A national census aims to count every single resident of a country, an ambitious goal given that it means reaching and recording every household across an entire nation's territory at once. In practice, no census ever achieves a perfectly exact count: some people get missed entirely, an undercount, while others end up recorded more than once or counted in error, an overcount. Because these errors are essentially unavoidable at national scale, demographers treat raw census results as a very good estimate that needs further statistical correction, not as a perfectly exact figure to be taken at face value.
Undercounting and overcounting happen for genuinely different, systematic reasons
Undercounting tends to concentrate among specific, identifiable groups: people experiencing homelessness, recent immigrants wary of engaging with government forms, and residents of remote or hard-to-reach areas are all more likely to be missed by census-takers than the general population, which means census undercounts aren't randomly scattered but systematically concentrated in particular groups. Overcounting happens for its own distinct reasons, most commonly when a person with more than one residence, a student living away at university, for instance, gets recorded at both their term-time and family addresses by mistake. Because these two kinds of error stem from different underlying causes, correcting for them requires genuinely different statistical techniques, not a single uniform adjustment.
Demographers use independent methods to estimate and correct for the gap
To estimate how far a census's raw count likely diverges from the true population, demographers commonly use follow-up survey methods, conducting an independent, smaller-scale recount of a sample of areas after the main census and statistically comparing the two counts to estimate the overall undercount or overcount rate. These estimated correction factors are then applied to adjust the raw census figures into a more statistically reliable final population estimate, which is precisely why the population figures demographers and governments actually rely on for planning and funding decisions are corrected estimates built on top of the raw census, not the literal, unadjusted headcount itself.
What we're still unsure about
That census undercounting and overcounting are real, systematic and statistically correctable phenomena is well established, extensively documented demographic methodology, applied consistently across national statistical agencies. What remains more genuinely contested, particularly in political contexts, is exactly how large a given census's undercount or overcount actually is for specific groups, since these correction estimates themselves carry real statistical uncertainty and can become the subject of significant political dispute when they affect how government funding or political representation gets allocated — statisticians and demographers continue refining correction methods, but the precise scale of undercount for hard-to-reach populations remains a genuinely debated, actively studied question rather than one with a single universally agreed number.
This sits inside Population Measurement & Census Methods, one of seven topics in Demographics, one of four domains in Sociology, one of seventeen subjects the app can quiz you on.