Statistics · Ch 1 — Collection of Data
Census Method versus Sample Method
Census Method versus Sample Method
Once it is decided that data must be collected, a further basic choice arises: should every single unit of the population be studied, or only a part of it?
Census (Complete Enumeration)
The Census Method collects information from every single unit of the population under study, with no unit left out. India's decennial Census of India (conducted every 10 years, most recently in the 2020s cycle) is the best-known example — every household in the country is enumerated.
Sample Method
The Sample Method collects information from only a representative part (a sample) of the population, and the results are then generalised to the whole population.
| Basis | Census Method | Sample Method |
|---|---|---|
| Coverage | Every unit of the population | Only a selected part (sample) |
| Accuracy | Free of sampling error (though not of non-sampling error) | Subject to sampling error, but often more free of non-sampling error since fewer units are handled with more care |
| Cost | Very high | Comparatively low |
| Time required | Very long | Comparatively short |
| Suitability | Small populations; when data on every individual unit is legally/administratively required (e.g. a national census, an electoral roll) | Large or infinite populations; when speed and cost matter; when testing is destructive (e.g. testing how long light bulbs last — you cannot test every bulb and still have any left to sell) |
A method of data collection in which every single unit of the defined population is studied, with none omitted — e.g., the dec …
A method of data collection in which only a selected, representative part (sample) of the population is studied, and conclusions are generalise …
The error that arises solely because conclusions about a population are drawn from studying only a sample of it rather than every unit; it is absent in a true census and tend …
Errors arising from causes other than sampling itself — e.g. a poorly worded questionnaire, non-response, recording or transcription mistakes — which can affect both census and sample data, and are often harde …