Household and family nucleus – what are these?

Posted on 6 September 2022, 14:58, population statistics expert Ene-Margit Tiit

Censuses introduce new terms that might be unfamiliar for most people, such as ‘household’, ‘family nucleus’, ‘partner’, and even such a peculiar concept as ‘non-family household’. What do all these terms mean and what is their purpose?

Statistical definition of a family

First, let’s look at the meaning of family nucleus. It is synonymous with family. A family nucleus, i.e. a family, can be:

  • a married/cohabiting couple without children;
  • a married/cohabiting couple with at least one child;
  • a lone parent with at least one child.

Thus, in a family nucleus, the members are in a relationship as partners or as parent and child.

For statistical purposes, married couples and partners in a consensual union are considered equivalent. The two persons in a couple are called partners regardless of whether they are married, in a registered partnership or in an unregistered consensual union. It could also be a same-sex couple who are married, in a registered partnership or in a consensual union. Statistics are not concerned with people’s views or attitudes; statistics just record the facts.

In the case of parent-child relationships, it is assumed that the child is the partners’ or one partner’s biological child. An officially adopted child is equivalent to a biological child, but a step-child or a child living in a foster family or in a substitute home is not a member of the family nucleus. A child remains a member of the family nucleus even as an adult. But once the child starts their own family, they no longer belong to the same family nucleus with their parents.  

A family nucleus does not include grandparents or grandchildren; it can only comprise members of two generations.

In reality, grandparents often live with a family nucleus. A son will not be thrown out of the house if he brings a girlfriend to live with him. This means that we need another term in addition to ‘family nucleus’ that has fewer restrictions on membership. That term is ‘household’. In Estonian, ‘family’ has been used instead of ‘household’ in some earlier censuses – family was defined as the people living in the same dwelling. Members of the same household often share finances, although this is not a universal requirement today.

Statistical definition of a household

There are various possible household structures.

  • Every person who lives alone counts as a separate household.
  • Each family nucleus living on its own is also considered a household.

The difference between family and household is that a household may include other members in addition to the family nucleus. For example, a household could consist of father, mother, child and grandmother, whereas the father, mother and child constitute the family nucleus while the grandmother is an additional member. There may be no ties of kinship between household members, although additional household members are usually related to a member of the family nucleus.

A household could also contain several family nuclei. For example, a mother and child, and the mother’s parents – in this case, this would be a two-family household. The mother used to belong to the same family as her parents, and now she and her child constitute a separate family nucleus. A person cannot belong to two families simultaneously. If a household member (based on ties of kinship) could belong to several family nuclei within a single household, the younger generation is preferred in enumeration.

If a household member moves away (i.e. goes to live in a different dwelling), he or she is no longer considered to belong to this household or this family, even though ties of kinship remain. Change of residence is not instantaneous in statistics, it usually happens gradually over a year.

There are also households without a family nucleus, such as when several students rent a shared dwelling.

How have the concepts of household and family evolved over time?

What is the origin of these terms and concepts? Who conceived them, and when? As we know, it has been internationally agreed which kind of information should be collected with a population census. In the 19th century, relevant recommendations were made by the International Statistical Congress. Today, these guidelines are issued by the United Nations and Eurostat (at the European level). This allows the production of international statistics including comparisons between countries.

The domestic situation has always been an important topic in censuses. In the first census conducted in Estonia in 1881, enumerators used a so-called house form where they listed all the people living in a house (i.e. basically the residents of each dwelling) and their status in the house. The marital status was also recorded for all enumerated persons.

House forms were not used in the next census in 1897. Enumerators did not list the people living in the house; instead, they wanted to know who belonged to the household and what their status was. They also determined how many persons belonged to the (main) family.

The first population census organised by the Republic of Estonia took place in 1922. There was a greater focus on collection of data about the ‘society’ which referred to households and families. Such a two-tier approach was very similar to the current understanding: the main difference was that a single person was not considered a household. Family was basically defined the same way as the family nucleus today. At that time, the majority of Estonians lived in rural areas, and a household meant a family living on a farm. The second census of the Republic of Estonia, which was conducted in 1934, was very well organised and is considered the most accurate population census in the territory of Estonia. The questionnaire of the 1934 census is almost the same as the questionnaire of the most recent census.

There was a census in 1941 during the German occupation, but data on families and households were not collected then. The Soviet-era censuses (in 1959, 1970, 1979 and 1989) did not use the term ‘household’, due to the capitalist connotations of the Estonian term for it. Still, families were more or less construed as households today, and enumerators noted both the number of household members living with the family and the number of members living separately from the family. Families were also grouped by structure – for example, family of a married couple with/without children; family of a married couple and one or more of the spouses’ parents or other relatives; family of a mother and child(ren); family of a father and child(ren); and other families. Persons living alone were counted separately.

New household types in re-independent Estonia

The first population census after the restoration of independence was carried out in 2000. The questionnaire was in full compliance with Eurostat’s standards. But, in addition to internationally required characteristics, there were also several national questions (about ethnic nationality, language, origin). The 2000 census adopted the contemporary definitions of household and family. Private and institutional households were defined and distinguished. An institutional household refers to an institution where the household members reside permanently (for a year or more). There was a new term that had to be defined – homeless people. In pre-war Estonia, it was the responsibility of each rural municipality to look after its residents and provide shelter for everyone (the worst-case scenario was having to go from farm to farm asking for shelter), whereas nowadays there is a considerable amount of people who have no home, mainly due to financial difficulties, with a few choosing to live this way. In our climate, homeless people mostly stay at homeless shelters which provide temporary residence.

When are partners actually partners, and what does this mean?

In the census, marital status is self-reported, just like all the other characteristics. Thus, it is possible for a man to (claim to) be married to several women, or vice-versa. Census data are aggregated, not presented in a personally identifiable format, so this does not cause misunderstandings. Since marital status was self-reported, it means that it was possible that an unregistered marriage (consensual union) was also reported as a marriage in earlier censuses. Today, most data for the census are obtained from registers, and the determination of unregistered partnerships is one of the most complicated tasks that requires the use of algorithms and data from many different registers.

Determining partnerships during an interview is not easy either, as this is seen as sensitive information that some respondents would rather not disclose to the enumerator. During the interview stage of the previous census, there were numerous examples where one of the partners confirmed the existence of a partnership, while the other partner claimed that they are an additional member of the household and do not have a relationship with any of the household members. Analysis of the distribution of non-relatives as household members in past censuses indicates that there are cases where partnership has been concealed – especially in earlier decades, for example, when there was a non-relative family member in a single-parent household.

The situation of same-sex couples is especially complicated, as there is ambivalence towards such relationships in the society and respondents aim to hide such relationships, despite all the data protection measures implemented in the census and during interviews. Nevertheless, in the previous census in 2011, we managed to enumerate about 100 partnerships between men and a similar number of partnerships between women; some couples also had children. Based on other countries’ experience, these figures are likely to be underestimated, but there is no good way to obtain better data for these characteristics.


In conclusion, considering increased geographic and residential mobility and the unwillingness of people to share personal information, it is not possible to obtain highly accurate data on a country’s population – the latest census data are also estimates, not exact enumeration figures. But the level of accuracy is sufficient for state-level decision-making. It is important to note that register-based statistics are (unlike traditional enumeration statistics) always timely and up to date, since the information in registers is regularly updated. It is the best way to get a good overview of our population, families and dwellings.