A national headcount sounds like an administrative exercise and is routinely one of the most contested processes a government runs. The reason is what the numbers are used for.
The count allocates representation
In many systems, legislative seats are distributed among regions in proportion to population, and boundaries are redrawn on the basis of census results.
A region that is undercounted loses representation for the whole period until the next count, which is typically a decade.
Because the effect lasts so long and cannot be corrected afterwards, every party has a direct interest in how enumeration is conducted in areas where it is strong or weak.
Funding formulas depend on the same numbers
Central transfers to regions and municipalities are frequently calculated using population, along with characteristics such as age structure, household size and deprivation.
An undercounted neighbourhood receives less for schools, health services and transport than its actual population requires, compounding the disadvantage that made it hard to count.
Businesses and infrastructure planners use the same figures, so the effects extend beyond the public sector.
Some populations are systematically harder to reach
People in informal housing, those moving frequently, students, undocumented residents and households without reliable postal addresses are all more likely to be missed.
Statistical agencies address this with follow-up visits, administrative data and modelled estimates for non-responding households, each of which is itself contested.
Groups that fear the information could be shared with enforcement agencies respond at lower rates, which is why confidentiality guarantees are treated as operationally critical.
Question design is a policy decision
How categories are defined determines what can be measured, and adding, removing or rewording a question changes which groups become visible in the data.
Ethnicity, language, religion, disability and household relationships are all areas where wording has been argued over extensively, because the results feed into legal protections.
Comparability across decades competes with accuracy today, since changing a question breaks the time series that policy analysis relies on.
Counting itself is expensive and slow to fix
A national census involves recruiting a very large temporary workforce, mapping every address and processing responses within a compressed period.
Errors discovered after publication are hard to remedy, since the allocations built on the figures are already in force.
Several countries have shifted towards registers and administrative data to reduce cost, which changes the politics rather than removing it, because the rules for combining datasets become the new battleground.