1Since the beginning, gender studies have emphasized the analysis of social categories and representations. They have underscored their gendered dimension, as with the professional qualifications described by Margaret Maruani and Chantal Nicole (1989), where the historically constructed professionalism of jobs held by men is contrasted with the persistent lack of recognized qualifications for so-called “women’s” work. Social construction, statistical representation : the tools and categories used by statisticians are not neutral in this respect. In the tables of INSEE (Institut national de la statistique et des études économiques – National Institute of Statistics and Economic Studies), the precise description of male jobs contrasts with the vague outline of female activity (Amossé 2004). Statistics do not merely reflectg reality, but contribute to its creation (Desrosières 2008). Regarding gender categories, the analysis of social representations as well as statistical publications has highlighted the concept of invisibility. In “La statistique saisie par le genre,” Annie Fouquet (2003) proposes a two-part history : the first, which extends from the immediate postwar period to the beginning of the 1970s, describes women’s statistical invisibility, while the second presents an era in which women move towards independence and gender equality. However persuasive and inspiring this narrative is, it does not fully explain more recent developments, early indications of which can be seen as we look back at the 1980s.

2Studies of the concept of a household, carried out over the last ten years by statistics specialists, and which are echoed today in the academic sphere, [1] encourage us to rethink the first two “eras” of this history of gender statistics and to add a third. In the 1960s and 1970s, as Fouquet (2003) noted, women “resisted” analytical models, and it is this resistance which revealed the obsolescence of the statistical categories, in which they were, indeed, poorly represented. This development was chracterized by a logic of emancipation, with the emergence of a statistical individual whose social role was no longer necessarily assigned by his or her position within the household. The recent reconsideration of the statistical household, from defining its boundaries to identifying the connections between its members, (cf. for example de Saint Pol, Deney, and Monso [2004]), is related to another development, where households break up and reform, in law and in fact. There is an apparent change in perspective in the statistical field today, from questioning the perception of inequalities between men and women to redefining their roles within households and, consequently, more broadly in society. The erosion of the statistical household reflects the weakening of an “institution” (according to the meaning defined by Mary Douglas [1999]), which is both a family unit, a unit of income solidarity, and a place to share one’s life. As the concept of household changes, the very framework of our definition of the place of men and women is once again, but differently, called into question.

3In this article, we therefore propose a re-interpretation of the history of statistical tools and categories, through the concepts of the household and the individual, and the shifts in perspective which these concepts have prompted in the study of gender. From a theoretical viewpoint, we follow in the wake of sociologists and historians of statistics, including Alain Desrosières and Laurent Thévenot, who helped develop a way of thinking about statistics that does not limit its understanding to its technical dimension. More specifically, we will use the approach chosen by Monso and Thévenot (2010) for the series of surveys on Formation et Qualification Professionnelle (FQP – Training and Occupational Skills). By examining the history of the survey questionnaires, and the publications which are associated with them, the authors define the key issues which link policies (i.e., both questions and measures), analytical tools (surveys, categories, methods), and explanatory theories (trends in the disciplines of economics, sociology, and political sciences). However, unlike Monso and Thévenot, we will not focus our analysis on a series of specific surveys, but will include a large range of sources, tools and statistical publications (see box). This freedom has a price. The distance permitted by the overview provides less empirical precision. Our article is therefore presented more like the draft of a historical reading put forward for discussion than an actual work by a statistical historian.

4The three “eras” that structure our analysis (and this text) should, moreover, be understood in the sense of sociologists’ eras, rather than historians’. They include overlapping periods, pioneering works, which were sometimes never followed up, and persisting features, once a new “era” has begun. What justifies the time-related terminology is the general trend, which shows successive shifts in focus over time, without any observed flashbacks. These “eras” are marked by dominant issues, in the sense of the term used by Monso and Thévenot (2010), but in a gender perspective. They are characterized by a certain way of addressing the question of women and, more generally, gender categories in statistics, and refer to the different import given to individuals and households at different times.

5Using the pivotal periods mentioned by Fouquet (2003), the first era separates the immediate postwar period, when the household surveys were developed by INSEE, from the era starting at the beginning of the 1970s when, following May 1968, a movement of individual liberation gradually changed the way society regarded itself. From the first to the second era, to say it simply, the focus of statistics moved from households to individuals, from the low professional visibility of women to their active participation in the labor market, from the differences linked to gender roles in the household to a goal of emancipation and equality between men and women. The emergence of the third era coincides with a shift back to the household, but this time with the focus on the individual, and an attempt to redefine household gender roles : the aim is to reconcile different roles (familial, economic, etc.), to understand the way in which different times are interrelated, and the way in which decisions are made. As we shall see, the beginning of this third era does not mean the end of the second and is also more difficult to date : it takes shape around a (re-)questioning of the concept of the household, to adapt to the changes currently underway. In the 1980s, studies of the individualization of the Emploi du temps (Time Use) surveys began at INSEE, [2] and surveys by INED (Institut national d’études démographiques – National Institute of Demographic Studies) tried to clarify changes in family situations. But it is mainly since the end of the 1990s that “form investments” were made in these two institutes to fund studies of the objective and subjective position of individuals within households.

First era : assignment of gender roles within households

6Since the Second World War, the household has been a central concept in public statistics in France. The household is defined as equivalent to housing unit, and has taken on a practical aspect, [3] as it is used as the basic unit in major social surveys (together with documents from the Public Records Office, which also served as a major source for demographers). As Claude Seibel (1989) points out, the importance of the household comes from the United States, where the methods of using sample household surveys originated (by Paul Thionet, Raymond Lévy-Bruhl, and Jacques Desabie). These surveys inspired by this, including the first Emploi survey in 1950, were used by INSEE’s “Population and Households” department. In the history of French statistics, the household is thus much more than a technical concept used to select random population samples. As the title of the collection published in 1989 in honor of Jacques Desabie illustrates (Les ménages, INSEE 1989), [4] the concept of the household symbolized a certain way of producing social statistics which prevailed during the first decades of INSEE (established in 1946). The statistical representation of society at the time was produced using tables based on types of household and major social groups, themselves determined by the socio-professional position of the “heads of household”. The analyses focused on working-class and urban households. It was the golden age of the nomenclature of socio-professional categories (Desrosières and Thévenot 1988).

7Of course, once the sampling had been completed and the interviewer had made initial contact, questions were posed to individuals, rather than households. However, the questionnaires and the published reports emphasized the household and not the individual. Most often individuals were surveyed only as representatives of the household, with roles assigned by statisticians, in line with the social norms of the period. As Monique Meron points out in Le travail du genre (2003), the household is the nexus between the familial, the personal, even the intimate (often reserved for women) and the economic, the public, and the social (reserved for men) spheres. However, far from being understood as a whole, as the melting pot of the most basic social relationships, the household was then analyzed by two distinct groups. Demographers looked at women and saw their role as confined by their reproductive function and their family role. In contrast, sociologists and economists focused only the situation of men because their analysis was focused on employment and work. The history of statistical tools and institutions (between INED and INSEE) is also the history of the interaction between these specific domains.

8One of the most striking examples of this allocation of roles is the Histoire familiale survey which, from its first edition in 1954 to that of 1975, remained focused on non-single women. As Thérèse Locoh, Jacqueline Hecht and Armelle Andro (2003) point out, there has always been an imbalance in studies of fertility that prioritize the mothers’ point of view, as if this were a “female” topic : “There is no requirement, other than a bias which stems from and reveals gender roles, that women should be the only interlocutors in all the major surveys on fertility” (306). The woman as potential wife and mother has long remained a compulsory statistical element. Here women as represented by statistics do not suffer from invisibility, but from a limited and imposed role. By focusing on fertility instead of the female condition, this representation is more “feminocentric” than feminist, to borrow an expression from Locoh, Hecht, and Andro (2003).

9The statistical analysis of women’s work and employment situation is also marked by certain representations. Fouquet (2003) emphasized the problems posed for statisticians by the “boundary” between women’s activity and inactivity. In the population censuseslike the Emploi surveys, for women from rural areas the changes in definition reveal the reluctance to consider their activity of helping on the family farm as a job, beyond its domestic or professional character. In the 1950s, these changes contributed, moreover, to a statistical short-sightedness concerning advances in female employment. From 1954 to 1962, although many women participated in the rural exodus and left the agricultural world, almost as many entered the work force. Thus, behind women’s apparently stable rate of employment a large-scale transformation was happening, which statisticians only identified several years after the fact. The published reports of the time essentially covered up this development whose emergence was as unwelcome as its implications were hard to predict.

10Finally, studying the first decades after the Second World War shows the degree to which the professional situation of women was a source of difficulty and incomprehension in the statistical field. The common categories of work and employment could not describe their situation. Furthermore, the analyses most often differed according to gender : the interest was in men’s work and women’s occupations. And when the tables provided equivalent data for men and for women, it was presented separately, back to back, as Annie Fouquet (2003) notes, not in relation to each other to allow a comparison. Discussions focused on the differences in the “nature” of gender roles, not on the reduction of inequalities (as we will see in our second section).

11In fact, during this time, men and women seemed modeled by statistics, each in their own role. But this logic took on a particular connotation concerning women, to the extent that the statisticians were predominantly men. An anecdote related by Michel Louis Lévy (1989) demonstrates this. He cites the INSEE statisticians’ reluctance to develop surveys on family situations. The scene took place in the 1960s, where “a small entourage of INSEE statisticians was walking around [. . .] with Jacques Desabie,” who asks : “Do you see us asking French women at what age they lost their virginity, or if their first child is indeed their husband’s ?” And then mentions the risk “of generally bringing INSEE into disrepute” (24). Through this account, from which Michel Louis Lévy distances himself—emphasizing that “the extreme caution [of the time] [5] [. . .] did not prevent a certain gradual extension in the questions posed, for example, lo and behold, the fertility of young single women was thenceforth included in the Famille survey,” and expressing the “wish [. . .] that it [INSEE] confirm some recent boldness, beyond the limits respected up to now” (26)—we measure the mood of the times, and how much things have changed in a few decades. This was indeed a world conceived by men for women who, at the time, were considered slightly impure (Bourdieu 1998).

12It was also a time when public statisticians were members of a State nobility associated with the major orientations of postwar reconstruction. The period was devoted to economic and social planning, aimed at revealing and reducing social inequalities, on the one hand, with, on the other hand, economic development having supposedly satisfied the need for labor ; it was a throwback to the issues initially developed in the FQP surveys (Monso and Thévenot 2010). In this context, “one” hesitates between the need for women to have and to care for children, and the chance to address a shortage of labor by employing women. The hesitations between these two demands made on women can be traced in the history of the Emploi du temps surveys. Its precursor (the first survey conducted at INED in 1948), formulated the first part of the alternative. “It [was] about knowing the cost of raising a child for society in working hours” (Fouquet and Charraud 1989, 54). After the Second World War, the demographic deficit and concern with the birth rate formed the background to this survey. An overall balance sheet (in hours) was drawn up for all domestic chores based on the time budgets of married women (the only ones surveyed). Mother and housewife, this was how statistics represented women. Given the scope selected for the survey, the allocation of domestic work between men and women was excluded from any possible analysis.

13Quite simply, it was not a question that was raised at that time.

14Certainly, the domestic role of women was identified as the main obstacle to their entering the workforce. But, there is no possible hesitation in choosing between the two : the choice is for the domestic at the expense of the professional. This is the message which emerges from the tables showing female activity in relation to the number of children they are caring for. The first role of a woman is to be a mother. After that or failing that, their obligation is to participate in the national reconstruction effort. Quoted by Fouquet and Charraud (1989, 55), Jacques Desabie reveals the spirit of the age in a 1952 study : “more than a quarter of the labor force consists of people who we may wish, for social reasons, would not work : [. . .] ten [percent] are young people under twenty, thirteen are over sixty and four [are] employed women caring for children.” Allocations are made by age, gender, and family situation.

15When, at the end of the 1950s, a shortage of workforce occurred (because of the development of the domestic market), the demands on women changed slightly. In 1958, when the new INED Budget temps (Time Budget) survey was carried out, one of the questions asked was : “Will women be able to free themselves from household chores to respond to this call [for labor] ?” (Fouquet and Charraud 1989, 56). As labor needs was growing and the labor force was waiting for the large baby-boom generation to reach working age, women were called upon, at least temporarily, to fill the labor gap. The main issue raised by the possibility of female employment was the need to identify different working arrangements, but not to review the balance of housework allocation, which was in fact impossible to demonstrate since men were not included in the survey. In statistical publications, women still appear to be objects in the hands of male planners. “Why not encourage their employment with part-time jobs ?” asks Alain Girard (1958). The same is true for INSEE and its survey Le travail feminin(Female Work) in 1958, which was meant to examine obstacles to women’s employment. In retrospect, the obstacles seem to be above all in the minds, because the comments stress that two thirds of non-working women under fifty-five reveal an unwillingness to work, no matter what happens, forgetting, along the way, the formidable reserve of labor force which the remaining third constitutes (Fouquet and Charraud 1989).

16As Locoh, Hecht and Andro (2003) point out, demographics is not just about available data. There were “unspoken gender factors” built on stereotypes, unasked questions, and ignored data, etc. Statistics and what is said about them are two different things. This was the case with the surveys Opinion sur la natalité et la population (Opinion on the Birth Rate and Population), conducted from the 1950s through the 1980s, where the small degree of attention given to gender differences, and the similarity of the answers for men and women, does not cease to surprise. Either women were invisible, or it was not possible to identify their specific features. Despite the changes, whose impetus we will try to understand, gender representations have not disappeared. Sometimes they continue to be based on the concept of the household, as a cover for gender differences. Thus, households have an opinion on saving, and thrifty behaviors, independently of the individuals who are questioned in practice, and of their gender (L’Hardy 1989 ; Vangrevelinghe 1989). [6] Admitedly, women did not have the right to vote and to stand as political candidates until 1944, and were not able to open a bank account or enter the workforce without their husband’s permission before 1965. . .

17Despite these situations of resistance, things changed nevertheless. The statistics of the first decades after the Second World War, alongside society itself, were gender blind. Using Fouquet’s work for INSEE (2003), and drawing upon the studies of Locoh, Hecht and Andro for INED (2003), we have been able to show the statistical obstacles that gender studies encountered along the way and the objective complicities which were maintained by the statistical categories and data in this approach. This was the golden age of household surveys, in which the individual was most often relegated to the background. Research was most often conducted in an asexual manner or in the “masculine neuter,” to use Margaret Maruani’s expression (2001). And the concept of household was largely part of this well-regulated representation where each person, depending on gender, sometimes also on age, was assigned a role according to the needs of economy or society. As we will see, the 1970s marked a turning point, with the gradual emergence of the statistical individual in France.

Second era : demonstration of gender inequalities

18The increase of women in the work place, finally identified in the mid-1960s, the events of May 1968, and the development of feminist movements in the following decade brought statistics into a new era. The legal differences between men and women were fought against and their actual differences were presented as inequalities to be purged.

19Of course, for technical reasons, the household remained the main starting point for statistical population surveys. But, as an analytical category, it gave way to the individual. We will return to the many factors that contributed to this development later. As Laurent Toulemon shows in this issue, the household composition has undergone profound changes since the 1970s : the law has changed, as have actual situations, with the increase in divorce, separations, and blended families. During the same period the birth rate has fallen sharply, which Gérard Calot (1989) presents in Les Ménages as the conquest of a new frontier of individual liberty. As a consequence of these major changes, which affected both the number and the structure of households, the individual easily became the preferred analytical unit. The social sciences, to which statisticians have become more open (Seibel 1989), have also affirmed the status of the individual, by increasingly referring to the paradigm of methodological individualism, whether in microeconomics or actor-centered sociology, to mention the most fashionable disciplinary trends within statistical institutes from the 1980s onwards. Finally, the increase in processing power and the technological revolution prompted by the introduction of micro-computing at the beginning of the 1990s made it possible both to gather complex individual data and to process this data. From then on a single person no longer summarized the household, which, in fact, faded into the background in favor of the individuals constituting it. Gradually, the surveys directed exclusively at the household level,—like the Budgets de famille (Family Budgets), Patrimoine (Household Assets) or Logement (Housing) surveys—began to be challenged because statisticians had become used to taking individuals into account in their reasoning.

20These developments were not neutral from a gender point of view. Analyzing surveys by taking into account the individual level allowed a systematic comparison of men’s and women’s situations. The decline of the social class model was accompanied by the emergence of new analytical categories, including gender, and new issues : according to Monso and Thévenot (2010), the theme of social inequalities (traditionally measured, as we have seen, through the category of “head of household”) notably disappeared in favor of the topic of human capital and discrimination. The input, then, is resolutely individual, with gender being one of the characteristics considered either a benefit or a disadvantage on the job market. In this way, women’s occupations and working situations, which had long been overshadowed by those of men, became clearer. Symmetrically, the role of men in the family became a subject for analysis in its own right. Demographers and economic statisticians followed opposing paths, but converged in their approach to men’s and women’s situation and, on an institutional level, led to an increasingproximity between INED and INSEE. The affirmation of a statistical individual, the homo statisticus, and this individual’s liberation from the household, allowed the disallocation of gender roles, that is to say allowed the individual to become a woman like any other.

21In the demographic field, the development of the Histoires familiales surveys is a good expression of these changes. Conducted from the mid-1950s to the mid-1970s, the first three editions remained limited to non-single women (respectively women from 45 to 54 years of age, then the under-70 age group, and finally those under 65). The first significant change occurred in 1982, when, by extending its field to include single people, the survey responded to the main criticisms made of the 1975 survey (in fact, this survey was not used very much because it did not take into account how family structures and representations associated with the family had changed). The questionnaire doubled in size, from two to four pages, and included questions on the dates when couples met and separated (no longer simply when they married and divorced), on the working life of the women questioned, and the type of childcare used. A sign of the times, the survey particularly emphasized women’s activity. In 1999, another change occurred when men were finally included in the survey. In 1954, the survey questioned women about the number of children they wanted to have, in 1962 and 1975, the focus was on fertility over time, while in 1982 and 1990 attention turned to women’s activities and to childcare respectively. It was not until 1999 that the subject of men’s fertility appeared in the survey. From then on, the questionnaires were completely identical for both genders, and men’s responses were not only addressed in terms of whether or not they were consistent with those of women. [7] This change appears to have come quite late. In fact, it is partly due to the distinction between INSEE and INED operations : the most complicated demographic indicators, which require very broad samples, are taken from INSEE’s post-census surveys (the Histoires familiales surveys), while INED conducted the household surveys. As it happens, men were asked questions about their fertility (past and planned) from the first edition of INED’s Situations familiales (Family Situations) survey in 1986.

22From that point on, these surveys made it possible to make comparisons between men and women, for example a comparison of the stages of coming into adulthood (Toulemon 1994). The convergence of men’s and women’s trajectories was no longer assumed but demonstrated, by comparing them from generation to generation : thus it was established that women have gradually acquired an adult status less dependent on that of their spouse. For demographers, the shift in focus from the “household” to the “individual” is rather remarkable, because many statistics were historically based on individual administrative documents (like registry office documents for birth, marriage and death.) However, the Histoires familiales and Situations familiales demographic surveys have changed over time from a study of households where the roles were gendered and fertility largely a female issue, to a period when men and women answered the same questions and could then be compared and analyzed in terms of inequalities. As Locoh, Hecht and Andro have noted (2003, 314), the development of the use of gender in demography, which coincides with this period, was accompanied by a deconstruction of its normative categories. “Understanding human reproduction, a phenomenon which is directly linked to biology, from the perspective of gender, involved showing to what extent the naturalization of women’s reproductive ability, which remains an insurmountable physiological ability, obscures the social organization of reproduction. [. . .] Integrating men into the analysis, through the study on fertility and men’s reproductive plans, allows us to take seriously the relational and asymmetric nature of gendered identities and issues.”

23The changes with regard to employment are more noticeable in the publications than in the surveys themselves. It is true that the Conditions de travail (Working Conditions) surveys, which began in 1991 to include on service-sector jobs (mostly filled by women), followed this trend, after having been strongly shaped (in 1977 and 1984) by the questionnaire’s orientation towards industrial employment. But the data of the Emploi, FQP, or Recensements de la population (Population Censuses) surveys already enabled analyses of the inequalities between men and women. Of course, the gendered dimension of socio-professional nomenclatures, which had been reaffirmed by default at the time of the 1982 redesign, was still an obstacle to analysis. However, women’s activities began to be examined in detail in statistical publications, and to be more systematically compared to those of men, whether from the viewpoint of access to employment, the recognition of qualifications, wages or working conditions. There was a shift from a presentation of the differences between men and women, to an analysis of inequalities. The demand “Equal pay for equal work” was echoed in INSEE publications, and the wages which women brought into their household were no longer only seen as a secondary income. Women were considered to be fully employed (Fouquet and Charraud 1989). This is precisely one of the problems that resulted from these changes, as they again emphasized the tension between fertility and professional activity—two roles that are difficult to reconcile, as highlighted in the Emploi du temps surveys.

24While the first survey at INSEE in 1967 examined both men and women, very few analyses were produced on gender inequalities. The 1975 survey, on the other hand, was characterized by the introduction of studies emphasizing these inequalities : the imbalance in the allocation of household chores between men and women was explicitly condemned (Rousse and Roy 1981), while, for the first time, an article provided a simultaneous hourly and monetary assessment of domestic work and professional work (Chadeau and Fouquet 1981). This revived an accounting approach already used in the INED survey of 1948 but, a sign of the change in era, counting time was now used to underline the differences between men and women, not to estimate the productive resources which women’s “free” time represented. However, it was not possible to examine the way in which the inequalities appeared at the household level, because only one person (selected at random) was questioned. It was not until the 1985–1986 survey that both spouses were questioned, if the household comprised a couple. It was only with the following survey (carried out in 1998–1999) that this analytical approach was really used (among others by Laurent Lesnard [2002]). In fact, analyzing the combination of different time uses within households belongs more to the third era in our (small) history of statistics.

25The history of the Emploi du temps surveys, all separated by more than a decade, reveals INSEE’s lack of enthusiasm in examining this theme. It was as if it were taken for granted that the differences between men and women were “structural,” that they would and could only change very slowly. The lack of more regular data has undoubtedly not helped awareness on this subject.

26During the first decades after the Second World War, as we have indicated, the statisticians’ perception of society seemed to be as important as the tools they used. Representations changed but resistances remained, as is shown by the article “Les ménages n’auront plus de chef” (Households Will No Longer Have a Head) (Courson 1982), which noted the transformation of the heads of households into “reference persons,” an evolution which in fact corresponds more to a terminological window dressing (certainly symbolic) than to a profound change. [8] The fact that the statisticians and demographers were men or women is not, in this respect, neutral. As Locoh, Hecht and Andro [2003] pointed out, the increase in the number of women among demographers was not unrelated to the consideration of gender themes. With regard to INSEE, opening up managerial positions to women also had an effect on the gender composition of statisticians. At INSEE, at the beginning of the 1980s, women such as Maryse Huet, Monique Meron, Annie Fouquet or Maryse Marpsat were pioneers of the introduction of gender in statistical studies. The presence of Fouquet and Marpsat on the publishing team for the collection Données sociales (Social Data) in 1984 and 1987, moreover, brought about an increased interest in women in these publications. The first of these publications showed the difficulty of accounting properly for the changes in society : “Relationships between the socio-professional category of individuals, which is used to study employment or salaries, and that of households, which is more commonly used to analyze family budgets or housing, for example, have become less simple at a time when women’s paid employment and relative social independence have grown” (INSEE 1984, 3).

27In the 1990s, gender mainstreaming established the inequalities between men and women as a central theme at the European level. The need to integrate the gender dimension in all the statistical tools and output is gradually becoming prevalent at the national level. Finally, ten or fifteen years later, one would hope that statistics have indeed become gender friendly. It is undoubtedly true to a large extent. Of course, regression is always possible, and progress is necessary on specific questions. But we can agree with Fouquet (2003) when she suggests that the work that is now needed on inequalities between men and women is probably more political than directly statistical.

28Nevertheless, we believe that the recent changes observed in the statistical field are not gender-neutral. We think that a third era, which encourages the re-inclusion of individuals into households, and an understanding of how inequalities between men and women are formed, and how the gender roles within households are defined or redefined, seems to have been emerging over the last few years.

Third era : an analysis of gender roles within the household

29This era, which characterizes the current situation in the statistical world, originated in reflections that began as early as the 1980s in the INED. Because they take the family unit as the central subject, demographers, well before other statisticians, began to open the household black box and to examine in more detail the roles of each of its members.

30The 1983–1984 survey on the Formation des couples (Formation of Couples) explicitly took the couple as its subject. Carried out twenty-five years after the Alain Girard survey (Le choix du conjoint – The Choice of a Spouse, 1958), it seeked to analyze the changes in the way in which people choose their partners. Focused on the relationship between men and women in couples, the survey was in step with a pivotal period where both social inequalities and gender inequalities were being studied. In the introduction to the survey, the developers wonder ifmen and women play the same role in choosing one another in all social milieux. At the same time, age differences are analyzed in terms of the power relationship within the couple (Bozon 1990), which presaged more recent lines of questioning. Another series of surveys, which are concerned with the Comportements sexuels des français (Sexual Behavior in France, 1970, 1992, 2006), shows that the relationships between men and women, in this case in their intimate life, has been the subject of demographic analysis for a long time. Changes in the questionnaires reflect the emergence of our third era : while the first survey in the series was mainly interested in the question of contraception (still seen only as a woman’s question, a woman’s problem at the time), the second broadened its scope to include men in order to better understand their behavior (the AIDS epidemic was growing). It was only with the third edition of the survey that the relationships between men and women were explicitly put forward as one of the main objectives : due to the extensive attention paid to female pleasure in the media, the perception of the couple and sexuality might have changed.

31These surveys, as well as those on Situations familiales (1985 and 1994) and Relations familiales et intergénérationnelles (Family and Intergenerational Relationships) (2005, 2009), [9] explore relationships in households both objectively and subjectively. The way of understanding sexuality, the place of fathers and mothers, the way of apprehending one’s family situation are thus new themes which were gradually established in quantitative research in the social sciences. In fact the erosion of marriage and the deconstruction of the family unit were powerful incentives for developing these themes, which evolve differently depending on whether one is a man or a woman. As Olivia Samuel (2009) notes in the Histoire de vie—Construction des identités survey, “women [particularly] express a form of self-affirmation which escapes the social norms of the family” (122). This survey is symbolic of a trend observed more broadly in sociology in the 1990s (Corcuff 1995), which presents individuals as plural beings, a combination of many social roles and facets of identity. Its goal is precisely to understand how these various aspects are related ; how the roles assigned by society are endorsed, asserted or, on the contrary, denied ; particularly those familial roles referring to positions within the household.

32The third era in our analytical grid is less obvious in the Histoires Familiales survey. In 1999, as in 2011, the questionnaires for men and women were identical, but not given to the same households. [10] Thus there was interest in these topics from a macrosocial rather than microsocial perspective, that is, without really taking the couple as an analytical unit (Locoh, Hecht, and Andro 2003). However, the changes can be observed by comparing the way in which the 1990 and 2011 surveys address the question of childcare arrangements. In the first case, it is a question for women, which then becomes “shared” in the second survey, as it is addressed to both men and women in the same way. A new way of considering conciliation emerges, less “cursed” for women since it also applies to men. In the case of men, it undoubtedly refers to professional arrangements, as well as domestic situations. Anyhow, in recent developments in social statistics, conciliation is now as much about the relationship of individuals within households as about the work-life balance. We also find traces of this change in INED’s 2005 Famille et employeurs (Family and Employers) survey.

33For demographers, the use of couples, or more broadly households, as a subject leads to the formulation of questions in terms of power, cooperation, and conflict. This is for example the approach taken for a study of decisions on fertility in West Africa, which are analyzed according to both the characteristics of individuals and those of couples (Andro 2001). Researchers from other countries (Canada) or working on other countries (Africa) instigated work on gender roles, as if the national resistance was easier to remove when it was about others, rather than oneself (Locoh, Hecht, and Andro 2003). With a goal of opening up households, the analyses of countries in the “South” emphasized in particular the importance of considering the household as a system for allocating resources between its members, rather than as a group sharing a system of joint production and consumption. For these countries, this was similar to gender mainstreaming, as was highlighted by European authorities, for example, and by the ILO (International Labor Organization) during the 1990s. This position was popularized by the FAO (Food and Agriculture Organization), [11] which plans to examine the contribution and access of women to household resources from a gender perspective.

34At INSEE, these issues took more time to emerge, or at least to become established. However now they are very visible. In 2010, the thematic module of the European panel on household income and living conditions (EU-SILC, SRCV for France [Statistiques sur les ressources et conditions de vie – Statistics on Resources and Living Conditions]) was concerned with “sharing resources within the household.” It was partly inspired by the module “Decision making in couples” in the latest Emploi du temps survey (2009–2010), which itself was part of the extended questions developed by INED’s Relations familiales et intergénérationnelles survey (2005). Put to a broad sample of couples this module, proposed and developed by Sophie Ponthieux, aims to examine the degree to which resources are pooled between spouses, and to better understand the factors which are at play in the decision-making processes. As the study note presenting the survey points out (INSEE 2009), the aim is to respond to a social demand, notably expressed in the CNIS (Conseil national de l’information statistique – National Council for Statistical Information), [12] and to an expectation of economists and sociologists. The questionnaire was specifically designed to take into account the various analytical and theoretical approaches that are found in the two disciplines : in sociology, those which result from studies on money in couples (especially that of Jan Pahl and Carolyne Vogler [1994]) ; in economics, those initiated by collective decision-making models, which no longer consider the household as only a place for sharing, but also as a place of power, negotiation, and conflict (cf. the article by Olivier Donni and Sophie Ponthieux in this issue). The new information which will be provided by the module on the financial organization of households, on the division of responsibilities and decisions, and on their determinants, [13] could suggest avenues which go beyond the limits of the unitary approach to the household, which currently forms the basis of most studies on individual poverty risks (cf. the article by Danièle Meulders and Sile O’Dorchai in this issue).

35Before relationships between individuals in households became a subject for statistical analysis in their own right, INSEE had studied the growing tension between individual and the household levels. In a more technical and less explicit way than at INED, the question of the relationship between these two levels had been raised as early as the 1980s. An article by Daniel Verger (1984) is a good illustration of this approach. It examines the individual use of appliances within the family, which can reveal the roles occupied by different members of the household : “child king or woman slave ?” he asks (85). Emphasizing the complexity of the phenomena (between use, possession, and power), he suggests a questionnaire to open the household “black box.” This proposal was not taken up, nor that of substantially developing questions on the individual or collective nature of the uses of money within households in the Budget des familles surveys. [14]

36In fact, reflecting the erosion of family units and the emancipation of the individual which was observed in the previous decade, the 1980s saw the convergence of a movement to deconstruct the household as the only relevant unit for analyzing social phenomena, and the rise in methodological individualism. At INSEE, several innovative experiments emerged in this way, as Claude Seibel (1989) recalls : the Lorraine panel aimed to monitor households, “units with changing shapes” (Padieu 1989, 31) ; the 1987 program on the theme of “domestic production” was intended to respond to the “conceptual and observational problems and difficulties [. . . which are posed by . . .] the development of the informal economy, changes in the division of chores between men and women within couples” (Seibel 1989, 16). But these themes were then put on hold to a certain extent, because of the complexity of the technical problems encountered, as well as the lack of theoretical frameworks enabling the analysis of such data.

37It was not until the end of the 1990s, after most surveys had focused on individuals, frequently forgetting that they belonged to households, that the question of the relationship between these two levels was once again, and explicitly, asked at INSEE. it took place in a context in which the concept of household was more broadly questioned. Statistical investigations were extended beyond “ordinary” housing by investigating “collective” households (prisons, as in the 1999 Histoires Familiales survey ; [15] retirement homes and health institutions, as in the 1999 Handicap, Incapacité, Dépendance [Handicap, Disability, Dependency] survey), as well as people “outside households,” such as “the homeless” (in 2001 for INSEE’s national survey). Above all, new reflections aimed to redefine the Tronc Commun des Ménages (TCM – Common Core of Household Survey Questionnaires), which, at the beginning of each survey, helps to identify the profiles of the household questioned, and the relationships between its members. Presented in an internal INSEE note (Guglielmetti et al. 2001), these reflections reconsidered the concept of “household/housing unit,” “which was established at a time when the stable monolocalized nuclear family, sharing the same resources, could be seen as the quasi-general model” (3). Dating back to the 1950s, this model lost its relevance with the increase in the number of divorces, the rise in university studies, changes in the stages towards adulthood, etc. The number of households split between several addresses has risen, as well as the number of couples who do not share the same home. In the end, the relevant analytical unit has been called into question and is less often strictly found within the confines of the household-housing unit.

38The concept of a “living unit” thus emerged, which defines all the people “who share one (or several) usual residence(s) and who, moreover, have common budget, activities, projects,” on the basis of housing [16] but without being bound to it. The use of this concept, applied at INSEE in the TCM since 2004, is part of a European process of harmonizing survey practices. [17] Most importantly, this decision reflects the possible disconnection between housing, the budget, and the family. The ties, especially family ties, between the different members of the household are also specified in the new TCM, which thus enables identification of same-sex couples. Even though the starting point remains the sampling of housing units, and therefore has limited empirical consequences, the change is far from being minor from a conceptual viewpoint. A statistic gradually emerges, in which the basic unit is the individual, whose position may be situated in households which, depending on the questions analyzed, could be defined as housing, family or economic units. The concept of “intimacy” used by the international “Gender and Generation Survey,” within which falls the Relations familiales et intergénérationnelles survey (2005) at INED, further expands the field of possible definitions of the concept of household. The model of the implicit unity and stability of the housing-budget-family trilogy [18] is gradually being replaced by a group of individuals who maintain different kinds of relationships, the permanence of which is no longer assumed. Multi-residency and shared family ties (cf. the article by Laurent Toulemon [1994] in this issue), but also intra- and extra-family monetary transactions, can be examined in a new way. One can measure the extent to which the positions and roles of men and women are no longer necessarily cast in the same way.

39The “reference person,” who has replaced the “head of the household,” is potentially being redefined and, in the future, without priority necessarily being given to a specific gender : for households seen from a budget perspective, this could be the person contributing the most resources ; in terms of housing, this could be the contractual owner of the dwelling ; in terms of the family unit, the need to summarize the household in a distinct individual is open to discussion, with fathers and mothers possibly being analyzed in concert. The Enquête Longitudinale Française depuis l’Enfance (ELFE – French Longitudinal Study since Childhood, 2011) presages this evolution, as the “reference person” in this survey is the mother, without obscuring the father’s position : the different survey questionnaires try to find a balance between the joint and related questioning of the father and the mother, and controlling the duration of survey interviews.

40* * *

41Just after the Second World War and the development in INSEE of household surveys, the way in which statistics represented individuals and households changed significantly. This was not without consequences for the statistical representation of the differences between men and women. After a period of several decades, in which the household masked changes in situations which were gender differentiated by assigning distinct roles to men and women, and therefore reifying them, the statistical field gradually focused on individuals, which allowed gender inequalities to be highlighted and analyzed. In parallel, although later, the questioning of the notion of the household encouraged a reappraisal of the differences between men and women, by reinscribing them within households.

42These changes correspond to a gradual normative shift, rather than real breaks in the history of gender statistics. Of course, they are not without exceptions, some surveys or studies appearing before, or after, their “time.” [19] And the perception of the changes can itself be oblivious to the past. [20] Finally, some themes are still mainly analyzed either at household level or at individual level. However, as we have shown, the three eras presented show the re-orientations which affect a considerable number of social statistics surveys. We believe that they have a broad impact.

43The statistical presentation of how men and women used their time seems to us a proper summation of these eras. At first statistics focused on women and their potential for repopulating postwar France and, if necessary, for providing an additional labor force. Time use surveys then enabled the differences in how domestic chores were shared to be analyzed in terms of inequality. In the most recent era, the aim has no longer been to only describe the differences perceived as natural, nor to reveal inequalities, but to understand how these differences develop, how the roles of men and women within couples are related, and which social norms guide social practices.

44These changes are summarized in the table below where, like Monso and Thévenot (2010), we compare political concerns, the place of the category of sex in statistics, the analytical tools and explanatory theories for each of the eras of our analysis.

Table

Political Concerns, the Category of Sex in Statistics, Analytical Tools, and Explanatory Theories

Era of our Analysis Grid Political Concerns The Category of Sex in Statistics Analytical Tools and Methods Explanatory Theories First Era : When individuals disappear behind the household Demographic, economic and social regulation (birth rate, need for labor, fight against social inequalities) Different questions for men and women, the differences presented as natural Indicators of fertility and domestic activity for women, socio-demographic accounts, CSP [Catégorie Socio-Professionnelle – Occupational Category] of the “head of household” Classic demography ; economics and sociology of work ; sociology of social reproduction (mobility) Second Era : Individual emancipation, omission of the household Fight against discrimination (gender equality), functioning of the labor market (conciliation, childcare) Identical questioning of men and women, differences analyzed as inequalities Tables compared between men and women, regressions with the gender category as an explanatory variable Economics of human capital and discrimination, studies on gender in sociology, demography, and economics Third Era : Individuals re-inscribed in households Relationship between employment, work, and social protection, between the individualization of social policies and changes in family structures, housing, etc. The ties between individuals within households : differences re-examined objectively and subjectively (in terms of power or negotiation) New Common Core Questionnaire for Household Surveys, non-unitary and optimum blending models, multi-level analyses New economics and demography of the family, the sociology economics of identity

Political Concerns, the Category of Sex in Statistics, Analytical Tools, and Explanatory Theories

45Compared to the first two eras, the last one seems a priori less normative as far as the relative position of men and women is concerned : the first, as we have emphasized, presented the position of each as natural, while the second pursued a goal of gender equality. With the third era there are more uncertainties, which recently surfaced in numerous controversies in gender studies (on care, as in this issue, and recently, over pension increases for fathers of three children.) If the political goals are less clear, the scientific discussions are lively. New questions are emerging, for example concerning the relationship between family ties and economic solidarity, which are raised by the individualization of public policies (such as the RSA – Revenu de Solidarité Active [Active Solidarity Income] and, perhaps in the near future, income tax), between the stability of couples and the trajectory of individuals, around gay parenting, and so on. Here, the relative roles and positions of men and women are widely re-examined.

46As these questions have developed, the second era continues to actively uncover gender inequalities through statistics, whether in recently studied areas (such as violence), or in others that are well known but where inequalities persist (the allocation of domestic chores, the level of salaries and wages, access to management positions in business and politics.) According to Fouquet (2003) there is no longer much to be done statistically on this subject, except to continue the momentum of previous decades : between State feminism and women’s mobilization, there is mainly some political work to be done, based on the numbers.