Decide what you need to count
Begin with the unit relevant to your question: people who might use a service, households that might buy it, or premises that might need installation. Those totals are not interchangeable. A property, an address, a family and a Census household should not be treated as automatic equivalents.
A household is a group of people
Under the Census 2011 instructions, a household usually comprises people who normally live together and share a kitchen, allowing for work-related absences. Members can be related, unrelated or both. A person living alone also constitutes a household. Consequently, a household count is not simply a count of married families. [1]
A census house is a building, or a separately recognised part of one with its own main entrance from a road, common courtyard or staircase. It may be occupied or vacant and used for residential, non-residential or mixed purposes. A building containing independent flats can contain several census houses. [1]
The instructions explicitly allow more than one household in a census house. Separate households sharing accommodation therefore need not imply separate buildings. Equally, a vacant or commercial census house does not automatically represent a resident household or a home available to purchase. [1]
Read the exclusions before interpreting a total
An institutional household comprises unrelated people living in an institution and sharing a kitchen; hostels and old-age homes are examples. Census definitions also recognise houseless households living outside buildings or census houses. These categories matter when matching a population numerator with a household denominator. [2]
Census 2011 table HL-01 counts households, excluding institutional households, by the condition of the census houses they occupy. It distinguishes residential from residence-cum-other use and categories of good, livable and dilapidated condition. Its observation unit is the household. A percentage calculated from this table must therefore retain that household denominator; it is not a percentage of every building or all vacant and occupied housing stock. [3]
For ownership, household size and dwelling rooms, use the appropriately labelled HL-04 table, which also excludes institutional households. Its room categories describe accommodation available to households; rooms are not an additional count of homes. Keep the ownership and occupancy measures separate when assessing a housing-related question. [4]
Use household-size assumptions transparently
Prefer a directly reported household count when it fits the geography, period and coverage. Otherwise, a scenario can divide the population living in the relevant household category by an assumed average household size. Numerator and denominator must describe the same area, date and population group. Combining all residents with a size measure that excludes institutions produces an inconsistent estimate.
For illustration only, 45,000 people divided by 4.5 people per household gives 10,000 households. Changing the assumption to 4 gives 11,250; using 5 gives 9,000. None of these divisors is asserted to be an official national or city mean. A calculator using 4.5 should label it as an editable assumption, show the range and preserve the original population reference year.
Translate household scale into the question you can answer
For a home-service study, document the reachable area, eligible occupied households and the evidence for service need. If institutions are potential customers, assess their purchasing arrangements separately instead of converting their residents into ordinary households. Keep assumed adoption, replacement frequency and expenditure visible as additional steps.
Record whether your result is a census count, a survey estimate or a calculation. State the reference period, exclusions and household-size assumption beside the answer. This lets another researcher change one assumption without mistaking the resulting scenario for an official housing statistic.