ‘crisis’.” (1998: 3). To reinforce his analysis, he refers to Halbwachs: “It is not just the poverty of unemployed workers, the bankrupts, insolvencies and ruins that are the immediate cause of many suicides. But an obscure feeling of oppression weighs on all souls, because there is less activity in general, and men participate less in an economic life that has passed them by, and are no longer looking outside, but are more concerned not only with their physical distress or mediocrity, but with all the personal reasons they may have to want to die.” ([1930] 2002: 394). Perhaps we should simply think that there is no direct link between unemployment and suicide because the effect of unemployment is mediated by elements whose social support is more or less effective for the individual. In addition, the harmful aspects of unemployment and economic depression also spread within the family due to the interpersonal tensions they induce, or as a depressive component in addition to others along the lines of “the straw that broke the camel’s back.” Unemployment also creates a new balance of power between employers and employees, which may generalize tensions to all employees, the decline in activity slowing down the activity of shopkeepers, etc.

49An examination of suicide rates by decadal age-groups and gender between 1979 and 2009 shows that suicide has not just affected those in work, but all individuals, whatever their ages. The categories that appear to be most drastically affected by the economic depression and the peaks in unemployment are people whose age makes them much less likely to be at work. [11] This is particularly clear for men aged over 55 years and for almost all age groups of women (Figure 3). Factors affecting suicide are multidimensional and to reduce the explanation of the higher suicide rate to only one of them is very likely to lead to erroneous conclusions. Several moderating comments can be advanced. First, the cohorts born in 1890s, 1900s, 1910s are more suicidal than those following. They were in 1985, in mean 95 years, 85 years and 75 years (Anguis, Cases and Surault 2002), which may explain, at least partially, the above average suicide rates for people over 75 years compared to earlier and later decades. Second, there has been a period effect which concerns all ages for both sexes, except for women aged 15-24, whose suicide rate is usually very low. Visually speaking the magnitude of the rate of suicide rates by age shows a strong contrast between the higher and lower rates (Figure 3).

Figure 3

Changes in gross rate of suicide mortality from 1979 to 2009

Figure 3

Changes in gross rate of suicide mortality from 1979 to 2009

Field: Population of France.
Source: CépiDc. Graph by authors

50Calculating a simple ratio of suicide rates overcomes this problem. It suffices to relate suicide rates during the period when there was a sharp rise in unemployment to those of the reference year where unemployment is “ordinary” or stable. Thus, the suicide rates for 1985 and 1986 are added in order to have sufficient numbers, then they are divided by two to smooth out the effects of these two pivotal years and finally, they are related to the rates in 1979, a year conventionally chosen for its moderate rates. The results show that the differences in ratios between each age group and sex are low (Table 8). Only women in the 45 to 54 age group have a suicide rate 1.4 times higher than in 1979, whilst for men it is those in the 75 to 84 age group who show a ratio about 1.3 times higher. Third, all male suicide rates in 2009 are lower than those observed before the start of the economic depression (taken here as 1979 because of data availability) for the same age groups, apart from 35-44. Similarly, the female suicide rate for all age groups declines in 2009 compared to 1979 (apart from 45-54 where rates increase by 0.5 per 100,000).

Table 8

Ratio of rates of suicide 1985 and 1986 over 1979

Table 8
Men Women 15-24 years old 1.18 1.02 25-34 years old 1.22 1.17 35-44 years old 1.21 1.25 45-54 years old 1.18 1.40 55-64 years old 1.09 1.10 65-74 years old 1.10 1.15 75-84 years old 1.28 1.15 85-94 years old 1.18 1.22

Ratio of rates of suicide 1985 and 1986 over 1979

51Changes in suicide rates over time suggest that:

  1. There have been worsening tensions since the 1970s, which could be seen in the 1980s and 1990s. Consequently, we can deduce that the rise of depression, closely correlated with suicide (Sher, Oquendo and Mann 2001; Heeringen and Hawton 2009), is not just due to improved screening and social acceptability but it has really increased;
  2. The peak of unemployment in 1985 is accompanied by a rise in suicides regardless of age (except those under 25 years) for both sexes;
  3. Levels of suicide among men and women are mostly lower in 2009 than the already relatively low level of 1979;
  4. Tensions due to new management cannot be seen in the suicide rates;
  5. The tensions due to the gradual and steady increase in the number of single people after the period 1965 to 1975 ought to have resulted in a continuous linear progression rather than an uneven one. In addition, there should be no reason why the level would sharply decline (Figure 2).

52It is noteworthy that the male suicide rate in the early 2000s should have returned to that of the early 1970s, whereas the economic climate has worsened. In addition, management has become more aggressive and the proportion of single people in the population has increased. This means that current rates of suicide should be even lower than the low levels of 1970, if these aggravating factors had been observed in the 1970s. This apparent paradox is somewhat surprising. Has this come about because of the development of self-control over violence, and therefore of self-violence itself in the continuous process of civilization (Elias [1939] 1990)? However, this type of societal change has a certain inertia and it is assumed that the fundamental movements are slow, gradual and imperceptible “to the naked eye,” in contrast to the great variations observed in suicide rates in recent decades. The alternative hypothesis would look to the “abnormally” disturbed socioeconomic conditions of the end of the post-war boom which, over time, have returned to their ordinary state. In fact, the post-1973 generations have only known this state of employment and of the economy, which gives them a certain ordinariness.

Discussion and conclusion

53The identical levels of depressiveness of single-living and widowhood reveal the shared aspects of these two ways of life: the common lack of conjugal integration (Table 5). This is not an insignificant result to be noted in that where other sociodemographic variables are controlled, those widowed and living alone react with the same level of depressivity to the lack of conjugal integration, e.g. the structural lack of social support from a partner. The distinction of the shock of the event of widowhood from the ordinary state of the widower was used to isolate two aspects of regulation during the course of this shock in our results: deregulation by the sudden deconstruction of everyday life and the emotional upheaval occasioned by the loss of a partner (loss of support through separation or death). Until now we have generally stayed within the Durkheimian model of social dysfunction and failure of the conjugal tie, particularly divorce, as leading to suicide. As life without a partner has become increasingly common, it becomes impossible to consider single-living, separation and widowhood a fortiori as social dysfunction, and we should simply observe them as being new ways of living in relationship with others. Halbwachs noted astutely: “But if suicides, on the contrary, increase mainly because social life is complicated, and the singular events that expose people to desperation multiply, they are still a bad thing, but perhaps a relatively bad thing. There is indeed a necessary complication which is the condition for a richer and more intense social life.” ([1930] 2002: 12).

54The frequent observations that society is becoming more complex and pathogenetic (Tocqueville [1840] 1951; Janet 1929; Halbwachs [1930] 2002; Bastide 1965; Ehrenberg 1998) raise questions at least about their scope if not their validity. Once the initial assessment has been accepted, we would then expect to see a continued deterioration in the mental health of individuals and consequently an increase in the prevalence of various expressions of malaise. But this is not what is observed from suicide. The impacts of economic depression and mass unemployment that have profoundly shaken French society from the end of the period of post-war boom, were transmitted to all individuals, even those who were less directly involved in production and less materially affected by its consequences (Figure 3 and Table 8). The findings on the impact of widowhood versus the state of widowhood converge with those of the strong rise in unemployment after the first oil shock, then its stabilization. The impact of widowhood is followed, after a period when its shock is being absorbed, by the ordinary state of being a widow(er). The impacts of mass unemployment and economic depressions would be gradually replaced, during the passage of time, by a new state which would be nothing more than the redefinition of the prior current state, i.e., as an ordinary state. Social impacts as events have given way to a more unstable social situation but one that becomes normality, all the more so for those with little or no experience of the previous situation of the post-war boom. The reference points of individuals are changed, transmuting the old abnormality into a new normality seeming to neutralize, at least partially, the harmfulness of yesterday. This explains why the repeated observations over time of an increasingly pathogenic society did not lead to a form of cannibal society, devouring its own members with an increasingly voracious appetite.

55The dialectical relationship of stress and “adaptation” over the passage of time opens up a wide field of sociological investigation. How much adjustment to tension is possible? Can we adapt to any type of shock? Are individuals all equally adaptable or is there a stratification indicating differences in resilience and coping (Pearlin and Schooler 1978)? What are the sources of adaptability? The plasticity of agents to new social and personal relationships (of which we know little either in terms of scope or mechanisms) and their level of resistance to mental stress, are a new avenue for research that is not easy to follow, but one that is potentially rich for understanding the trio of societies, tensions and individuals. To mitigate these reflections, we must remember that the adjustment period we observed runs over several decades and that it was particularly pathogenic (Figures 2 and 3, Table 8). Then if in relation to suicide unemployment seems less harmful than in the past, the evidence about depression still shows that it has a more erosive effect on the strength of the unemployed than on people in employment (Table 7). Finally, and more prosaically, the people who experienced the post-war boom are also those who are leaving or have already left the labour market.

56The joint study of depression and suicide can overcome the temptation of strict equality: the sociology of mental tensions equals the sociology of inscription of social and deregulationary dysfunction in the privacy of the mind. In fact, some mental illnesses arise because of toxic social situations, such as new management styles (Boltanski and Chiapello 1999), but we should not conclude too readily that all tensions are due to malfunctions and that if they were eliminated we would suddenly achieve a world without tensions. To state this would simply show the naivety of expecting such an improbability. Aneshensel, summarizing Merton (1938), argues that “social orders permitting normal emotional functioning also generate circumstances in which emotional disorder constitutes a normal or predictable response. The occurrence of social stress, therefore, can be seen as an inevitable consequence of social organization” (Aneshensel 1992: 33, our emphasis). Merton pointed out the contradictions generating tensions between strong cultural norms that promote financial success and a social structure allowing only part of the population access to it. Aneshensel goes further, arguing that “the imperatives associated with the maintenance of the social system inevitably create tension between the individual and the collectivity. These systemic sources of tension can be changed from one location in the system to another but cannot be eliminated entirely. Thus, stressful life circumstances and their emotional consequences may be and often are experienced by perfectly ordinary people integrated into the normative structures of society” (ibid.. 33). In fact, the differences in prevalence of mental tensions between men and women, the selectivity of these by gender come also from a “normal” functioning of a society where the regulation of individuals varies according to their socialization and their assigned positions (particularly by gender) and not only because of social dysfunctions. The shift of the regulatory regime of the individual from obedience to autonomy means that “it is necessarily accompanied by a change in how suffering is experienced” (Ehrenberg 2010: 348). More generally, the process of civilization is accompanied by “changes that affect social norms regarding human behaviour and feelings, and therefore the diseases associated with them” (Elias [1988] 2010: 115). The sociology of mental tensions not only reveals that any system of restraint and authorization contains pathological aspects, but, more simply, that any regulatory system (values and norms) necessarily produces toxic effects. Societies, even in their logic of harmonious functioning, produce tensions and mental illnesses that are specific to them.

57In a social dynamic where conjugal unions are rarer, less restrictive and more volatile, we are led to believe that the likely increase in the group of people without a partner will continue to increase the numbers of various forms of malaise including depression. Marital integration has eroded, especially since the second demographic transition, and in thirty years has contributed to the 15% increase in the prevalence of depression. It is a real paradox that the liberation of the couple, once seen as reactionary, and the extension of the autonomy of individuals has resulted in pathologies that reveal their weaknesses. The growing burden of the number of those in depression following the weakening of conjugal integration will, as in the past thirty years, come to affect separated people and single parents. But we should be wary of univocal interpretations and hasty judgments. Entering into adult life without a partner is also a period of learning about relationships with others, contributing to the construction of self and to refinement in the choices and personality of individuals. In addition, unions are far from being always harmonious ones. Conflicts between partners may result in alcoholism in men and depression in women (Horwitz and White 1991). Separation, including for single parents and over and above material difficulties, may be less harmful than marital discord (Simon 2002). The unbinding of conjugal ties and individualization of the subject are therefore at the expense of the pathologies of modernity or, less melodramatically, we should simply note that various societal configurations are all accompanied by their particular forms of mental tension.


Appendix 1. Statistical Blinder-Oaxaca decomposition applied to logits

58The method proposed by Fairlie (2005) has several steps. An indicator identifying the 2002 survey is introduced to capture the specific effects of this year. The logistic model is estimated on stacked data from “Health” surveys in 1970 and 2002. The numbers of the two surveys are equalized on the smaller sample. The numbers in the larger initial sample are randomly drawn a number of times (here 25) to avoid artifacts due to the deformation of the sample deriving from the truncation. A chain of calculations is carried out at each iteration. The coefficients of the parameters estimated using the logistic model are those from these two stacked surveys. They correspond to the impact of each category of the model’s variables on depression. The contributions of each independent variable to the difference between 2002 and 1970 are then calculated based on these common factors and independent variables for each survey. Calculations of individual probabilities of being depressed then vary according to the composition of the single population at each survey date. The results are thus rid of period effects. The formalization of the decomposition of an equation:Y = F(X figure im12) is written as follows (Fairlie 2005):


61F represents the cumulative function of the logistic distribution;

62figure im14 represents the difference between the risk of being depressive between 1970 and 2002;

63N70 and N02 represents the numbers from 1970 and 2002;

64X70 and X02 represents the independent variable vectors in 1970 and 2002;

65figure im15 and figure im16represents the vectors of the estimated coefficients of the independent variables in 1970 and 2002.

66The first term in brackets represents the portion of the difference between two dates which is due to differences in the actual distributions of X (structure) and the second part due to differences in the processes that determine the levels of Y. This latter member also captures the effects of unobserved and unobservable variables.

67The contributions of each independent variable to the difference between 2002 and 1970 are then calculated. Identifying the contribution of differences in specific variables to the difference in depression between two dates is not straightforward. The contribution of each variable to the difference of depressiveness between two dates is equal to the change in the average predicted probability of replacing the 1970 distribution with the 2002 distribution of this variable while holding the distributions of other variables constant. The contributions of each variable to changes in depression between the periods are estimated successively and in a sliding manner. Thus, the contribution of X1 to the difference between the two dates is expressed by:

69The contribution of X2 to the difference is obtained by:

71Where: N70 and N02 represent the numbers in 1970 and 2002;

72?* corresponds to the constant;

73figure im19 and figure im20 represent the same independent variable in 1970 and 2002 for individual i;

74figure im21 andt figure im22 represent the estimated coefficients for the stacked samples of 1970 and 2002 for the first and second independent variables.

Appendix 2. Risk of being depressive, logistic models

tableau im23
1970 Survey 2002 Survey Constant - 4.3*** - 4.3*** Sex 0.3 1.5*** Men Women Men Women Single 1.2 ** 0.2 - 1.1 *** 0.8 *** Widow(er) 1.1 - - 0.1 - 1.5 *** 0.7 *** Separated - 12.2 - 1.0 - 1.2 *** 1.1 *** Single parent 65 or over - 10.9 - 0.1 - 1.6 - - 0.1 - Single parent 65 or under - 12.3 - 1.0 *** 1.3 *** 0.8 *** Couple with child(ren) Ref. - Ref. - Ref. - Ref. - Couple without child(ren) 0.0 - 0.1 - 0.4 * 0.2 - Other - 0.2 - - 0.6 - 0.2 - - 0.4 - 18-29 years old - 2.2 ** - 0.5 - - 1.4 *** - 1.2 *** 30-39 years old - 0.1 - 0.1 - - 0.1 - - 0.5 *** 40-49 years old Ref. - Ref. - Ref. - Ref. - 50-59 years old - 0.2 - - 0.1 - - 0.1 - 0.0 - 60-69 years old - 2.4 *** - 0.7 * - 0.5 - - 0.3 * 70 years old and over - 3.1 ** - 1.1 ** - 1.0 ** - 0.7 *** Farmer 0.0 - 0.8 - - 0.8 - - 0.4 - Business owner - 0.8 - 0.6 - - 0.1 - - 0.7 * Senior manager - 0.9 - - 0.8 - - 0.1 - - 0.7 *** Manager - 0.4 - 0.1 - - 0.2 - - 0.4 ** Employee Ref. - Ref. - Ref. - Ref. - Manual worker - 0.4 - - 0.6 - 0.0 - - 0.2 - Unemployed - 12.3 - 0.5 - 1.0 *** 0.5 *** Retired 1.0 - 0.4 - 0.4 - 0.3 * Other inactive 1.6 ** 0.2 - 1.9 *** 0.4 *** None/Not declared - 0.4 - 0.0 - 0.2 - - 0.2 ** CEP, BEPC, CAP, BEP Ref. - Ref. - Ref. - Ref. - Ed. level baccalauréat or higher 0.4 - 0.5 - - 0.1 - - 0.1 -
Note: *** p < 1 %; ** p < 5 %; * p < 10 %
A logistic regression model for each survey (interacted with gender variables).
Field: Reference persons and partners aged at least 18.
Source: “Health” (Santé) surveys 1970 and 2002

Appendix 3. Risk of being depressive, logistic models and decompositions, 1970 and 2002

Note: *** p < 1 %; ** p < 5 %; * p < 10 %
This table presents the risk of being depressed (logit) calculated for each sex by stacking the samples of 1970 and 2002, and the contributions of each category to the rate of depressives and the relative proportion they have in the contribution to the difference between the rate of depression in 2002 and 1970 are calculated.
Field: Reference persons and partners aged at least 18.
tableau im24
Men Women Rate of depressives in 2002 2.02 5.62 Rate of depressives in 1970 0.65 1.82 Difference 1.37 3.80 Logit Contribution Logit Contribution Net Relative (%) Net Relative (%) Constant - 0.89 *** - - - 1.14 *** - - Single 1.03 *** 0.07 5.4 0.70 *** 0.09 2.3 Widow(er) 1.38 *** 0.01 0.9 0.60 *** 0.07 1.8 Separated 1.10 *** 0.10 7.3 1.11 *** 0.29 7.7 Single parent 65 or over 1.19 - - 0.01 - 0.4 - 0.10 - 0.00 0.0 Single parent 65 or under 1.14 *** 0.00 0.1 0.81 *** 0.21 5.4 Couple with child(ren) Ref. - - - Ref. - - - Couple without child(ren) 0.28 * 0.11 8.0 0.14 - 0.08 2.0 Other 0.11 - - 0.01 - 0.9 - 0.46 * 0.12 3.0 18-29 years old - 1.51 *** - 0.17 - 12.4 - 1.05 *** - 0.17 - 4.6 30-39 years old 0.12 - 0.00 - 0.3 - 0.38 *** 0.00 0.0 40-49 years old Ref. - - - Ref. - - - 50-59 years old - 0.14 - - 0.02 - 1.2 0.00 - 0.00 0.0 60-69 years old - 0.84 *** 0.00 0.1 - 0.39 ** 0.02 0.6 70 years old and more - 1.33 *** - 0.15 - 11.2 - 0.78 *** - 0.32 - 8.3 Farmer - 0.39 - 0.04 2.9 - 0.16 - 0.00 0.0 Business owner - 0.30 - 0.02 1.5 - 0.34 - 0.02 0.5 Senior manager - 0.23 - - 0.01 - 0.4 - 0.66 *** - 0.06 - 1.5 Manager moyen - 0.24 - 0.01 0.6 - 0.34 ** - 0.05 - 1.2 Employee Ref. - - - Ref. - - - Manual worker - 0.14 - 0.06 4.0 - 0.26 - 0.05 1.3 Unemployed 0.91 *** 0.11 8.0 0.44 *** 0.15 3.8 Retired 0.56 - 0.05 3.8 0.38 *** 0.28 7.2 Other inactive 1.81 *** 0.30 21.6 0.38 *** - 0.61 - 16.0 None/Not declared 0.06 - - 0.02 - 1.2 - 0.18 ** 0.20 5.2 CEP, BEPC, CAP, BEP Ref. - - - Ref. - - - Ed. level baccalauréat or higher - 0.07 - - 0.02 - 1.8 - 0.14 - - 0.17 - 4.5 Sum of contributions - - 0.47 34.4 - - 0.19 4.9
Source: “Health” (Santé) surveys 1970 and 2002, stacked samples

Appendix 4. Duke Depression score

75The Duke depression score is part of a survey to determine the health profile of people using seventeen questions of which five issues concern depression. The items that make up the depressive component are:

  1. I give up too easily
  2. I have difficulty concentrating
  3. (During the past week how much trouble have you had with:) Sleeping
  4. (During the past week how much trouble have you had with:) Getting tired easily
  5. (During the past week how much trouble have you had with:) Feeling depressed or sad

76According to item, three answers are proposed: 1: “None” or “No this does not describe me at all” 2: “Some” or “Somewhat describes me,” 3: “A lot” or “Yes, describes me exactly.” Points are awarded (0, 1, 2) and the addition multiplied by 10 is the Duke depression score (Guillemin et al. 1997).


  • [1]
    Cited by Bastide (1965: 27).
  • [2]
    Regulation is a system of norms (Latin norma: square, ruler) that prevails in a given society, varying in part according to the roles endorsed by a given individual (spouse, father or mother, actively employed or unemployed, occupation, etc.). Deregulation is the malfunction of this system.
  • [3]
    According to the International Classification of Diseases (ICD10), the depressed person “shows a lowering of mood, reduction of energy and a decrease in activity. There is an impaired ability to experience pleasure, loss of interest, decreased ability to concentrate, commonly associated with significant fatigue, even after minimal effort. Sleep disturbances and loss of appetite are also generally associated. There is almost always a decrease in self-esteem and self-confidence, and frequently, ideas of guilt or worthlessness, even in mild forms” (WHO 1996).
  • [4]
    The CES-D (Center for Epidemiology Studies-Depression Scale) score was constructed with twenty questions relating to depressed mood, sadness, fatigue, loss of energy, appetite disorders and sleep, feelings of inferiority, or difficulty concentrating. All responses allow positioning those with depressiveness on a scale ranging from 0 to 60. From 17, a depressive symptomatology is suspected; the presence of proven depressive symptoms is expected from 23.
  • [5]
    The survey of health expenditure 1970-71 was carried out by INSEE and CRÉDOC with more than 7,000 households (more than 23,000 individuals), the “Health” 2002-03 survey was collected by INSEE from 17,000 households (approximately 41,000 individuals). The weighting is not available for the 1970 data, so the results presented in this article are not weighted, with the exception of descriptive statistics for the 2002 survey only. The variations with or without weights are low.
  • [6]
    Depressed people may not consume antidepressants (Lamboy, Leon and Guilbert 2007) and vice versa consumers are not all considered clinically depressed (Amar and Balsan 2004).
  • [7]
    Currently, the identification of depression in general population surveys are conducted more or less indirectly through a depressivity scale whose levels are determined using a score calculated from several questions (often 5-20) highlighting the presence of depressive symptoms (e.g., CES-D score or Duke score).
  • [8]
    We rely on Bowlby’s observation that the duration of the affliction of mourning is variable, but exceeds a year on average ([1984] 2008: 133-7).
  • [9]
    Some differences are observed between our indicator and one constructed from a depressivity score (CES-D). The decrease in depression after 70 years, seen from our index, while conversely the CES-D indicates that depressive prevalence rose sharply after age 65 for women and to a lesser extent for men (Cousteaux and Pan Ké Shon 2008, 2010). Only a thorough methodological study could explain these differences. For our part, we note that ageing, even outside of depression, is linked to at least four items of the CES-D score: lack of appetite, concentration, sleep and drive.
  • [10]
    It would be more apt to refer to “widows” as widowhood is 85% feminine.
  • [11]
    Incidentally, it should be emphasized that the rate of suicide among older people, although spectacularly high, receives very little attention, revealing a kind of intensity scale of compassion according to age.

From 1970 to 2002, depression increased in France mainly due to the combined effects of the weakening of marital ties and changes in the social rules governing the interiority of the individual. An increase in the numbers of single people was the expression of this lower marital integration. The economic depression that followed the post-war boom, and social pressures to be a distinct and powerful self, occurred simultaneously, altering the regulation of the subject. By using data on depression and suicide, and combining micro and macro approaches, it is possible to evaluate the effects of the growth of various categories of individuals on depression and to better understand the impact of tensions related to the economic crisis of the 1980s. Examination in terms of the shocks associated with the state of widowhood or unemployment leads to the hypothesis of a model of adaptation to stress, substituting a new normality for the old, rather than continuous progression.


  • depression
  • loneliness
  • suicide
  • plasticity
  • economic depression


  • Amar, E., Balsan, D., 2004. “Les ventes d’antidépresseurs entre 1980 et 2001.” Études et résultats 285.
  • OnlineAneshensel, C.S., 1992. “Social Stress: Theory and Research.” Annual Review of Sociology 18: 15-38.
  • OnlineAneshensel, C.S., Rutter, C.M., Lachenbruch, P.A., 1991. “Social Structure, Stress, and Mental Health: Competing Conceptual and Analytic Models.” American Sociological Review 56(2): 166-78.
  • Anguis, M., Cases, C., Surault, P., 2002. “L’évolution des suicides sur longue période: le rôle des effets d’âge, de date et de génération.” Études et Résultats 185.
  • Aouba, A., Péquignot, F., Camelin, L., Jougla, É., 2011. “Évaluation de la qualité et amélioration de la connaissance des données de mortalité par suicide en France métropolitaine, 2006.” Bulletin Épidémiologique Hebdomadaire 47-48: 497-500.
  • OnlineAseltine, R.H., Kessler, R.C., 1993. “Marital Disruption and Depression in a Community Sample.” Journal of Health and Social Behavior 34(3): 237-51.
  • Bastide, R., 1965. Sociologie des maladies mentales. Paris: Flammarion.
  • Baudelot, C., Establet, R., [1984] 2002. Durkheim et le suicide. Paris: Presses Universitaires de France [6th ed.].
  • Baudelot, C., Gollac, M., Bessières, C., Coutant, I., Godechot, O., Serre, D., Viguier, F., 2003. Travailler pour être heureux? Le bonheur et le travail en France. Paris: Fayard.
  • Beaud, S., Pialoux, M., 2003. Violences urbaines, violence sociale. Genèse des nouvelles classes dangereuses. Paris: Fayard.
  • Beck, F., Guilbert, P., Gautier, A., 2007. Baromètre santé 2005, Attitudes et comportements de santé. Saint-Denis: INPES:
  • Besnard, Ph., 1987. L’anomie: ses usages et ses fonctions dans la discipline sociologique depuis Durkheim. Paris: Presses Universitaires de France.
  • Bibring, E., 1953. “The Mechanisms of Depression.” In Affective Disorders. Pyschoanalytic Contribution to their Study. P. Greenacre, ed., New York, NY: International Universities Press: 13-48.
  • Boltanski, L., Chiapello, È., 1999. Le nouvel esprit du capitalisme. Paris: Gallimard.
  • Bourdieu, P., 1994. Raisons pratiques. Paris: Le Seuil.
  • Bowlby, J., [1984] 2008. Attachement et perte. Vol.3: La perte, tristesse et dépression. Paris: Presses Universitaires de France.
  • Cassan, F., Mazuy, M., Clanché, F., 2001. “Refaire sa vie de couple est plus fréquent pour les hommes.” Insee Première 797.
  • Chauvel, L., 1998. “La croissance du suicide et les problèmes de la société française après les Trente Glorieuses.” Working Paper for the Secondes Journées Nationales de Prévention du Suicide. Paris: OFCE.
  • Coster, S. de, 2005. “Depression and Law Violation: Gendered Responses to Gendered Stresses.” Sociological Perspectives 48(2): 155-87.
  • OnlineCousteaux, A.-S., Pan Ké Shon, J.-L., 2008. “Le mal-être a-t-il un genre? Suicide, risque suicidaire, dépression et dépendance alcoolique.” Revue Française de Sociologie 49(1): 53-92.
  • OnlineCousteaux, A.-S., Pan Ké Shon, J.-L., 2010. “Is Ill-Being Gendered? Suicide, Risk for Suicide, Depression and Alcohol Dependence.” Revue Française de Sociologie 51(Supplement): 3-40.
  • Dejours, C., Bègue, F., 2009. Suicide et travail: que faire? Paris: Presses Universitaires de France.
  • Demazière, D., 1996. “Chômage et dynamiques identitaires.” In L’exclusion, l’état des savoirs, S. Paugam. Paris: La Découverte: 335-43.
  • Du Roy, I., 2009. Orange stressé. Le management par le stress à France-Télécom. Paris: La Découverte.
  • Durkheim, É., [1897] 1997. Le suicide. Étude de sociologie. Paris: Presses Universitaires de France [9th ed.].
  • Ehrenberg, A., 1998. La fatigue d’être soi. Dépression et société. Paris: Odile Jacob.
  • OnlineEhrenberg, A., 2001. “De la névrose à la dépression. Remarques sur quelques changements de l’individualité contemporaine.” Figures de la psychanalyse 4: 25-41.
  • Ehrenberg, A., 2010. La société du malaise. Paris: Odile Jacob.
  • Elias, N., [1939] 1990. La dynamique de l’Occident. Paris: Presses Pocket.
  • Elias, N., [1988] 2010. “Civilisation et psychosomatique” In Au-delà de Freud. Sociologie, psychologie, psychanalyse. Established and presented by M. Joly. Paris: La Découverte: 113-30.
  • OnlineFairlie, R.W., 2005. “An Extension of the Blinder-Oaxaca Decomposition Technique to Logit and Probit Models.” Journal of Economic and Social Measurement 30(4): 305-16.
  • Fokkema, T., Dykstra, P.A., 2002. “Differences in Depression Between Married and Divorced Women in the Netherlands: in Search of an Explanation.” The Netherlands Journal of Social Sciences 38(1): 24-47.
  • Fredén, L., 1983. Aspects psychosociaux de la dépression: en sortir? Bruxelles: Pierre Mardaga.
  • Gaymu, J., Festy, P., Poulain, M., Beets, G., 2008, Future Elderly Living Conditions in Europe. Paris: INED.
  • OnlineGerstel, 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 64(1): 84-101.
  • OnlineGove, W.R, 1972. “The Relationship between Sex Roles, Marital Status, and Mental Illness.” Social Forces 51(1): 34-44.
  • OnlineGove, W.R, Tudor J.F., 1973. “Adult Sex Roles and Mental Illness.” American Journal of Sociology 78(4): 812-35.
  • Guillemin, F., Paul-Dauphin, A., Virion, J.M., Bouchet, C., Briancon, S., 1997. “Le profil de santé de Duke: un instrument générique de mesure de qualité de vie liée à la santé.” Santé Publique 1: 35-44.
  • Halbwachs, M., [1930] 2002. Les causes du suicide. Paris: Presses Universitaires de France [22th ed.].
  • Halbwachs M., 1947. “L’expression des émotions et la société” In Échanges sociologiques. M. Halbwachs, Paris, Centre de Documentation Universitaire:
  • OnlineHawton, K., Heeringen, K. van, 2009. “Suicide.” The Lancet 373(9672): 1372-81.
  • OnlineHorwitz, A.V., White, H.R., 1991. “Becoming Married, Depression and Alcool Problems Among Young Adults.” Journal of Health and Social Behavior 32(3): 221-37.
  • OnlineHorwitz, 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 60(2): 505-14.
  • OnlineHughes, M., Gove, W.R., 1981. “Living Alone, Social Integration, and Mental Health.” American Journal of Sociology 87(1): 48-74.
  • INSEE, 2012. “L’économie française.” INSEE Références.
  • Jacobson, E., 1971. Depression. New York, ny: International Press.
  • Janet, P., 1923-24. “La tension psychologique et ses oscillations.” In Traité de psychologie, G. Dumas. Paris: Alcan: 919-52.
  • Janet, P., 1929. L’évolution psychologique de la personnalité. Paris: A. Chahine.
  • Jong-Gierveld, J. de, Tilburg, T.G. van, 1987. “The Partner as Source of Social Support in Problem and Non-Problem Situations.” Journal of Social Behaviour and Personality 2(2): 191-200.
  • Karasek, R., Theorell, T., 1990. Healthy Work: Stress, Productivity, and the Reconstruction of Working Life. New York, NY: Basic Books.
  • Kaufmann, J.-C., 1999. La femme seule et le prince charmant: enquête sur la vie en solo. Paris: Nathan.
  • OnlineLamboy, B., Léon, C., Guilbert, P., 2007. “Troubles dépressifs et recours aux soins dans la population française à partir des données du Baromètre Santé 2005.” Revue d’Épidémiologie et de Santé Publique, 55(3): 222-7.
  • Le Pape, A., Lecomte, T., 1999. “Prévalence et prise en charge médicale de la dépression en 1996-1997.” Questions d’Économie de la Santé 21.
  • Leroux, I., Morin, T., 2006. “Facteurs de risque des épisodes dépressifs en population générale.” Études et Résultats 545.
  • Lesthaeghe, R., 1995. “The Second Demographic Transition in Western Countries: an Interpretation.” In Gender and Family Change in Industrialized Countries. K.O. Mason, A.-M. Jensen, eds. Oxford: Clarendon Press, New York, NY: Oxford University Press: 17-62.
  • Maruani, M. ed., 2005. Femmes, genre et sociétés: l’état des savoirs. Paris: La Découverte.
  • OnlineMerton, R.K., 1938. “Social Structure and Anomie.” American Sociological Review 3(5): 672-82.
  • OCDE, 2012. Santé mentale et emploi. Mal-être au travail? Mythes et réalités sur la santé mentale et l’emploi. Paris: OCDE.
  • Pan Ké Shon, J.-L., 2002. “Être seul.” In Données sociales. La société française 2002-2003. Paris: INSEE: 587-94.
  • Pearlin, L.I., Schooler, C., 1978. “The Structure of Coping.” Journal of Health and Social Behavior 19(1): 2-21.
  • OnlineRushing, W.A., 1979. “Marital Status and Mental Disorder: Evidence in Favor of a Behavioral Model.” Social Forces 58(2): 540-56.
  • OnlineSapinho, D., Chan-Chee, C., Briffault, X., Guignard, R., Beck, F., 2008., “Mesure de l’épisode dépressif majeur en population générale: apports et limites des outils.” Bulletin Épidémiologique Hebdomadaire 35-36: 313-7.
  • Sermet, C., Cambois, E., 2002. “Mesurer l’état de santé.” In Démographie: analyse et synthèse. Vol.3: Les déterminants de la mortalité. G. Caselli, J. Vallin, G. Wunsch, eds. Paris: INED: 25-52.
  • Sher, L., Oquendo, M.A., Mann, J.J., 2001. “Risk of Suicide in Mood Disorders.” Clinical Neuroscience Research 1(5): 337-44.
  • OnlineSimon, R.W., 2002. “Revisiting the Relationships Among Gender, Marital Status, and Mental Health.” American Journal of Sociology 107(4): 1065-96.
  • OnlineSingly, F. de, 2007. Sociologie de la famille contemporaine. Paris: Armand Colin [3rd ed.].
  • Singly, F. de, 2011. Séparée. Vivre l’expérience de la rupture. Paris: Armand Colin.
  • Tocqueville, A., [1840] 1951. De la démocratie en Amérique. Tome 2. Paris: Éditions M.-Th. Génin/Librairie de Médicis.
  • Tousignant, M., 1992. Les origines sociales et culturelles des troubles psychologiques. Paris: Presses Universitaires de France.
  • Turner, R.J., Turner, J.B., 1999. “Social Integration and Support.” In Handbook of the Sociology of Mental Health. C.S. Aneshensel, J.C. Phelan, eds. New York, NY: Kluwer Academic-Plenum Publishers: 301-20.
  • 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 33(1): 10-24.
  • OnlineVallin, J., Meslé, F., 1998. “Comment suivre l’évolution de la mortalité par cause malgré les discontinuités de la statistique? Le cas de la France de 1925 à 1993.” In Enjeux des classifications internationales en santé. G. Pavillon, ed. Paris: Inserm: 113-56.
  • Van de Velde, C., 2004. Devenir adulte. Sociologie comparée de la jeunesse en Europe. PhD in Sociology. Paris: Institut d’Études Politiques.
  • Waldron, I., Hughes, M.E., Brooks, T.L., 1996. “Marriage Protection and Marriage Selection. Prospective Evidence for Reciprocal Effects of Marital Status and Health.” Social Science & Medicine 43(1): 113-23.
  • OnlineWorld Health Organization–who, 1996. Classification statistique internationale des maladies et des problèmes de santé connexes. Geneva: World Health Organization.
Jean-Louis Pan Ké Shon
60, rue Étienne Dolet
92240 Malakoff
Géraldine Duthé
Institut National d’Études Démographiques (INED)
133, boulevard Davout
75980 Paris cedex 20
Translated by
Peter Hamilton
Uploaded on on 21/01/2015
Distribution électronique pour Presses de Sciences Po © Presses de Sciences Po. Tous droits réservés pour tous pays. Il est interdit, sauf accord préalable et écrit de l’éditeur, de reproduire (notamment par photocopie) partiellement ou totalement le présent article, de le stocker dans une banque de données ou de le communiquer au public sous quelque forme et de quelque manière que ce soit.
Loading... Please wait