HOUSEHOLD AND FAMILY CHARACTERISTICS

The UNECE Recommendations for the 2000 round of censuses recognised major changes in the structure of households and families compared to the 1990 recommendations. One important change was that references to consensual unions were included systematically in the new 2000 recommendations. Other important revisions concerned de jure/de facto place of residence, the distinction between private and institutional households, the concept of child, and the concept of reconstituted family.

The significant changes in living arrangements and the emergence of new household types reported in many countries in the UNECE region seen in the previous decade have continued since the 2000 round. Trends have included, for instance, a later start of family life, increased cohabitation, larger numbers of one person households and lone-parent families as a result of divorce, more reconstituted families, and increased proportions of people living in more than one household.

The 2010 CES Recommendations noted that household and family composition can be examined from several different points of view. In considering topics related to households it is important that countries are aware of the different concepts relating to households and families. Some of the key issues (but by no means all) were specifically investigated in the UNECE survey, and this chapter reviews the practices in the 2010 census round regarding the adoption of a number of these concepts, definitions and classifications associated with the structure and characteristics of households (both private and institutional) and families³³.

Concepts and definitions of terms Private and institutional households

The CES Recommendations offers two different concepts for defining private households. The so-called housekeeping concept defines (at paragraph 479) a household as being either:

“…. a one-person household, that is a person who lives alone in a separate housing unit or who occupies, as a lodger, a separate room (or rooms) of a housing unit but does not join with any of the other occupants of the housing unit to form part of a multi-person household as defined below; or

a multi-person household, that is a group of two or more persons who combine to occupy the whole or part of a housing unit and to provide themselves with food and possibly other essentials for living. Members of the group may pool their incomes to a greater or lesser extent.”

Countries adopting this concept do not assume that the number of private households is equal to the number of housing units and thus the incidence of households living in ‘shared’ dwellings is possible.

However, some countries are unable to collect data based on the concept of the ‘common housekeeping’ of household members, in particular where they have to rely on register-based information. Many of these countries use the household-dwelling concept which considers all persons living in a housing unit to be members of the same household, such that there is one household per occupied housing unit.

The results of the UNECE survey showed that 36 of the responding countries (71 per cent) reported that they used the ‘housekeeping’ concept, whereas 15 reported using the ‘household-dwelling’ concept, of which only 3 (France, Liechtenstein and Spain) could estimate the number of private households according to the housekeeping concept.

33 The material in this chapter has been taken largely from a paper prepared by Howard Hogan (US Census Bureau) and presented at the Joint UNECE-Eurostat Work Session on Population and Housing Censuses, held in Geneva from 30 September to 3 October 2013 (http://www.unece.org/stats/documents/2013.10.census1.html)

166

The CESR went on define to (at paragraph 484) an institutional household as comprising:

“… persons whose need for shelter and subsistence were being provided by an institution.”

Most countries (90 per cent) broadly adopted this definition. Of those that did not, two reported that the definition used was imposed either by legislation (Germany) or by the register used as the data source (Norway), while the United Kingdom reported using a definition that better reflected the needs of its main data uses and provided for consistency with previous censuses and surveys (though in the bulk of cases the categories of institution recognised were broadly comparable with those recommended by the CES). In the Canadian census certain types of collective dwellings that were considered to be institutions in the CES Recommendations were classified as non-institutional collective dwellings, though the people living there were distinguished from people living in private households.

Indeed, all responding countries distinguished the population living in private households from those living in institutional or other households, although in the United Kingdom there was no attempt to collect information on private households and families living within communal institutions.

In some countries, people live in specialized housing estates in which the occupants live in a semi-independent arrangements but where various care services are provided in a centralized manner.

In 13 countries, the population living in this kind of housing estate was classified as living in a private household, whereas in 17 countries the population was classified as living in an institutional household. In another 14 countries no such housing was reported. Six countries sometimes classify the population either as institutional or as living in private households depending on various criteria.

For example, in Belarus, if meals were provided by Social Protection then the population was considered institutional, but where people pay for the services themselves, they were considered as living in a private household. In Canada, the level of care was used as a criterion. In the UK the proportion of the population with facilities to enable them to cook for themselves was the deciding factor.

In thirteen countries, information on ‘other’ households was collected. Seven countries included the ‘homeless’ population in this extracurricular group, while other population groups also treated separately included sailors and others living in boats or mobile/temporary living quarters (note that a more detailed review of the concept of ‘homelessness’ is reported in Chapter 19). Four countries reported that the population which was temporarily absent (for up to one year) were separately identified. Switzerland categorized persons who do not fit in the private or institutional category as ‘administrative’ households, while Canada identified the population in non-institutional collective dwellings.

Twenty-seven countries reported conducting a pre-census living quarters validation check for the purpose of, among other things, identifying the nature of the collective living quarters or the potential presence of private households living within institutions. Only two of the countries with register-based census (Iceland and Slovenia) reported doing this.

Child

The CESR gives (at paragraph 495) the definition of a ‘child’ as being: “a blood, step- or adopted son or daughter (regardless of age or marital status) who has usual residence in a household of at least one of the parents, and who has no partner or own child(ren) in the same household.

Grandsons and granddaughters who have usual residence in the household of at least one grandparent while there are no parents present may also be included”.

The CESR went on to note that foster children are not considered to be a child within this definition, nor are any (grand)son or (grand)daughter who lives with a spouse, with a registered partner, with a consensual partner, or with one or more own children.

167 All but five countries reported adopting this definition fully. In Israel, however, a child was defined as anyone under the age of 18 living in the household regardless of familial relations, while in the United States, any offspring of head of household was considered as a child regardless of whether or not they have a partner in the same household. In Bulgaria grandchildren as specified above were not included as children.

Couple

The CES Recommendations defines the concept of a ‘couple’ (at paragraph 496) to include:

“… married couples, registered couples, and couples who live in a consensual union. Two persons are understood to be partners in a consensual union when they have usual residence in the same household, are not married to each other, and report to have a marriage-like relationship to each other”.

All but four countries reported adopting this definition fully. Poland did not recognise people in same-sex relationships as couples. In Norway persons in same-sex marriages are included as couples, but not same-sex cohabitants (data are not available). Romania reported that the concept of a

‘registered couple’ did not exist in national legislation at the time of the census. Kyrgyzstan did not respond on this issue.

The nuclear family

The CESR defined (at paragraph 493) a family nucleus as: “… two or more persons who live in the same household and who are related as husband and wife, as cohabiting partners, as a married (or registered) same-sex couple, or as parent and child. Thus a family comprises a couple without children, or a couple with one or more children, or a lone parent with one or more children”.

The family concept as defined above limits relationships between children and adults to direct (first-degree) relationships, that is between parents and children. In some countries, numbers of ‘skip generation’ households, that is households consisting of grandparent(s) and one or more grandchild(ren), but where no parent of those grandchildren is present, are considerable. Therefore, countries may include such skip generation households in their family definition. The CESR went on to note that ‘family nuclei’ are usually identified at the processing stage on the basis of marital status, sex, age, and relationship to the reference member of the household. In the case of multi-family households, however, these data are often not sufficient to provide a reliable basis for allocating persons to particular family nuclei. It is left to countries to decide whether family nuclei in these households should be distinguished by asking the respondent to list the members of each family nucleus in consecutive order, or in some other way.

Four countries (Azerbaijan, Italy, the Russian Federation and the United Kingdom) reported in the survey that they used a different definition of a ‘family’ than the nuclear concept recommended by CESR. But a closer examination of their more detailed responses suggests that in each case the concepts are broadly compatible.

Three-generation household

The CES Recommendations defines the concept of a ‘three-generation household’ (at paragraph 496) to consist of: “… two or more separate family nuclei or one family nucleus and (an)other family member(s), containing at least three generations and where the youngest two generations always constitute one family nucleus”.

The results of the survey showed that data on three-generation households as fully defined can be produced in 35 out of the 46 countries that responded. A number of countries reported specifically why they could not do so. Bulgaria, for example, noted the lack of interest at the national level in data on such households. Liechtenstein reported that a three-generation household could only be identified

168

if a member of the second generation was the head of the household; in the case where the head of the household was in the oldest generation, the grand-children would just be classified relatives of the head of the household. Conversely, the United States could identify three-generation households indirectly by the relationship question only when the head of household was a grandparent. Norway reported that although data could be produced from its registers, the topic was not considered to be a census variable.

Reconstituted family

The CES Recommendations defines the concept of a ‘reconstituted family’ (at paragraph 498) as being: “… a family consisting of a married or cohabiting couple or a married (or registered) same-sex couple, with one or more children, where at least one child is a non-common child, that is the child of only one member of the couple”.

A little more than half of the countries (27) reported that data on reconstituted families can be produced from their census. Among these, 24 countries fully complied with CESR definition, including 5 out of the 9 with register-based censuses. Three countries adopted a definition that deviated slightly, including Canada where the ‘step-family’ concept used was a close approximation.

Also, although the United Kingdom noted that data on stepchildren had been collected, reconstituted families were not produced as standard output of the census.

In reporting the method used to derive information on reconstituted families, 12 out of 27 countries referred the use of a household relationship matrix (including three of the countries with register-based census), and in 3 other countries a partial matrix had been used. Some 12 countries used the information about the relationship to the reference person – but only Norway of the register-based countries did so; 4 of these used it together with relationship matrix or partial matrix information (more details on the way countries collected information on relationship is given below).

A unique method of matching based on birth dates of children was used in Hungary. In the register-based census of Slovenia the required information was also available in registers. In Poland and in Portugal, besides the relationship to the reference person, direct identification on parents was also used to produce data on reconstituted families.

The CESR went on to suggest (at paragraph 539) how reconstituted families might be classified, but an enquiry into the extent that such a classification was adopted in the 2010 round was not covered in the UNECE survey.

Extended family

The CESR suggested (at paragraph 501) that some countries may wish to derive information on

‘extended families’, data which can have certain advantages not only for studying the economic relationships of families as spending units, but also in classifying families from a demographic point of view. The CESR suggested that an ‘extended family’ be defined as: “… a group of two or more persons who live together in the same household and who do not constitute a family nucleus but are related to each other (to a specified degree) through blood, marriage or adoption”.

Thirty-five countries out of 48 reported that extended families can be identified from their census data, though Belgium noted that the process was difficult, while in Latvia and in the United Kingdom data is available from the census but not produced in standard output. Four of the countries, however, reported some deviation from the CES recommend definition. Hungary developed an algorithm to do this, and in the United Stated, such families can, to a large extent, be identified based on data on the relationship to the head of household.

Canada explained that extended family members can be identified as those people who were in the same ‘economic’ family (related to each other) and who were not members of a ‘census’ family (that is they do not constitute a family nucleus as defined above). The economic family concept there

169 requires only that family members be related. By contrast, the census family concept requires that family members be a male or female married spouse, a male or female cohabiting partner, a male or female lone parent, or a child with a parent present. The economic family and its associated classifications and variables were derived according to the responses to the questions on sex, date of birth, marital status, cohabitation status, and relationship to the reference person. In addition, consideration was given to the order in which household members were listed on the questionnaire.

The CESR went on to encourage those countries that derived information on this type of family unit to use the suggested classifications proposed for the non-core topics ‘extended family status’ (see paragraph 532) and ‘type of extended family’ (paragraphs 543-544), but – as with the proposed classification of reconstituted families above - these non-core derived topics were not covered in the UNECE survey.

Collecting information on relationship within the household

The household and family status of persons within private households is primarily based on the information collected on the (core) topic of relationship between household members. The 2010 CESR noted (at paragraph 506) that in previous censuses, the selection of the one reference person in the household to whom all other household members report or designate their relationship was the recommended method for mapping household structures. When the household's reference person is chosen carefully, this method gives accurate information for most household and family types. In certain cases, however, such as in multiple family households, this method will not always give the precise information that is required. Therefore, a more elaborative method - the household relationship matrix approach - has been developed by some countries. This household relationship matrix allows for the collection of all relationships between all household members.

Some countries have good experience with the household relationship matrix method in their censuses. But other countries have noted problems with this approach, due to its complicated character. Therefore, the CESR recommended that countries consider the relationship matrix only as one possible method for mapping household structures.

The majority of countries (31 – almost two thirds) used information on the relationship to a single person — the household reference person (HRP) - to derive household relationship, while 12 countries (four each from those countries adopting a traditional, register-based or combined methodology) collected information on the inter-relationship between each household member using the more complex ‘matrix’ approach. In addition, three countries (Iceland, Lithuania and Tajikistan) reported using a combination of these two methods, and several others (such as Portugal, Romania, the Russian Federation and Spain) used the relationship to a single person but also collected information on the person’s parents if they were living in the same household.

It might be expected that the ‘matrix’ approach would be more common among traditional census taking countries - where information might be more easily collected from such a specific question on a census form - but of these, only Bosnia and Herzegovina, Greece, Luxembourg and the United Kingdom did so. However, four of the nine register-based countries (Austria, Finland, Iceland and Sweden) reported using this method. On the other hand, Norway noted the difficulty of doing so by reporting that data from its registers identifies relationship between spouses, cohabitants and children/parents only, and that all other types of relationship were not known.

Although the CES Recommendations recommended that information on the relationship to the HRP should be collected, it was left to countries to choose among several different criteria to determine who the reference person should be. The 2010 CESR noted (at paragraph 513) that “… the selection of the one reference person in a household to whom all other persons in the household report, or designate, their relationship requires careful consideration. In the past the person considered to be the 'head' of the household was generally used as the reference person, but this concept is no longer considered appropriate in many countries of the region. It has also sometimes

170

been proposed that the person designated as the reference person should be the oldest person in the household or the one who contributes the most income.”

However, the CESR went on to recognise that given that the primary purpose of such identification was to assign family status and to assign individuals into families, both of these approaches have weaknesses. The automatic selection of the oldest person, for example, may be undesirable because in multi-generational households the broadest range of explicit kin relationships can be reported where the reference person is selected from the middle generation. Similarly, the selection of the person with the highest income may be a person who will not solicit the broadest range of explicit kin relationships. The CESR noted that there was some evidence though to suggest that the following criteria for selection of the reference person would yield the most fruitful range of

However, the CESR went on to recognise that given that the primary purpose of such identification was to assign family status and to assign individuals into families, both of these approaches have weaknesses. The automatic selection of the oldest person, for example, may be undesirable because in multi-generational households the broadest range of explicit kin relationships can be reported where the reference person is selected from the middle generation. Similarly, the selection of the person with the highest income may be a person who will not solicit the broadest range of explicit kin relationships. The CESR noted that there was some evidence though to suggest that the following criteria for selection of the reference person would yield the most fruitful range of