Population measurement and sources · Lesson 2

Census, register and survey

4 min read

Lesson objectives

  • Choose between a census, a population register and a survey for a defined question, then qualify coverage.
Reading tools

Census, register and survey

Choose a source by the question it can answer, not by its reputation. A census aims to cover a defined population at a reference date and can provide detailed small-area information. Yet it is periodic, and some measurements may age before the next round. An administrative register is updated when a particular transaction occurs, sometimes frequently, but entry follows service and registration rules rather than necessarily the desired population definition. A survey can ask detailed questions absent from registers, based on a sample. Its estimate for a small district may be imprecise if only a few sampled households lie there. Do not reduce these tradeoffs to slogans such as ‘census accurate, register recent, survey cheap.’ Examine the universe, coverage, reference date, sample design and missing information in each case.

A service planner may ask two distinct questions: how many usual residents lived in each district at the baseline date, and what share of households uses a service? A recent district-level census may provide a baseline for the first question after boundary and quality checks. A trustworthy population register could also help if its resident definition and coverage are known and relevant. The second question measures use, not simply registration. A user register omits nonusers and may also reflect enrolment rules. A suitable probability survey can estimate use with uncertainty if its sample represents the target population. Without such design, a citywide share is not an exact district fact.

Build a source decision matrix with four columns: question, candidate source, reason it fits, and limitation to disclose. Record an alternative if the first source fails. Combining sources may help but creates alignment work. Do not add service register rows to a census as if every row were a new resident. Do not compare an annual survey percentage with an old census count without stating the time gap. Each result should name the source, edition or period, and access date, distinguishing an observed count from a sample estimate.

Lesson summary

Choose a population source by question, coverage, date and unit, keeping resident counts separate from household use shares.