A census would get data from 100% of the population (or at least close to 100%). Sampling would be to get data from some of the population (much less than 100%).
Sampling gives good insight of the choosen sample
It is a judgment call based on the accuracy required, time allotted and costs. If you need data quickly, at a reasonable cost and time (of course depending on how many individuals there are in the population), you would take a sample. If the population is a reasonable number that all individuals in the population could be used, then go for the census.
A questionnaire has little to do with sampling technique. Sampling technique is to do with who gets the questionnaire and that can be any sampling technique: the questionnaire can be sent to everyone (census), to a random sample, stratified random samples, to random samples in clusters, by quota or convenience. Or a pile of questionnaires can be left for respondents to pick up - self-selection.
Multistage sampling is a form of cluster sampling where instead of using the entire cluster, random samples from each cluster are used. This is typically used when doing opinion polls or surveys.
Sapling? Sampling??
The main difference between sampling and census is that in sampling, a subset of the population is selected and studied to make inferences about the entire population, while in a census, data is collected from every individual or element in the population. Sampling is more cost-effective and less time-consuming compared to a census, which requires resources to collect information from every unit in the population.
boobs
A census would get data from 100% of the population (or at least close to 100%). Sampling would be to get data from some of the population (much less than 100%).
Margaret Gurney has written: 'Sampling applications of the 1970 census publications, maps, and public use summary files' -- subject(s): Census, 19th, 1970, Sampling (Statistics)
Less time and less cost for a sample
Sampling is necessary in a few places. It could be when eating, painting and building.
census
Sampling gives good insight of the choosen sample
If it is too time consuming and/or expensive to analyse the whole population of interest you can take a sample instead. If the survey is conducted using correct sampling techniques (e.g. randomised selection, adequate sample size, etc.) the survey can tell you just as much as basing your results on a census.
There are several advantages of sampling over census (i.e. selection of wholepopulation for analysis).Firstly, the costs on sampling should be much lower than that on census. For example,for the government by-census (note: population census is usually conducted onceevery ten years and a by-census is conducted in the middle of the intercensal period),one fifth of the population is large enough to declare what the government wants toknow. There is no need to spend several times of dollars to interview the entirepopulation in the society.Secondly, a quality guru (Deming, 1960) argued that the quality of a study was oftenbetter with sampling than with a census. He suggested that, "Sampling possesses thepossibility of better interviewing(testing), more thorough investigation of missing,wrong , or suspicious information, better supervision, and better processing than ispossible with complete coverage". Research findings substantiate this opinion. Morethan 90% of survey error in one study was from non-sampling error1, and 10% or lesswas from sampling error2. (Donald et al., 1995)Thirdly, sampling can save the time. The speed of execution reduces the time betweenthe recognition of a need for information and the availability of that information.1 Non-sampling error is the error of research due to factors other than the sample size and samplingmethod, including non-response, bad communication with interviewees, measurement error, etc.2 Sampling error is the error during research due to the sample size and sampling method.
It is more accurate, unbiased and includes every item in the population, whereas sampling may be biased, and sampling is not totally representative.