Surveying Households in Their Main Residence

3A population census is conducted in hierarchical, interlocking steps. The INSEE, which is in charge of the census, collects information on homes, then households within the homes, then families within the households, then individuals within the families. Each step corresponds to definitions chosen to meet the key objectives and issues of the census, namely to define the legal population of each municipality and eliminate redundancies and double counting. The goal therefore is to identify the place where each individual is counted or recorded (in principle, only once), with the exception of rare cases that are treated identically throughout the census to ensure equal treatment of the municipalities. Some individuals are left out or “counted separately.”

4For the purposes of the census, “home” means either main residences (for households), or vacant homes (secondary residences, empty homes), or communities (staff housing, university campuses or dormitories, nursing homes, long-stay hospitals, religious communities, halfway houses, prisons, barracks). People living in non-permanent mobile homes, sailors, and the homeless are not counted in statistics on households.

5Each main residence is thus home to a household that, by definition, brings together all permanent residents of the home. The concepts of main residence and permanent resident are complex when applied in practice (see below). They aim to place each person in a single household. Thus, some residents in the home are not counted as “permanent residents.”

6Within the household, for the purpose of displaying results, the census then identifies the families (usually one; two at the most). A “family” is defined as a couple (heterosexual, per the census definition), with or without children, or a single person and his or her children, all of whom reside in the home. Children must be unmarried, not living as part of a couple, and must not have children living in the home. Thus a three-generation household (grandparents, parents, children) comprises two families: a couple without children (the grandparents) and a couple with children (the parents). Until 1982, only people under the age of 25 could be considered children. This age limit disappeared with the 1990 census, but sometimes in publications the age of the children is emphasized (under 18 or 25) to avoid situations where, for example, a household consisting of an elderly parent living with a single, 60-year-old, adult child would be recorded as a single-parent family (Daguet 2007; Chardon, Daguet, and Vivas 2009). People who are not part of a couple and who have no children in the household are called “isolated;” a household that is home to a single grandparent, parents, and children comprises one isolated person and one family.

7By assigning one household and one household only to residents, the census is not looking to identify multi-residence situations, where people divide their time between two usual residences. The census uses a series of conventions to determine whether people should be counted in the home. It distinguishes between permanent residents (those who live in this home most of the year)—who are included in the first list on the home’s census sheet and who must fill out an individual form—and other individuals living in the home but not counted there (because, in theory, they are identified elsewhere). Permanent residents in the home include spouses who have a different home for professional reasons but for whom this home is the family residence; students under the age of 18 who reside elsewhere for their studies but are attached to the parental home; students over the age of 18 who reside in the home while pursuing their studies; tenants, au pairs, etc. who live in the home. Children who divide their time between the homes of each of their separated parents are recorded in the home where they live “most of the time.” In the case of an even match, they must be counted in the home where they slept the night before the census. Symmetrically, the following categories of residents—which are listed on separate lists on the home’s census sheet—are excluded from the list of the home’s permanent residents: adult children housed elsewhere during their studies (recorded in the place of their student housing), even if they regularly return to the family home; minors (recorded at their parents’ home) living in the home during their studies; people who live in the home for professional reasons but who return to their family home on weekends; and people who regularly return to the home but who are housed in a facility (e.g. a nursing home) in which they are recorded. These rules, which have been used in censuses since 2004, are amended regularly to take into account international recommendations (UNECE/Eurostat 2006).

8Based on these conventions, what has been observed since 1968 is that the number of persons per household has declined steadily (Table 1) to reach 2.3 in 2006. In 1968, nearly one in five households included either several families or individuals outside the family (“complex households”). In 2006, they numbered no more than 5%, [1] and most households comprised only one family. One-person households were on the rise (33% in 2006, against 20% in 1968). [2]

Table 1

Family Structure of Households in Censuses from 1968 to 2006

Family structure 1968 1975 1982 1990 1999 2006 Man alone (%) 6.4 7.4 8.5 10.1 12.5 13.5 Woman alone (%) 13.8 14.8 16.0 17.1 18.5 19.5 Single-parent family (%) 2.9 3.0 3.6 6.6 7.4 7.7 Couple without children (%) 21.1 22.3 23.3 23.7 24.8 26.1 Couple with child/ren (%) 36.0 36.5 36.1 36.4 31.5 28.0 Complex household (%) 19.8 16.0 12.5 6.1 5.3 5.3 Number of households (in millions) 15,778 17,745 19,589 21,542 23,808 26,069 Average number of people in household 3.06 2.88 2.70 2.57 2.40 2.30

Family Structure of Households in Censuses from 1968 to 2006

The age limit of 25 for children was eliminated in 1990; this resulted in a significant drop in complex households and raise in families with children.
Scope: Mainland France.

9Single-person households comprise women more often than men, but the gap has been decreasing over the past 40 years. In 2006, 59% of unattached individuals were women, against 68% in 1968 (Daguet 2007). This development is concentrated in the 25–50 age bracket. The increased frequency of breakups leads more men to live alone, whereas women live with their children after the breakup, as evidenced by the increase in single-parent families comprising a mother and her children. Single fathers represent only 16% of this type of family unit (see below). Other changes, on the other hand, have contributed to the increasing number of women living alone. After the age of 50, after the children have left the nest, women are more frequently alone than men, who have a greater tendency to become involved again in a live-in relationship following a breakup. At older ages, the decline in mortality contributes to delaying the moment where widows and widowers live alone, but it is still most often women who remain alone after the death of their spouse. Their increasing autonomy, thanks to improvements in their health and financial situation, then allows them to stay longer in their own homes (Daguet and Niel 2010).

10In 2006, complex households no longer represented one in 20 households: households comprising one family and one isolated person (outside the nucleus family, as defined by the census, usually an ascendant or a relative) now represented only 2% of all households, while 2.8% of households comprised several isolated persons (friends, roommates, couples not identified as such in the census); and only 0.5% of households comprised two families. However, the disappearance of the age limit in 1990 contributed to the decrease in complex households in favor of families, with single people over the age of 24 no longer being counted as “isolated” if living with a parent; using the same 1982 age limits for children, the proportion of complex households would have been 10.2% in 2006, as opposed to 5.3% following the new criteria. [3]

11These definitions of households and families do not include the entire population. In 2006, 2.4% of people in France (2.3% women and 2.5% men) were not listed as being part of a household. They were mainly housed by medium or long-stay programs (1.6% women and 1.3% men), with men belonging more frequently to other categories [4] and women less often categorized as “homeless” (Marpsat 1999). Within households, 18% of people were not attached to any family (mostly single people: 12% men, 16% women, who represent 33% of the households, see Table 3); 36% of households included no family [5] (see Table 2).

Table 2

Population Situation in the 2006 Census

Situation at the time of the census Numbers (in thousands) Hommes Femmes Ensemble Familles Ménages Population of families 24,280 24,624 48,904 16,854 16, 732 Population outside families, in households 4,692 6 332 11 024 - 9,338 Household population 28,973 30,956 59,928 16,854 26,070 Population outside of households 733 711 1,443 - - Total population 29,706 31,666 61,372 16,854 26,070 Situation at the time of the census Distribution (in%) Hommes Femmes Ensemble Familles Ménages Population of families 81.7 77.8 79.7 100.0 64.2 Population outside of families, in households 15.8 20.0 18.0 - 35.8 Household population 97.5 97.8 97.6 100.0 100.0 Population outside of households 2.5 2.2 2.4 - - Total population 100.0 100.0 100.0 100.0 100.0

Population Situation in the 2006 Census

The first three colums describe the situation of the residents; the fourth and fifth colums the situation of the reference person in the family and the household (the man in a couple, otherwise the oldest employed person, otherwise the oldest person).

12In the census, families were identified based on the family relationships reported on the housing sheet. Ever since 2004, the question “Are you living in a couple?” has been added to the individual form. It helps identify factual situations, but the rationale used to redefine families still excludes two types of couples, namely same-sex couples and couples who have two homes, when each spouse is counted in a home. Similarly, children are only included in the family if they are recorded in the home.

13With these definitions, we can further specify the marital status of the people living in the households. Nearly half the population (48%) lived as couples, a proportion to which we can add the children of these couples (23%). Overall therefore, 72% of the population lived as part of a family comprising one couple with or without children (Table 3). In 2006, the total household population included 107 women for every 100 men. Men live more often in a nuclear family [6] either as adults or as children. [7] There are by definition equal numbers of men and women living as couples, but women leave the family nest at an earlier age than men. Further, men are much less likely than women to have a child or a partner living with them while they are still living in their parent’s home.

Table 3

Family Structure of Persons and Households in the Census

Situation in the household Numbers (in thousands) Men Women Combine Families Households Adult in a couple without children 7,102 7,102 14,203 7,102 7,052 Adult in a couple with children 7,543 7,543 15,087 7,543 7,523 Child of a couple 7,418 6,561 13,979 - - Adult of a single-parent family 344 1,865 2,209 2,209 2,157 Child of a single-parent family 1,873 1 553 3 427 - - Person living alone 3,521 5,094 8,615 - 8 615 Outside of family, not living alone 1 172 1 237 2,409 - 723 Combined 28,973 30,956 59,928 16,854 26,070 Situation in the household Distribution (in%) Men Women Combined Families Households Adult in a couple without children 24.5 22.9 23.7 42.1 27.1 Adult in a couple with child/ren 26.0 24.4 25.2 44.8 28.9 Child of a couple 25.6 21.2 23.3 - - Adult of a single-parent family 1.2 6.0 3.7 13.1 8.3 Child of a single-parent family 6.5 5.0 5.7 - - Person living alone 12.2 16.5 14.4 - 33.0 Outside of family, not living alone 4.0 4.0 4.0 - 2.8 Combined 100.0 100.0 100.0 100.0 100.0

Family Structure of Persons and Households in the Census

The first three colums describe the situation of the residents; the fourth and fifth colums the situation of the reference person in the family and the household.

14Single-parent families make up 9.4% of the population of households (8% men and 11% women). The proportion of single-parent families varies greatly depending on the selected approach: 23% of families with children were single-parent families; the proportion of children living in a single-parent family was slightly lower, at 20%, because, on average, single-parent families have fewer children residents in the household than couples with children (Chardon, Daguet, and Vivas 2009); but the proportion of parents living alone with one or more children was 13% (by definition, single-parent families only count one parent, whereas couples count two).

15Women were more often parents of single-parent families than men (6.0% against 1.2%), which translated as 541 mothers for every 100 fathers in these families. For children, it was the opposite, even more so than with the children of couples. There were only 83 daughters for every 100 sons in single-parent families, and this ratio was lower for children living with their fathers (70 daughters for every 100 sons) than for those living with their mothers (85 for every 100); separated fathers rarely lived alone with their children, especially if they were girls, who, on average leave the family nest earlier.

16There remains a blind spot in the census, involving blended families. If a family comprises a couple and children, the census does not clarify the relationship between the parents and the children of the couple (Chaleix and Haag 1999). The census needs to specify, for each child, whether the child belongs to one, or the other, or to both parents, but the census housing sheet does not make this distinction.

Surveying Households and Their Family Relationships

17Household surveys are only conducted with people living as households and therefore exclude, with some exceptions, people in communal living or in special situations (sailors, homeless people, etc.). The interview begins with the filling out of a list of residents in the home, supplemented by questions about family relationships between household members, and specifics on the existence, for each resident, of another regular residence. Between 1999 and 2001, a new “Common Core for Household Surveys” (TCM) was developed by INSEE (Herpin, Toulemon, and Verger, 2001). Ever since 2004, most INSEE surveys have been based on a common “table of residents in a home” that provides information on the other usual homes of the members of the household and sometimes includes questions about “living units” in the housing.

18Only households defined by the census as “ordinary” are thus included in the scope of the surveys conducted among households. [8] This helps limit the number of “unreachable” people or “long-term absences,” and lowers the risk of double counting, but it distorts the way people are recorded, with people with multiple residences only being surveyed in homes that are the principal residence of a household.

19The employment survey is included in the EU LFS (European Labor Force Survey) and meets those guidelines. Starting out with a complete list of the residents of a home, a question asks if they “live in this home on a permanent basis.” In the 2005–2006 employment survey, 1.6% of households included at least one person who did not. In surveys using the TCM, conversely, all the residents, even the occasional ones, are included, if they spend more than 30 days per year in the home. The list of residents in the home specifies, for each person, whether that person lives in the home “all year long or almost,” or, through various modalities, only part of the time. Then a new question asks if that person “also lives elsewhere.” Finally, once the list of residents has been completed, a new question ensures that nobody is left out: “Is there anyone else who usually lives here, even if it is not regularly and that person also lives elsewhere?” All these residents, even the casual ones, are therefore included in the scope of these surveys.

20The census, wherein questionnaires are filled out directly by respondents, limits itself to a list in which the relationship with the reference person is briefly described, whereas other surveys, which are filled out by professional questioners, identify the relationships between household members more comprehensively. In addition to the relationship with the reference person of the household, the employment survey specifically questions the relationship status for each person; whether the person is involved with a member of the household on the list, and if so, which one; and includes the children of each adult on the resident list. The new TCM is even more specific. It first identifies, for each person, the number of the father and mother if they live in the house, then the number of the partner, and the family relationship of each individual with all the others (provided this was not already identified by the questions on the parents and the partner). In addition, the question about romantic relationships includes potential partners that do not reside in the home.

21These questions help to define the family situation of residents in the home and their family relationships. They identify blended families comprising a couple and children, some of whom were born to only one of the partners (Chardon and Vivas 2009). Blended families can be described from the perspective of the adults, who mix different parenting roles within the home—as parents of children born before the union; as parents of children born to the new couple; and finally, as stepparents, when the couple lives with children born before the union. There are equal percentages of blended families that either include children or not (Chardon and Vivas 2009). Compared to the men, the women were more often mothers and less often stepmothers. According to the 2006 SRCV survey, [9] 9% of couples with at least one minor child were blended families; 6.6% of families counted one child belong to the mother, 3.2% one child belonging to the father (with 0.8% of the families including children belonging to each spouse). In addition, women did not always report raising stepchildren who came to live with the couple, whereas men, who more often than women lose touch with their children after a relationship breakup, were more consistently prone to consider that they were raising their stepchildren (Toulemon 2005).

One or More “Usual Residences?”

23Situations of multiple residences, in which adults and children divide their time between several usual residences, account for most of the differences observed from one statistical source to another.

24Living in multiple residences corresponds to a variety of situations that are strongly differentiated by age. This includes children of separated parents who go back and forth between their parents’ homes; young adults who have left the parental home, but return to it frequently; adults who work in a different city from the location of their family home, or who have a home near their workplace; young retirees who split their time between several homes; elderly people who are housed part-time in the homes of their children or in nursing homes and live in their own home the rest of the time; etc. In addition, many transitions are gradual rather than immediate and result from transition processes that take time—whether linked to family (leaving home, couples forming and breaking up), profession (training, fixed-term contracts, trial periods), or residence (retirement planning, buying but not yet occupying a home). These transitions often correspond to temporary situations of living in multiple residences. Thus, more than half of people in a situation of multiple residences at a given date were no longer in that situation one year later (with the exception of children aged 5–15, for whom multi-residence is more lasting) (Toulemon and Pennec 2009).

25Multi-residence situations were common: one home out of 10 (10.1% in the 2006 SRCV survey) included at least one person who also lived elsewhere, but they were poorly identified in the employment survey (in which this applied to only 1.6% of households). At the individual level, 6.3% of the SRCV survey respondents also resided elsewhere than in the home where the survey took place. Taking into account the risk of double counting for people with two usual homes, we can estimate that multi-residence affects a minimum 3.7% of the population. Standing at around 2% in very young children, multi-residence reached 12% for young adults, men or women; it was rare among middle-aged adults (less than 2% at age 40) and became more common again in older adults (4% between ages 55 and 85) (Toulemon and Pennec 2010a).

Life as a Couple and its Boundaries

26This gray area in family situations is echoed in the definition of couples in the census: 951,000 people (or 1.5% of people in households) lived in a household and spontaneously reported being in a couple in the census without being considered as such by INSEE family codification; conversely, 274,000 people (0.4%) were seen as being part of a cohabitating couple whereas they reported not living in a couple [10] (Table 4).

Table 4

Living as a Couple, INSEE Codification and Responses to the Census Survey

“Are you living in a couple?” In thousands In percentages Living arrangements (INSEE) Combined Yes No Combined Men Women In a couple 29,290 29,015 274 99.1 99.1 99.0 Not in a couple, of which: 30,639 951 29,688 3.1 3.4 2.8 - living alone 8,615 278 8,337 3.2 4.1 2.7 - outside of family 2,409 494 1,915 20.5 22.7 18.5 - single-parent adult 2,209 110 2,100 5.0 11.2 3.8 - child 17,406 70 17,336 0.4 0.4 0.4 All households combined 59,928 29,966 29,962 50.0 51.8 48.3 Outside households 1,443 129 1,314 9.0 10.4 7.5 Combined 61,372 30,096 31,276 49.0 50.8 47.4

Living as a Couple, INSEE Codification and Responses to the Census Survey

Scope: Mainland France
Left side: numbers in the thousands. Right side: percentages by sex

27Almost all those labeled by INSEE codification as living in a couple (99.1%) reported living with a partner in the census. Alternatively, many people considered themselves living in a couple without forming a couple in the INSEE family sense, due to coding choices (which exclude couples in which only one member is registered in the home as well as same-sex couples). This is often the case for people viewed as living outside the family in a multi-person household (21%) and, to a lesser extent, for parents in single-parent families (5.0%, 11.2% men and 3.8% women) or living alone according to the census (3.2%, 4.1% men and 2.7% women). It may be noted that men more than women tend to report living in a couple without the partner being counted as part of the home.

28The question “Are you living in a couple?” was introduced in the census in 2004, but in 1999, the Famille (Family) survey linked with the census included a similar question about couples, which added: “Even if your partner lives in another home for professional reasons.” According to the census, one single-parent family out of 10 (10%; 8% women and 25% men) in the census had thus been recoded as a couple (Algava 2005).

29In the SRCV survey, 95.6% of people reporting living in a couple permanently resided in the same home; 2.7% were “partially coresidents” and 1.7% said they never lived together (1.9% men and 1.5% women), even though they answered “yes” to the question “Are you living in a couple?” According to preliminary work on three years of SRCV surveys, most couples who cohabitate partially or not at all do so temporarily (for one survey, but not the survey the previous year nor the forllowing year), which means that they live together before and after the period of partial coresidence or no coresidence. More than a transition phase in the making and unmaking of couples, these periods seem to correspond to a temporary distance between the partners due to constraints (professional, studies related) outside the relationship (Toulemon and Pennec 2010b).

30Periods of life in a couple without cohabitation can be distinguished from “stable romantic relationships,” of which there are many more and which usually correspond to a sexually exclusive relationship but without an explicit commitment as conjugal partners (Duncan and Philips 2010). In the ERFI survey (Étude des relations familiales et intergénérationnelles—Study of Family and Intergenerational Relationships), there were thus 9% of adults aged 18–69 who reported being in such a relationship (Regnier-Loilier, Beaujouan, and Villeneuve-Gokalp 2009), compared to 1% who reported being in a couple with a partner who did not live in the same home (non-coresident couples). The English concept of “living apart together” (LAT) is ambiguous on this point. Originally conceived to describe couples living apart due to external constraints, it also refers to “stable romantic relationships” in which the question of cohabitation as a couple is not clearly defined by the people themselves and which public statistics therefore have a hard time defining and identifying.

31A symmetrical situation, which is also very difficult to identify in statistical surveys, was recently described in the United States, that is, couples living together without being involved (living together apart, LTA) who, having broken up, continue to live together, either because of financial and housing constraints or for family reasons (mostly related to children) (Cross-Barnet, Cherlin, and Burton 2008).

32The PACS (Pacte civil de solidarité—Civil Solidarity Pact) status has enabled same-sex couples to gain legitimacy and social visibility, but the number of same-sex PACS couples (civil unions) remains low; in early 2009, there were 44,000 people in a same-sex civil union (28,000 men and 16,000 women), or 0.15% of the 30 million people living in couples and 6% of all civil unions (Davie 2011). If they actually reflected the reality of the number of same-sex people in cohabitation, these low numbers would raise three observational problems. First, in surveying the general population, very large numbers would be required to observe enough same-sex couples in the sample to be able to turn it into a statistical study. Second, the question “Are you in a couple?” without specifying the identity of the partner can lead to construing “pairs of people as couples” who may not be in a relationship together (e.g. two women whose husbands/partners work far from home). Finally, even when the partner is identified, couples considered to be of the same sex do not all fall into that category, because all it takes is an error on the variable indicating the sex of either partner to create, in a statistical file, a couple wrongly identified as being of the same sex. These errors are very rare, but if the couples involved were also rare, they would lead to a significant proportion, in relative terms, of “false same-sex couples.” It is therefore necessary either to validate the information on the sex of both partners who report being in same-sex couples (as is the case for surveys that return to question again for people whose sex is already known) or, more simply, to specify the type of relationship (the Canadian census, for example, makes the distinction between same-sex couples and different-sex couples in the question about couples), or the gender of the partner. In the Family and Housing survey (Famille et logements) that INSEE conducted among 350,000 adults in early 2011 as part of the census, the question “Your spouse, partner is: 1. A man; 2. A woman,” asked to both members of a couple, helped resolve the ambiguity and made possible, for the first time in France, studies of same-sex couples based on a survey that incorporated questions on family situation among a representative sample of sufficient size (Rault et al. 2011).

The Statistical Gray Area of Children’s Family Situations

33The family situations of children and couples raise similar questions—residence in the home, multi-residence, and relationships with other members of the household—but multi-residences have more impact on the family situation of children than on the family situation of adults. In cases of multi-residence, adults are often alone in one of their homes, whereas children most often have “two families” when they reside in two usual homes.

34For children whose parents are separated, the family environment is no longer limited to the parental couple. First, children can live in shared residence, and these situations develop with rulings establishing joint custody between the parents. Second, in the homes of children of separated parents (or in each home for those children with two residences), a stepparent could be living there also (if the parent who has custody has begun living in another couple after the breakup of the previous couple), as could younger half-brothers and half-sisters, born to the new couple, and “stepbrothers” or “stepsisters” that the step­parent may have brought with him/her.

35During the transition to adulthood, the offspring’s departure from the home is often gradual and young adults often divide their time between several “usual” homes. At age of 20 or so, more than one young person out of 10 reports having two homes. Studies on family relationships show that young people see themselves as more independent than what their parents describe (Attias-Donfut and Segalen 1998) and that this results in a discrepancy of more than six months, depending on whether information on the departure age is provided by the young adults themselves or by their parents (Villeneuve-Gokalp 2005).

36Table 5 first shows the distribution of minors based on their family status in 2004, according to information from the SRCV survey on family relationships within the home and on the possible existence of another home. The ways of life in the other home were estimated by assuming the other homes to be identical on average to the homes in the sample where children live in multi-residence. Children who have two usual homes represent 6.4% of the children described in the 2004 SRCV survey. But for most of these children, their second home is the home of the other parent, which is also included in the scope of the survey. After correcting double counting in the sample of children with two eligible homes (Toulemon and Pennec, 2010a), the percentage of children living in two homes stands at 3.8%, the second home being either a regular household (2.7%) or a home in a collective lodging (1.1%). The resulting multi-residence is due either to the fact that the parental couple has two usual homes, with the whole family going back and forth between homes; to the fact that the parents are separated, with the children going back and forth between the homes of each parent; or to the fact that the child also resides in a different home, without either of his/her parents. Only 0.9% of children under the age of 18 whose parents were living together in 2004 had two usual homes, but the percentage reached 15.7% among children whose parents were separated.

Table 5

Family Situation of Children

Family situation in the surveyed home Numbers in the survey Distribution Percentage living in two homes in the srcv survey in the srcv survey Combined Collective lodging Regular household Two parents, one home 4,729 79.8 81.8 0.5 0.5 0.0 Two parents together, two homes 87 1.1 32.6 5.5 27.1 Single-parent family, with the mother 680 10.1 10.8 7.1 1.6 5.5 Blended family, mother and stepfather 279 4.4 4.1 11.3 4.1 7.2 Single-parent family, with the father 146 1.6 1.3 53.1 3.8 49.3 Blended family, father and stepmother 149 1.9 1.3 41.2 12.5 28.7 Children not living with either parent 77 1.2 0.7 21.3 2.9 18.4 All children combined 6,147 100 100 3.8 1.1 2.7 including: Children living with only one parent 1,254 17.9 17.5 15.7 3.6 12.2 Children living with their mother 959 14.5 14.9 8.4 2.4 6.0 Children living with their father 295 3.5 2.7 46.6 8.6 38.0 Children living with both parents 4,816 80.9 81.8 0.9 0.5 0.4

Family Situation of Children

Scope: children under the age of 18 living in mainland France.

37The children interviewedin the SRCV survey are distributed based on their family situation in the home where the survey took place. The total of the lines above the line “All children combined” makes up 100%. Beneath that are specific situations, which are not mutually exclusive.

38In 2006, more than four out of five children (81%) lived with both parents; some of them lived in two homes, when the parents themselves had two homes or when they also lived in a home without their parents.

39Children dividing their time between the homes of both parents were counted twice in the SRCV survey, when multi-residence was reported by both parents. In the employment survey, in which the information is not available, those same children were more often counted in the home of their mother. While the proportion of children with separated parents are similar in both surveys, the employment survey appears to underestimate the proportion of children living with their father (2.7% against 3.5%), especially in the case of blended families.

40Children who live with their mothers are less likely to have a stepparent than those who live with their father; but as the vast majority of children with separated parents live with their mother, overall there are more children living with a stepfather than with a stepmother (4.4% against 1.9%). The responses provided by separated fathers and mothers were not entirely consistent regarding the multi-residence of their children According to mothers separated from their former partner (with whom 14.5% of children live), 6% of them live in another regular home, which corresponds to 0.9% of all children dividing their time between the two parental homes, and observed by their mother; according to the fathers (with whom 3.5% children live), 38% of them live in another ordinary home, i.e. 1.3% of all children. In total, it is therefore 2.2% of children who share their time between maternal and paternal homes. Additional data on the time spent in the different homes shows that some separated mothers tend not to report that their children, who spend more than half their time with them, also live with their fathers; unless fathers tend to overestimate the amount of time spent in common residence with their children who spend less than half the time living with them.

41Taking into account the family situation in all homes, it is thus estimated that in 2006 2.2% of children lived part time with their father and part time with their mother; only 2.1% lived exclusively with their father (1.3% with a stepmother and 0.8% in a single-parent family); 13.6% lived only with their mother (4.1% with a stepfather and 9.5% in a single-parent family).

42The increase in the proportion of children living with their father is mainly due to the increased frequency of shared residence, a likely consequence of legal rulings concerning joint custody (Toulemon and Pennec 2010a). The description of the family situation of children with separated parents and of the parents themselves thus makes it increasingly necessary to take into account all homes in which adults and children usually reside. Women live with their partner less often than men. Moreover, they less often report being part of a couple when they have a partner or boyfriend enumerated elsewhere. When they are in a single-parent situation, their children are less likely to be multi-residents, and they are less likely than the fathers to report another residence where their children live. Overall, family situations built across several homes concern men more than they do women.

Taking Multi-Residency into Account

43Taking into account all the homes in which adults and children usually live raises three problems. First, the interlocking structure of homes containing households made up of individuals can no longer be used. Second, the unit of measure is no longer the individual or the family; instead, it is portions of individuals in proportion to the time spent in different situations, or situations that include double counting because of the very way they are constructed. Third, the very nature of households and families becomes blurred if one accepts to take into account the situation of multi-residency.

44In the census, including families in the households, and individuals in the families, helps build partitions. Belonging to a family is an equivalence relation, symmetrical and transitive (if A and B belong to the same family, and if the same applies to B and C, then it is also the case for A and C). While an individual can belong to several families or several households, these no longer form partitions of the population, as the membership relation is no longer transitive. This is already the case with siblings, if we include half-sisters, half-brothers, stepbrothers and stepsisters (if A is the brother or sister of B and B is the brother or sister of C, A and C are not necessarily siblings).

45The solution of correcting the weighting of individuals so as to increase the probability of including those who live in more than one home in a sample (the distribution of weight rationale) amounts to equating a person living in two homes with two separate half-people. This helps avoid double counting but does not provide an accurate description of family situations.

46A more specific description is required to construct specific categories, as we have done for children. The situation of multi-resident people and their families then becomes difficult to define: what is a person living alone? Should we include people who are alone in a home only part of the time and live with other people (in either the same home or another one) the rest of the time? On what timescale do we measure multi-residency?

47The last problem we face is that the units grouping individuals together then become blurred. Measures as seemingly simple as the number of households with one person, or the number of children in a family, must be redefined to specify how partial members of these groups are included or not.

48These questions are not merely theoretical. For example, the measurement of poverty is based on the measurement of living standards, that is, the relationship between all the resources contributed by the members of the household and the number of consumption units in the household. But what weight should be given to people who only live in the home part of the time? How can we evaluate how much of their resources or consumption applies to the household? The article by Danièle Meulders and Sile O Dorchai on this issue sheds light on the effect of the choices made. Depending on the questions at hand, double counting will be accepted or not. A child needs a bed in each of his/her homes, and when counting families according to their size, children who are present part of the time are generally accepted. Conversely, a person divides his/her resources between the expenses of his/her various homes and, when taking into account the household’s resources, the same income is not counted twice. The multi-residency of the children also defines the situation of the parents, in a way that differs between fathers and mothers; half of all children who live with their fathers in a single-parent family also live with their mothers and the presence of a step-mother is associated with a decline in multi-residency. Multi-residence thus emerges as a way to allow the coexistence of fathers and children without undermining the special bond between mothers and their children.

49* * *

50As shown by the above examples, all of which came from surveys conducted by INSEE, the statistical system is gradually adapting to changes in family models. For example, unmarried couples are now well identified in the census and in surveys. But this adaptation is gradual: the question “Are you in a couple?” wasn’t introduced in the employment survey until 1990 and in the census until 2004.

51New family behaviors (same-sex couples, situations of multi-residency) pose new challenges to official statistics in France as elsewhere (Freguja and Valente 2010), and the mining of the wealth of data from the new TCM surveys is just starting. The accumulation of surveys and the use of the “stacked” files that include data from several surveys will help describe more precisely situations that are rare today but must be understood for accurate and precise measurement of the specific family and residential situations of the residents.

52The next Family and Housing survey, conducted as part of the 2011 census, which contains questions similar to those of the TMC surveys, should help in understanding how these new situations are approached in the census and in proposing new ways to address them. On the one hand, certain situations are invisible when it comes to the census (blended families), or poorly understood (same-sex couples with or without children); on the other hand, seemingly simple situations (single-parent families, single people) are only imperfectly recorded.

53The residential family situations of women are more diverse than those of men, with the latter living more frequently in a couple or as the child of a couple. But the Family and Housing survey, conducted among men and women, will help test the hypothesis by which, while men living in a couple are more often multi-residents for professional reasons than women (Bessière and Laferrère 2004), multi-residency among adults and children affects the family situation of men more than the family situation of women when neither are living with a partner.