A household is a measurement choice
It is tempting to treat a household as a natural unit, like a house or a family. In practice it is a definition, written by statisticians so that people can be counted consistently. Different countries and different surveys often use somewhat different wording, but a common idea is a person or group of people who live at the same address and share certain arrangements, such as meals or housekeeping. The details matter, because each phrase in that definition decides who is counted where.
This is why household structure is a good topic for learning how to read trend claims. A statement such as “more people now live alone” or “families are getting smaller” can be accurate under one definition and misleading under another.
Household, family and dwelling are different
Three terms are often used loosely.
- A dwelling is the physical home: a house, flat or room.
- A household is the people living there, as defined by the survey.
- A family is usually defined by relationships, such as couples or parents with children, and may be counted separately.
One dwelling can contain several households, and one household can contain more than one family or none at all. A group of unrelated adults sharing a flat is a household but typically not a family in the official sense. Claims about “the family” that quietly rely on household counts can therefore mix up two different things.
Where definitions get difficult
Several everyday situations do not fit neatly into a household definition, and how they are handled tends to affect the totals.
Shared meals and finances. If a definition requires sharing food or bills, an adult renting a room might be a separate household. If it requires only a common address, the same person is part of a larger one.
Students and people away from home. A young adult at university may be counted at the parental home, at a term-time address, or in a separate institutional category, depending on the rules. The choice can change how many young adults appear to live with parents.
Shared custody. Children who spend time in two homes may be counted in one, both or neither, depending on the survey. This affects statements about single-parent households and about children’s living arrangements.
Institutions and temporary housing. People in care homes, hospitals, prisons, boarding schools or shelters are often counted outside the household population. If a trend involves older people moving into care, household figures may leave out much of the change.
Informal arrangements. People who live together without formal ties, those who split time between addresses for work, and people with no fixed address are all hard to place.
How the definition changes the story
Consider the common claim that more people live alone. The number of one-person households can rise for several reasons that mean quite different things: more people choosing to live alone, more older people outliving partners, more people delaying partnership, or simply a population that is older overall. A rise in one-person households is a count of housing arrangements, not of loneliness or preferences, a distinction that loneliness research is careful to keep.
Average household size has a similar problem. It is calculated by dividing people by households, so it can fall because families are having fewer children, because adult children move out earlier, or because a growing share of older adults live alone. The single figure cannot say which. Likewise, a decline in multigenerational living may reflect changes in housing costs, migration patterns or cultural norms, and the same figures may be consistent with several explanations, which is a general feature of rival explanations for a trend.
Definitions also change over time. If a census adds a new category, such as recording cohabiting couples separately from married ones, or changes how it treats students, then a comparison across census years may partly reflect the new rules. Reputable statistical agencies usually note such changes, but secondary reporting often drops the caveat.
Surveys and censuses count differently
Households are counted through two main routes. A census attempts to include everyone at a set time, while sample surveys ask a smaller group more detailed questions between censuses. A census gives broad coverage but limited detail, while a survey can ask about relationships and money but is subject to sampling error and non-response. Neither is perfect. Hard-to-reach groups, such as people without a stable address, are often undercounted, which can lower estimates of unusual living arrangements.
Who answers is another issue. In many surveys one member reports for the whole household, and that person may describe relationships differently from others living there.
Questions to ask of a household claim
A few checks tend to help.
- What definition of household was used, and did it stay the same across the years being compared?
- Are institutions, students and people without fixed addresses included or excluded?
- Does the claim concern households, families or individuals? These can move in different directions.
- Is an average being used to describe a diverse set of arrangements?
- Could the population itself have changed, for example by ageing, and does the claim account for it?
Reading household figures with these questions in mind usually turns a dramatic-sounding statement about family change into a more modest one: something has probably shifted, but how much, and why, depends on the rules used to count.